Changeset 219 for doc/html/libEF_8h-source.html

Timestamp:

12/18/08 20:38:46 (16 years ago)

Author:

smidl

Message:

doc

Files:

• doc/html/libEF_8h-source.html (modified) (4 diffs)

doc/html/libEF_8h-source.html

@@ -11 +11 @@
       <li><a href="index.html"><span>Main&nbsp;Page</span></a></li>
       <li><a href="pages.html"><span>Related&nbsp;Pages</span></a></li>
+      <li><a href="modules.html"><span>Modules</span></a></li>
       <li><a href="annotated.html"><span>Classes</span></a></li>
       <li class="current"><a href="files.html"><span>Files</span></a></li>
@@ -38 +39 @@
 <a name="l00044"></a>00044         <span class="keyword">virtual</span> <span class="keywordtype">double</span> <a class="code" href="classeEF.html#69e5680dac10375d62520d26c672477d" title="logarithm of the normalizing constant, ">lognc</a>() <span class="keyword">const</span> =0;
 <a name="l00046"></a><a class="code" href="classeEF.html#a89bef8996410609004fa019b5b48964">00046</a>         <span class="keyword">virtual</span> <span class="keywordtype">void</span> <a class="code" href="classeEF.html#a89bef8996410609004fa019b5b48964" title="TODO decide if it is really needed.">dupdate</a> ( mat &amp;v ) {it_error ( <span class="stringliteral">"Not implemented"</span> );};
-<a name="l00048"></a><a class="code" href="classeEF.html#48cdd33d0e20d1a1aa45683c956bc61c">00048</a>         <span class="keyword">virtual</span> <span class="keywordtype">double</span> <a class="code" href="classeEF.html#48cdd33d0e20d1a1aa45683c956bc61c" title="Evaluate normalized log-probability.">evalpdflog_nn</a> ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const</span>{it_error ( <span class="stringliteral">"Not implemented"</span> );<span class="keywordflow">return</span> 0.0;};
-<a name="l00050"></a><a class="code" href="classeEF.html#6466e8d4aa9dd64698ed288cbb1afc03">00050</a>         <span class="keyword">virtual</span> <span class="keywordtype">double</span> <a class="code" href="classeEF.html#6466e8d4aa9dd64698ed288cbb1afc03" title="Evaluate normalized log-probability.">evalpdflog</a> ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const </span>{<span class="keywordtype">double</span> tmp;tmp= <a class="code" href="classeEF.html#48cdd33d0e20d1a1aa45683c956bc61c" title="Evaluate normalized log-probability.">evalpdflog_nn</a> ( val )-<a class="code" href="classeEF.html#69e5680dac10375d62520d26c672477d" title="logarithm of the normalizing constant, ">lognc</a>();it_assert_debug(std::isfinite(tmp),<span class="stringliteral">"why?"</span>); <span class="keywordflow">return</span> tmp;}
-<a name="l00052"></a><a class="code" href="classeEF.html#c71faf4b2d153efda14bf1f87dca1507">00052</a>         <span class="keyword">virtual</span> vec <a class="code" href="classeEF.html#6466e8d4aa9dd64698ed288cbb1afc03" title="Evaluate normalized log-probability.">evalpdflog</a> ( <span class="keyword">const</span> mat &amp;Val )<span class="keyword"> const </span>{
+<a name="l00048"></a><a class="code" href="classeEF.html#41c70565b4d3fb424599817d008f0c71">00048</a>         <span class="keyword">virtual</span> <span class="keywordtype">double</span> <a class="code" href="classeEF.html#41c70565b4d3fb424599817d008f0c71" title="Evaluate normalized log-probability.">evallog_nn</a> ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const</span>{it_error ( <span class="stringliteral">"Not implemented"</span> );<span class="keywordflow">return</span> 0.0;};
+<a name="l00050"></a><a class="code" href="classeEF.html#357512dd565e199904d367294b7dd862">00050</a>         <span class="keyword">virtual</span> <span class="keywordtype">double</span> <a class="code" href="classeEF.html#357512dd565e199904d367294b7dd862" title="Evaluate normalized log-probability.">evallog</a> ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const </span>{<span class="keywordtype">double</span> tmp;tmp= <a class="code" href="classeEF.html#41c70565b4d3fb424599817d008f0c71" title="Evaluate normalized log-probability.">evallog_nn</a> ( val )-<a class="code" href="classeEF.html#69e5680dac10375d62520d26c672477d" title="logarithm of the normalizing constant, ">lognc</a>();it_assert_debug(std::isfinite(tmp),<span class="stringliteral">"Infinite value"</span>); <span class="keywordflow">return</span> tmp;}
+<a name="l00052"></a><a class="code" href="classeEF.html#cff03a658aec11b806c3e3d48f37b81f">00052</a>         <span class="keyword">virtual</span> vec <a class="code" href="classeEF.html#357512dd565e199904d367294b7dd862" title="Evaluate normalized log-probability.">evallog</a> ( <span class="keyword">const</span> mat &amp;Val )<span class="keyword"> const </span>{
 <a name="l00053"></a>00053                 vec x ( Val.cols() );
-<a name="l00054"></a>00054                 <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i=0;i&lt;Val.cols();i++ ) {x ( i ) =<a class="code" href="classeEF.html#48cdd33d0e20d1a1aa45683c956bc61c" title="Evaluate normalized log-probability.">evalpdflog_nn</a> ( Val.get_col ( i ) ) ;}
+<a name="l00054"></a>00054                 <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i=0;i&lt;Val.cols();i++ ) {x ( i ) =<a class="code" href="classeEF.html#41c70565b4d3fb424599817d008f0c71" title="Evaluate normalized log-probability.">evallog_nn</a> ( Val.get_col ( i ) ) ;}
 <a name="l00055"></a>00055                 <span class="keywordflow">return</span> x-<a class="code" href="classeEF.html#69e5680dac10375d62520d26c672477d" title="logarithm of the normalizing constant, ">lognc</a>();
 <a name="l00056"></a>00056         }
@@ -88 +89 @@
 <a name="l00127"></a>00127         vec <a class="code" href="classenorm.html#60b47544f6181ffd4530d3e415ce12c5" title="Returns a sample,  from density .">sample</a>() <span class="keyword">const</span>;
 <a name="l00129"></a>00129         mat <a class="code" href="classenorm.html#60b47544f6181ffd4530d3e415ce12c5" title="Returns a sample,  from density .">sample</a> ( <span class="keywordtype">int</span> N ) <span class="keyword">const</span>;
-<a name="l00130"></a>00130         <span class="keywordtype">double</span> eval ( <span class="keyword">const</span> vec &amp;val ) <span class="keyword">const</span> ;
-<a name="l00131"></a>00131         <span class="keywordtype">double</span> <a class="code" href="classenorm.html#c1e3dcba256b0153cfdb286120e110be" title="Evaluate normalized log-probability.">evalpdflog_nn</a> ( <span class="keyword">const</span> vec &amp;val ) <span class="keyword">const</span>;
+<a name="l00130"></a>00130 <span class="comment">//      double eval ( const vec &amp;val ) const ;</span>
+<a name="l00131"></a>00131         <span class="keywordtype">double</span> <a class="code" href="classenorm.html#50cb0a083d97a7adbbd97c92e712c46c" title="Evaluate normalized log-probability.">evallog_nn</a> ( <span class="keyword">const</span> vec &amp;val ) <span class="keyword">const</span>;
 <a name="l00132"></a>00132         <span class="keywordtype">double</span> <a class="code" href="classenorm.html#b289a36a69db59d182bb6eba9c05d4a8" title="logarithm of the normalizing constant, ">lognc</a> () <span class="keyword">const</span>;
 <a name="l00133"></a><a class="code" href="classenorm.html#50fa84da7bae02f7af17a98f37566899">00133</a>         vec <a class="code" href="classenorm.html#50fa84da7bae02f7af17a98f37566899" title="return expected value">mean</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> <a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a>;}
@@ -102 +103 @@
 <a name="l00144"></a>00144 
 <a name="l00146"></a><a class="code" href="classenorm.html#7a5034b25771a84450a990d10fc40ac9">00146</a>         sq_T&amp; <a class="code" href="classenorm.html#7a5034b25771a84450a990d10fc40ac9" title="returns pointers to the internal variance and its inverse. Use with Care!">_R</a>() {<span class="keywordflow">return</span> <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>;}
-<a name="l00147"></a>00147 
-<a name="l00149"></a>00149 <span class="comment">//      mat getR () {return R.to_mat();}</span>
-<a name="l00150"></a>00150 };
-<a name="l00151"></a>00151 
-<a name="l00158"></a><a class="code" href="classegiw.html">00158</a> <span class="keyword">class </span><a class="code" href="classegiw.html" title="Gauss-inverse-Wishart density stored in LD form.">egiw</a> : <span class="keyword">public</span> <a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> {
-<a name="l00159"></a>00159 <span class="keyword">protected</span>:
-<a name="l00161"></a><a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442">00161</a>         <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a> <a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>;
-<a name="l00163"></a><a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453">00163</a>         <span class="keywordtype">double</span> <a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a>;
-<a name="l00165"></a><a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e">00165</a>         <span class="keywordtype">int</span> <a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a>;
-<a name="l00167"></a><a class="code" href="classegiw.html#c70d13d86e0d9f0acede3e1dc0368812">00167</a>         <span class="keywordtype">int</span> <a class="code" href="classegiw.html#c70d13d86e0d9f0acede3e1dc0368812" title="Dimension of the regressor.">nPsi</a>;
-<a name="l00168"></a>00168 <span class="keyword">public</span>:
-<a name="l00170"></a><a class="code" href="classegiw.html#056c094f01ca1cc308d72162f47617c9">00170</a>         <a class="code" href="classegiw.html#056c094f01ca1cc308d72162f47617c9" title="Default constructor, if nu0&amp;lt;0 a minimal nu0 will be computed.">egiw</a> ( <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>, mat V0, <span class="keywordtype">double</span> nu0=-1.0 ) : <a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> ( rv ), <a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a> ( V0 ), <a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a> ( nu0 ) {
-<a name="l00171"></a>00171                 <a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a> = rv.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() /<a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>.<a class="code" href="classldmat.html#96dfb21865db4f5bd36fa70f9b0b1163" title="access function">rows</a>();
-<a name="l00172"></a>00172                 it_assert_debug ( rv.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ==<a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a>*<a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>.<a class="code" href="classldmat.html#96dfb21865db4f5bd36fa70f9b0b1163" title="access function">rows</a>(),<span class="stringliteral">"Incompatible V0."</span> );
-<a name="l00173"></a>00173                 <a class="code" href="classegiw.html#c70d13d86e0d9f0acede3e1dc0368812" title="Dimension of the regressor.">nPsi</a> = <a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>.<a class="code" href="classldmat.html#96dfb21865db4f5bd36fa70f9b0b1163" title="access function">rows</a>()-<a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a>;
-<a name="l00174"></a>00174                 <span class="comment">//set mu to have proper normalization and </span>
-<a name="l00175"></a>00175                 <span class="keywordflow">if</span> (nu0&lt;0){
-<a name="l00176"></a>00176                         <a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a> = 0.1 +nPsi +2*<a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a> +2; <span class="comment">// +2 assures finite expected value of R</span>
-<a name="l00177"></a>00177                         <span class="comment">// terms before that are sufficient for finite normalization</span>
-<a name="l00178"></a>00178                 }
-<a name="l00179"></a>00179         }
-<a name="l00181"></a><a class="code" href="classegiw.html#18c1bf6125652a6dcbca68dd02dddd8d">00181</a>         <a class="code" href="classegiw.html#056c094f01ca1cc308d72162f47617c9" title="Default constructor, if nu0&amp;lt;0 a minimal nu0 will be computed.">egiw</a> ( <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>, <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a> V0, <span class="keywordtype">double</span> nu0=-1.0 ) : <a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> ( rv ), <a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a> ( V0 ), <a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a> ( nu0 ) {
-<a name="l00182"></a>00182                 <a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a> = rv.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() /<a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>.<a class="code" href="classldmat.html#96dfb21865db4f5bd36fa70f9b0b1163" title="access function">rows</a>();
-<a name="l00183"></a>00183                 it_assert_debug ( rv.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ==<a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a>*<a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>.<a class="code" href="classldmat.html#96dfb21865db4f5bd36fa70f9b0b1163" title="access function">rows</a>(),<span class="stringliteral">"Incompatible V0."</span> );
-<a name="l00184"></a>00184                 <a class="code" href="classegiw.html#c70d13d86e0d9f0acede3e1dc0368812" title="Dimension of the regressor.">nPsi</a> = <a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>.<a class="code" href="classldmat.html#96dfb21865db4f5bd36fa70f9b0b1163" title="access function">rows</a>()-<a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a>;
-<a name="l00185"></a>00185                 <span class="keywordflow">if</span> (nu0&lt;0){
-<a name="l00186"></a>00186                         <a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a> = 0.1 +nPsi +2*<a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a> +2; <span class="comment">// +2 assures finite expected value of R</span>
-<a name="l00187"></a>00187                         <span class="comment">// terms before that are sufficient for finite normalization</span>
-<a name="l00188"></a>00188                 }
-<a name="l00189"></a>00189         }
-<a name="l00190"></a>00190 
-<a name="l00191"></a>00191         vec <a class="code" href="classegiw.html#3d2c1f2ba0f9966781f1e0ae695e8a6f" title="Returns a sample,  from density .">sample</a>() <span class="keyword">const</span>;
-<a name="l00192"></a>00192         vec <a class="code" href="classegiw.html#6deb0ff2859f41ef7cbdf6a842cabb29" title="return expected value">mean</a>() <span class="keyword">const</span>;
-<a name="l00193"></a>00193         <span class="keywordtype">void</span> mean_mat ( mat &amp;M, mat&amp;R ) <span class="keyword">const</span>;
-<a name="l00195"></a>00195         <span class="keywordtype">double</span> <a class="code" href="classegiw.html#2ab1e525d692be8272a6f383d60b94cd" title="In this instance, val= [theta, r]. For multivariate instances, it is stored columnwise...">evalpdflog_nn</a> ( <span class="keyword">const</span> vec &amp;val ) <span class="keyword">const</span>;
-<a name="l00196"></a>00196         <span class="keywordtype">double</span> <a class="code" href="classegiw.html#70eb1a0b88459b227f919b425b0d3359" title="logarithm of the normalizing constant, ">lognc</a> () <span class="keyword">const</span>;
-<a name="l00197"></a>00197 
-<a name="l00198"></a>00198         <span class="comment">//Access</span>
-<a name="l00200"></a><a class="code" href="classegiw.html#533e792e1175bfa06d5d595dc5d080d5">00200</a> <span class="comment"></span>        <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a>&amp; <a class="code" href="classegiw.html#533e792e1175bfa06d5d595dc5d080d5" title="returns a pointer to the internal statistics. Use with Care!">_V</a>() {<span class="keywordflow">return</span> <a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>;}
-<a name="l00202"></a><a class="code" href="classegiw.html#a46c8a206edf80b357a138d7491780c1">00202</a>         <span class="keyword">const</span> <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a>&amp; <a class="code" href="classegiw.html#533e792e1175bfa06d5d595dc5d080d5" title="returns a pointer to the internal statistics. Use with Care!">_V</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> <a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>;}
-<a name="l00204"></a><a class="code" href="classegiw.html#08029c481ff95d24f093df0573879afe">00204</a>         <span class="keywordtype">double</span>&amp; <a class="code" href="classegiw.html#08029c481ff95d24f093df0573879afe" title="returns a pointer to the internal statistics. Use with Care!">_nu</a>()  {<span class="keywordflow">return</span> <a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a>;}
-<a name="l00205"></a>00205         <span class="keyword">const</span> <span class="keywordtype">double</span>&amp; <a class="code" href="classegiw.html#08029c481ff95d24f093df0573879afe" title="returns a pointer to the internal statistics. Use with Care!">_nu</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> <a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a>;}
-<a name="l00206"></a><a class="code" href="classegiw.html#036306322a90a9977834baac07460816">00206</a>         <span class="keywordtype">void</span> <a class="code" href="classegiw.html#036306322a90a9977834baac07460816" title="Power of the density, used e.g. to flatten the density.">pow</a> ( <span class="keywordtype">double</span> p ) {<a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>*=p;<a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a>*=p;};
-<a name="l00207"></a>00207 };
-<a name="l00208"></a>00208 
-<a name="l00217"></a><a class="code" href="classeDirich.html">00217</a> <span class="keyword">class </span><a class="code" href="classeDirich.html" title="Dirichlet posterior density.">eDirich</a>: <span class="keyword">public</span> <a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> {
-<a name="l00218"></a>00218 <span class="keyword">protected</span>:
-<a name="l00220"></a><a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7">00220</a>         vec <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>;
-<a name="l00221"></a>00221 <span class="keyword">public</span>:
-<a name="l00223"></a><a class="code" href="classeDirich.html#ac7e6116f3575c3860d07355e96cd4af">00223</a>         <a class="code" href="classeDirich.html#ac7e6116f3575c3860d07355e96cd4af" title="Default constructor.">eDirich</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>, <span class="keyword">const</span> vec &amp;beta0 ) : <a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> ( rv ),<a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a> ( beta0 ) {it_assert_debug ( rv.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ==<a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>.length(),<span class="stringliteral">"Incompatible statistics"</span> ); };
-<a name="l00225"></a><a class="code" href="classeDirich.html#55cccbc5eb44764dce722567acf5fd58">00225</a>         <a class="code" href="classeDirich.html#ac7e6116f3575c3860d07355e96cd4af" title="Default constructor.">eDirich</a> ( <span class="keyword">const</span> <a class="code" href="classeDirich.html" title="Dirichlet posterior density.">eDirich</a> &amp;D0 ) : <a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> ( D0.<a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a> ),<a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a> ( D0.<a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a> ) {};
-<a name="l00226"></a><a class="code" href="classeDirich.html#23dff79110822e9639343fe8e177fd80">00226</a>         vec <a class="code" href="classeDirich.html#23dff79110822e9639343fe8e177fd80" title="Returns a sample,  from density .">sample</a>()<span class="keyword"> const </span>{it_error ( <span class="stringliteral">"Not implemented"</span> );<span class="keywordflow">return</span> vec_1 ( 0.0 );};
-<a name="l00227"></a><a class="code" href="classeDirich.html#4206e1da149d51ff3b663c9241096b73">00227</a>         vec <a class="code" href="classeDirich.html#4206e1da149d51ff3b663c9241096b73" title="return expected value">mean</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>/sum ( <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a> );};
-<a name="l00229"></a><a class="code" href="classeDirich.html#688a24f04be6d80d4769cf0e4ded7acc">00229</a>         <span class="keywordtype">double</span> <a class="code" href="classeDirich.html#688a24f04be6d80d4769cf0e4ded7acc" title="In this instance, val is ...">evalpdflog_nn</a> ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const </span>{<span class="keywordflow">return</span> ( <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>-1 ) *log ( val );};
-<a name="l00230"></a><a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77">00230</a>         <span class="keywordtype">double</span> <a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a> ()<span class="keyword"> const </span>{
-<a name="l00231"></a>00231                 <span class="keywordtype">double</span> gam=sum ( <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a> );
-<a name="l00232"></a>00232                 <span class="keywordtype">double</span> lgb=0.0;
-<a name="l00233"></a>00233                 <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i=0;i&lt;<a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>.length();i++ ) {lgb+=lgamma ( <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a> ( i ) );}
-<a name="l00234"></a>00234                 <span class="keywordflow">return</span> lgb-lgamma ( gam );
-<a name="l00235"></a>00235         };
-<a name="l00237"></a><a class="code" href="classeDirich.html#6409d0362143a23976b43641ff19e53a">00237</a>         vec&amp; <a class="code" href="classeDirich.html#6409d0362143a23976b43641ff19e53a" title="access function">_beta</a>()  {<span class="keywordflow">return</span> <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>;}
-<a name="l00239"></a><a class="code" href="classeDirich.html#c842acb2e1cce5cc9000769ff06c086d">00239</a>         <span class="keywordtype">void</span> <a class="code" href="classeDirich.html#c842acb2e1cce5cc9000769ff06c086d" title="Set internal parameters.">set_parameters</a> ( <span class="keyword">const</span> vec &amp;beta0 ) {
-<a name="l00240"></a>00240                 <span class="keywordflow">if</span> ( beta0.length() !=<a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>.length() ) {
-<a name="l00241"></a>00241                         it_assert_debug ( <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>.<a class="code" href="classRV.html#c114a6f3ff06796cc2f4dacba74291eb" title="Return length (number of entries) of the RV.">length</a>() ==1,<span class="stringliteral">"Undefined"</span> );
-<a name="l00242"></a>00242                         <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>.<a class="code" href="classRV.html#70b24c39c5130b1e4753fa2eef495433" title="access function">set_size</a> ( 0,beta0.length() );
-<a name="l00243"></a>00243                 }
-<a name="l00244"></a>00244                 <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>= beta0;
-<a name="l00245"></a>00245         }
-<a name="l00246"></a>00246 };
-<a name="l00247"></a>00247 
-<a name="l00249"></a><a class="code" href="classmultiBM.html">00249</a> <span class="keyword">class </span><a class="code" href="classmultiBM.html" title="Estimator for Multinomial density.">multiBM</a> : <span class="keyword">public</span> <a class="code" href="classBMEF.html" title="Estimator for Exponential family.">BMEF</a> {
-<a name="l00250"></a>00250 <span class="keyword">protected</span>:
-<a name="l00252"></a><a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5">00252</a>         <a class="code" href="classeDirich.html" title="Dirichlet posterior density.">eDirich</a> <a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>;
-<a name="l00254"></a><a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6">00254</a>         vec &amp;<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>;
-<a name="l00255"></a>00255 <span class="keyword">public</span>:
-<a name="l00257"></a><a class="code" href="classmultiBM.html#7d7d7e78c129602bcde96078359dc6e5">00257</a>         <a class="code" href="classmultiBM.html#7d7d7e78c129602bcde96078359dc6e5" title="Default constructor.">multiBM</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classBM.html#af00f0612fabe66241dd507188cdbf88" title="Random variable of the posterior.">rv</a>, <span class="keyword">const</span> vec beta0 ) : <a class="code" href="classBMEF.html" title="Estimator for Exponential family.">BMEF</a> ( rv ),<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a> ( rv,beta0 ),<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a> ( <a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>._beta() ) {<a class="code" href="classBMEF.html#308cf5d4133cd471fdf1ecd5dfa09d02" title="cached value of lognc() in the previous step (used in evaluation of ll )">last_lognc</a>=<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>();}
-<a name="l00259"></a><a class="code" href="classmultiBM.html#b92751adbfb9f259ca8c95232cfd9c09">00259</a>         <a class="code" href="classmultiBM.html#7d7d7e78c129602bcde96078359dc6e5" title="Default constructor.">multiBM</a> ( <span class="keyword">const</span> <a class="code" href="classmultiBM.html" title="Estimator for Multinomial density.">multiBM</a> &amp;B ) : <a class="code" href="classBMEF.html" title="Estimator for Exponential family.">BMEF</a> ( B ),<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a> ( <a class="code" href="classBM.html#af00f0612fabe66241dd507188cdbf88" title="Random variable of the posterior.">rv</a>,B.<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a> ),<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a> ( <a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>._beta() ) {}
-<a name="l00261"></a><a class="code" href="classmultiBM.html#42e36804041e551d3ceea6c75abc0562">00261</a>         <span class="keywordtype">void</span> <a class="code" href="classmultiBM.html#42e36804041e551d3ceea6c75abc0562" title="Sets sufficient statistics to match that of givefrom mB0.">set_statistics</a> ( <span class="keyword">const</span> <a class="code" href="classBM.html" title="Bayesian Model of the world, i.e. all uncertainty is modeled by probabilities.">BM</a>* mB0 ) {<span class="keyword">const</span> <a class="code" href="classmultiBM.html" title="Estimator for Multinomial density.">multiBM</a>* mB=<span class="keyword">dynamic_cast&lt;</span><span class="keyword">const </span><a class="code" href="classmultiBM.html" title="Estimator for Multinomial density.">multiBM</a>*<span class="keyword">&gt;</span> ( mB0 ); <a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>=mB-&gt;<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>;}
-<a name="l00262"></a><a class="code" href="classmultiBM.html#11eeba7e97954e316e959116f90d80e2">00262</a>         <span class="keywordtype">void</span> <a class="code" href="classmultiBM.html#11eeba7e97954e316e959116f90d80e2" title="Incremental Bayes rule.">bayes</a> ( <span class="keyword">const</span> vec &amp;dt ) {
-<a name="l00263"></a>00263                 <span class="keywordflow">if</span> ( <a class="code" href="classBMEF.html#538d632e59f9afa8daa1de74da12ce71" title="forgetting factor">frg</a>&lt;1.0 ) {<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>*=<a class="code" href="classBMEF.html#538d632e59f9afa8daa1de74da12ce71" title="forgetting factor">frg</a>;<a class="code" href="classBMEF.html#308cf5d4133cd471fdf1ecd5dfa09d02" title="cached value of lognc() in the previous step (used in evaluation of ll )">last_lognc</a>=<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>();}
-<a name="l00264"></a>00264                 <a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>+=dt;
-<a name="l00265"></a>00265                 <span class="keywordflow">if</span> ( <a class="code" href="classBM.html#bf6fb59b30141074f8ee1e2f43d03129" title="If true, the filter will compute likelihood of the data record and store it in ll...">evalll</a> ) {<a class="code" href="classBM.html#5623fef6572a08c2b53b8c87b82dc979" title="Logarithm of marginalized data likelihood.">ll</a>=<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>()-<a class="code" href="classBMEF.html#308cf5d4133cd471fdf1ecd5dfa09d02" title="cached value of lognc() in the previous step (used in evaluation of ll )">last_lognc</a>;}
-<a name="l00266"></a>00266         }
-<a name="l00267"></a><a class="code" href="classmultiBM.html#13e26a61757278981fd8cac9a7ef91eb">00267</a>         <span class="keywordtype">double</span> <a class="code" href="classmultiBM.html#13e26a61757278981fd8cac9a7ef91eb">logpred</a> ( <span class="keyword">const</span> vec &amp;dt )<span class="keyword"> const </span>{
-<a name="l00268"></a>00268                 <a class="code" href="classeDirich.html" title="Dirichlet posterior density.">eDirich</a> pred ( <a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a> );
-<a name="l00269"></a>00269                 vec &amp;<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a> = pred.<a class="code" href="classeDirich.html#6409d0362143a23976b43641ff19e53a" title="access function">_beta</a>();
-<a name="l00270"></a>00270 
-<a name="l00271"></a>00271                 <span class="keywordtype">double</span> lll;
-<a name="l00272"></a>00272                 <span class="keywordflow">if</span> ( <a class="code" href="classBMEF.html#538d632e59f9afa8daa1de74da12ce71" title="forgetting factor">frg</a>&lt;1.0 )
-<a name="l00273"></a>00273                         {beta*=<a class="code" href="classBMEF.html#538d632e59f9afa8daa1de74da12ce71" title="forgetting factor">frg</a>;lll=pred.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>();}
-<a name="l00274"></a>00274                 <span class="keywordflow">else</span>
-<a name="l00275"></a>00275                         <span class="keywordflow">if</span> ( <a class="code" href="classBM.html#bf6fb59b30141074f8ee1e2f43d03129" title="If true, the filter will compute likelihood of the data record and store it in ll...">evalll</a> ) {lll=<a class="code" href="classBMEF.html#308cf5d4133cd471fdf1ecd5dfa09d02" title="cached value of lognc() in the previous step (used in evaluation of ll )">last_lognc</a>;}
-<a name="l00276"></a>00276                         <span class="keywordflow">else</span>{lll=pred.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>();}
-<a name="l00277"></a>00277 
-<a name="l00278"></a>00278                 beta+=dt;
-<a name="l00279"></a>00279                 <span class="keywordflow">return</span> pred.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>()-lll;
-<a name="l00280"></a>00280         }
-<a name="l00281"></a><a class="code" href="classmultiBM.html#3988322f8f51b153622036f461f62a67">00281</a>         <span class="keywordtype">void</span> <a class="code" href="classmultiBM.html#3988322f8f51b153622036f461f62a67" title="Flatten the posterior according to the given BMEF (of the same type!).">flatten</a> ( <span class="keyword">const</span> <a class="code" href="classBMEF.html" title="Estimator for Exponential family.">BMEF</a>* B ) {
-<a name="l00282"></a>00282                 <span class="keyword">const</span> <a class="code" href="classmultiBM.html" title="Estimator for Multinomial density.">multiBM</a>* E=<span class="keyword">dynamic_cast&lt;</span><span class="keyword">const </span><a class="code" href="classmultiBM.html" title="Estimator for Multinomial density.">multiBM</a>*<span class="keyword">&gt;</span> ( B );
-<a name="l00283"></a>00283                 <span class="comment">// sum(beta) should be equal to sum(B.beta)</span>
-<a name="l00284"></a>00284                 <span class="keyword">const</span> vec &amp;Eb=E-&gt;<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>;<span class="comment">//const_cast&lt;multiBM*&gt; ( E )-&gt;_beta();</span>
-<a name="l00285"></a>00285                 <a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>*= ( sum ( Eb ) /sum ( <a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a> ) );
-<a name="l00286"></a>00286                 <span class="keywordflow">if</span> ( <a class="code" href="classBM.html#bf6fb59b30141074f8ee1e2f43d03129" title="If true, the filter will compute likelihood of the data record and store it in ll...">evalll</a> ) {<a class="code" href="classBMEF.html#308cf5d4133cd471fdf1ecd5dfa09d02" title="cached value of lognc() in the previous step (used in evaluation of ll )">last_lognc</a>=<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>();}
-<a name="l00287"></a>00287         }
-<a name="l00288"></a><a class="code" href="classmultiBM.html#66cdfd83a70bc281840ab0646b941684">00288</a>         <span class="keyword">const</span> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>&amp; <a class="code" href="classmultiBM.html#66cdfd83a70bc281840ab0646b941684" title="Returns a reference to the epdf representing posterior density on parameters.">_epdf</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> <a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>;};
-<a name="l00289"></a><a class="code" href="classmultiBM.html#66a0fa6966e40bb6c3e7ba22d26e9d35">00289</a>         <span class="keyword">const</span> <a class="code" href="classeDirich.html" title="Dirichlet posterior density.">eDirich</a>* <a class="code" href="classmultiBM.html#66a0fa6966e40bb6c3e7ba22d26e9d35" title="Returns a pointer to the epdf representing posterior density on parameters. Use with...">_e</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> &amp;<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>;};
-<a name="l00290"></a>00290         <span class="keywordtype">void</span> set_parameters ( <span class="keyword">const</span> vec &amp;beta0 ) {
-<a name="l00291"></a>00291                 <a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>.<a class="code" href="classeDirich.html#c842acb2e1cce5cc9000769ff06c086d" title="Set internal parameters.">set_parameters</a> ( beta0 );
-<a name="l00292"></a>00292                 <a class="code" href="classBM.html#af00f0612fabe66241dd507188cdbf88" title="Random variable of the posterior.">rv</a> = <a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>.<a class="code" href="classepdf.html#ca0d32aabb4cbba347e0c37fe8607562" title="access function, possibly dangerous!">_rv</a>();
-<a name="l00293"></a>00293                 <span class="keywordflow">if</span> ( <a class="code" href="classBM.html#bf6fb59b30141074f8ee1e2f43d03129" title="If true, the filter will compute likelihood of the data record and store it in ll...">evalll</a> ) {<a class="code" href="classBMEF.html#308cf5d4133cd471fdf1ecd5dfa09d02" title="cached value of lognc() in the previous step (used in evaluation of ll )">last_lognc</a>=<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>();}
-<a name="l00294"></a>00294         }
-<a name="l00295"></a>00295 };
-<a name="l00296"></a>00296 
-<a name="l00306"></a><a class="code" href="classegamma.html">00306</a> <span class="keyword">class </span><a class="code" href="classegamma.html" title="Gamma posterior density.">egamma</a> : <span class="keyword">public</span> <a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> {
-<a name="l00307"></a>00307 <span class="keyword">protected</span>:
-<a name="l00309"></a><a class="code" href="classegamma.html#376cebd8932546c440f21b182910b01b">00309</a>         vec <a class="code" href="classegamma.html#376cebd8932546c440f21b182910b01b" title="Vector .">alpha</a>;
-<a name="l00311"></a><a class="code" href="classegamma.html#cfc5f136467488a421ab22f886323790">00311</a>         vec <a class="code" href="classegamma.html#cfc5f136467488a421ab22f886323790" title="Vector .">beta</a>;
-<a name="l00312"></a>00312 <span class="keyword">public</span> :
-<a name="l00314"></a><a class="code" href="classegamma.html#4b1d34f3b244ea51a58ec10c468788c1">00314</a>         <a class="code" href="classegamma.html#4b1d34f3b244ea51a58ec10c468788c1" title="Default constructor.">egamma</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a> ) :<a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> ( rv ) {};
-<a name="l00316"></a><a class="code" href="classegamma.html#8e348b89be82b70471fe8c5630f61339">00316</a>         <span class="keywordtype">void</span> <a class="code" href="classegamma.html#8e348b89be82b70471fe8c5630f61339" title="Sets parameters.">set_parameters</a> ( <span class="keyword">const</span> vec &amp;a, <span class="keyword">const</span> vec &amp;b ) {<a class="code" href="classegamma.html#376cebd8932546c440f21b182910b01b" title="Vector .">alpha</a>=a,<a class="code" href="classegamma.html#cfc5f136467488a421ab22f886323790" title="Vector .">beta</a>=b;};
-<a name="l00317"></a>00317         vec <a class="code" href="classegamma.html#8e10c0021b5dfdd9cb62c6959b5ef425" title="Returns a sample,  from density .">sample</a>() <span class="keyword">const</span>;
-<a name="l00319"></a>00319 <span class="comment">//      mat sample ( int N ) const;</span>
-<a name="l00320"></a>00320         <span class="keywordtype">double</span> <a class="code" href="classegamma.html#de84faac8f9799dfe2777ddbedf997ef" title="TODO: is it used anywhere?">evalpdflog</a> ( <span class="keyword">const</span> vec &amp;val ) <span class="keyword">const</span>;
-<a name="l00321"></a>00321         <span class="keywordtype">double</span> <a class="code" href="classegamma.html#d6dbbdb72360f9e54d64501f80318bb6" title="logarithm of the normalizing constant, ">lognc</a> () <span class="keyword">const</span>;
-<a name="l00323"></a><a class="code" href="classegamma.html#44445c56e60b91b377f207f8d5089790">00323</a>         <span class="keywordtype">void</span> <a class="code" href="classegamma.html#44445c56e60b91b377f207f8d5089790" title="Returns poiter to alpha and beta. Potentially dengerous: use with care!">_param</a> ( vec* &amp;a, vec* &amp;b ) {a=&amp;<a class="code" href="classegamma.html#376cebd8932546c440f21b182910b01b" title="Vector .">alpha</a>;b=&amp;<a class="code" href="classegamma.html#cfc5f136467488a421ab22f886323790" title="Vector .">beta</a>;};
-<a name="l00324"></a><a class="code" href="classegamma.html#6ab5ba56f7cdb2e5921c3e77524fa50a">00324</a>         vec <a class="code" href="classegamma.html#6ab5ba56f7cdb2e5921c3e77524fa50a" title="return expected value">mean</a>()<span class="keyword"> const </span>{vec pom ( <a class="code" href="classegamma.html#376cebd8932546c440f21b182910b01b" title="Vector .">alpha</a> ); pom/=<a class="code" href="classegamma.html#cfc5f136467488a421ab22f886323790" title="Vector .">beta</a>; <span class="keywordflow">return</span> pom;}
-<a name="l00325"></a>00325 };
-<a name="l00326"></a>00326 <span class="comment">/*</span>
-<a name="l00328"></a>00328 <span class="comment">class emix : public epdf {</span>
-<a name="l00329"></a>00329 <span class="comment">protected:</span>
-<a name="l00330"></a>00330 <span class="comment">        int n;</span>
-<a name="l00331"></a>00331 <span class="comment">        vec &amp;w;</span>
-<a name="l00332"></a>00332 <span class="comment">        Array&lt;epdf*&gt; Coms;</span>
-<a name="l00333"></a>00333 <span class="comment">public:</span>
-<a name="l00335"></a>00335 <span class="comment">        emix ( const RV &amp;rv, vec &amp;w0): epdf(rv), n(w0.length()), w(w0), Coms(n) {};</span>
-<a name="l00336"></a>00336 <span class="comment">        void set_parameters( int &amp;i, double wi, epdf* ep){w(i)=wi;Coms(i)=ep;}</span>
-<a name="l00337"></a>00337 <span class="comment">        vec mean(){vec pom; for(int i=0;i&lt;n;i++){pom+=Coms(i)-&gt;mean()*w(i);} return pom;};</span>
-<a name="l00338"></a>00338 <span class="comment">        vec sample() {it_error ( "Not implemented" );return 0;}</span>
-<a name="l00339"></a>00339 <span class="comment">};</span>
-<a name="l00340"></a>00340 <span class="comment">*/</span>
-<a name="l00341"></a>00341 
-<a name="l00343"></a>00343 
-<a name="l00344"></a><a class="code" href="classeuni.html">00344</a> <span class="keyword">class </span><a class="code" href="classeuni.html" title="Uniform distributed density on a rectangular support.">euni</a>: <span class="keyword">public</span> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a> {
-<a name="l00345"></a>00345 <span class="keyword">protected</span>:
-<a name="l00347"></a><a class="code" href="classeuni.html#ef42cd8d7645422048d46c46ec5cdac1">00347</a>         vec <a class="code" href="classeuni.html#ef42cd8d7645422048d46c46ec5cdac1" title="lower bound on support">low</a>;
-<a name="l00349"></a><a class="code" href="classeuni.html#71b6d6b41aeb61a7f76f682b72119231">00349</a>         vec <a class="code" href="classeuni.html#71b6d6b41aeb61a7f76f682b72119231" title="upper bound on support">high</a>;
-<a name="l00351"></a><a class="code" href="classeuni.html#52a6ff4a54010f88a6a19fca605c64a4">00351</a>         vec <a class="code" href="classeuni.html#52a6ff4a54010f88a6a19fca605c64a4" title="internal">distance</a>;
-<a name="l00353"></a><a class="code" href="classeuni.html#63105490e946e43372d6187ad1bafdda">00353</a>         <span class="keywordtype">double</span> <a class="code" href="classeuni.html#63105490e946e43372d6187ad1bafdda" title="normalizing coefficients">nk</a>;
-<a name="l00355"></a><a class="code" href="classeuni.html#f445a0ce24f39d14c1a4eed53fc8e2c3">00355</a>         <span class="keywordtype">double</span> <a class="code" href="classeuni.html#f445a0ce24f39d14c1a4eed53fc8e2c3" title="cache of log( nk )">lnk</a>;
-<a name="l00356"></a>00356 <span class="keyword">public</span>:
-<a name="l00358"></a><a class="code" href="classeuni.html#2537a6c239cff52e3ba814851a1116cd">00358</a>         <a class="code" href="classeuni.html#2537a6c239cff52e3ba814851a1116cd" title="Defualt constructor.">euni</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a> ) :<a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a> ( rv ) {}
-<a name="l00359"></a>00359         <span class="keywordtype">double</span> eval ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const  </span>{<span class="keywordflow">return</span> <a class="code" href="classeuni.html#63105490e946e43372d6187ad1bafdda" title="normalizing coefficients">nk</a>;}
-<a name="l00360"></a><a class="code" href="classeuni.html#06af95d514a6623ad4688bd2ad50ad71">00360</a>         <span class="keywordtype">double</span> <a class="code" href="classeuni.html#06af95d514a6623ad4688bd2ad50ad71" title="Compute log-probability of argument val.">evalpdflog</a> ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const  </span>{<span class="keywordflow">return</span> <a class="code" href="classeuni.html#f445a0ce24f39d14c1a4eed53fc8e2c3" title="cache of log( nk )">lnk</a>;}
-<a name="l00361"></a><a class="code" href="classeuni.html#4a0e09392be17beaee120ba98fc038cd">00361</a>         vec <a class="code" href="classeuni.html#4a0e09392be17beaee120ba98fc038cd" title="Returns a sample,  from density .">sample</a>()<span class="keyword"> const </span>{
-<a name="l00362"></a>00362                 vec smp ( <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() );
-<a name="l00363"></a>00363 <span class="preprocessor">#pragma omp critical</span>
-<a name="l00364"></a>00364 <span class="preprocessor"></span>                UniRNG.sample_vector ( <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>(),smp );
-<a name="l00365"></a>00365                 <span class="keywordflow">return</span> <a class="code" href="classeuni.html#ef42cd8d7645422048d46c46ec5cdac1" title="lower bound on support">low</a>+elem_mult ( <a class="code" href="classeuni.html#52a6ff4a54010f88a6a19fca605c64a4" title="internal">distance</a>,smp );
-<a name="l00366"></a>00366         }
-<a name="l00368"></a><a class="code" href="classeuni.html#4fd7c6a05100616ad16ece405cad7bf2">00368</a>         <span class="keywordtype">void</span> <a class="code" href="classeuni.html#4fd7c6a05100616ad16ece405cad7bf2" title="set values of low and high ">set_parameters</a> ( <span class="keyword">const</span> vec &amp;low0, <span class="keyword">const</span> vec &amp;high0 ) {
-<a name="l00369"></a>00369                 <a class="code" href="classeuni.html#52a6ff4a54010f88a6a19fca605c64a4" title="internal">distance</a> = high0-low0;
-<a name="l00370"></a>00370                 it_assert_debug ( min ( <a class="code" href="classeuni.html#52a6ff4a54010f88a6a19fca605c64a4" title="internal">distance</a> ) &gt;0.0,<span class="stringliteral">"bad support"</span> );
-<a name="l00371"></a>00371                 <a class="code" href="classeuni.html#ef42cd8d7645422048d46c46ec5cdac1" title="lower bound on support">low</a> = low0;
-<a name="l00372"></a>00372                 <a class="code" href="classeuni.html#71b6d6b41aeb61a7f76f682b72119231" title="upper bound on support">high</a> = high0;
-<a name="l00373"></a>00373                 <a class="code" href="classeuni.html#63105490e946e43372d6187ad1bafdda" title="normalizing coefficients">nk</a> = prod ( 1.0/<a class="code" href="classeuni.html#52a6ff4a54010f88a6a19fca605c64a4" title="internal">distance</a> );
-<a name="l00374"></a>00374                 <a class="code" href="classeuni.html#f445a0ce24f39d14c1a4eed53fc8e2c3" title="cache of log( nk )">lnk</a> = log ( <a class="code" href="classeuni.html#63105490e946e43372d6187ad1bafdda" title="normalizing coefficients">nk</a> );
-<a name="l00375"></a>00375         }
-<a name="l00376"></a><a class="code" href="classeuni.html#8050087e421a9cfd1b4b1f8bd33b1cc1">00376</a>         vec <a class="code" href="classeuni.html#8050087e421a9cfd1b4b1f8bd33b1cc1" title="return expected value">mean</a>()<span class="keyword"> const </span>{vec pom=<a class="code" href="classeuni.html#71b6d6b41aeb61a7f76f682b72119231" title="upper bound on support">high</a>; pom-=<a class="code" href="classeuni.html#ef42cd8d7645422048d46c46ec5cdac1" title="lower bound on support">low</a>; pom/=2.0; <span class="keywordflow">return</span> pom;}
-<a name="l00377"></a>00377 };
-<a name="l00378"></a>00378 
-<a name="l00379"></a>00379 
-<a name="l00385"></a>00385 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00386"></a><a class="code" href="classmlnorm.html">00386</a> <span class="keyword">class </span><a class="code" href="classmlnorm.html" title="Normal distributed linear function with linear function of mean value;.">mlnorm</a> : <span class="keyword">public</span> <a class="code" href="classmEF.html" title="Exponential family model.">mEF</a> {
-<a name="l00387"></a>00387 <span class="keyword">protected</span>:
-<a name="l00389"></a><a class="code" href="classmlnorm.html#b76ee2171ace4fb3ff95a131ae8fc421">00389</a>         <a class="code" href="classenorm.html" title="Gaussian density with positive definite (decomposed) covariance matrix.">enorm&lt;sq_T&gt;</a> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>;
-<a name="l00390"></a>00390         mat A;
-<a name="l00391"></a>00391         vec mu_const;
-<a name="l00392"></a>00392         vec&amp; _mu; <span class="comment">//cached epdf.mu;</span>
-<a name="l00393"></a>00393 <span class="keyword">public</span>:
-<a name="l00395"></a>00395         <a class="code" href="classmlnorm.html#3a5ad4798d8a3878c5e93b8e796c8837" title="Constructor.">mlnorm</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classmpdf.html#f6687c07ff07d47812dd565368ca59eb" title="modeled random variable">rv</a>, <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classmpdf.html#acb7dda792b3cd5576f39fa3129abbab" title="random variable in condition">rvc</a> );
-<a name="l00397"></a>00397         <span class="keywordtype">void</span> <a class="code" href="classmlnorm.html#f95dfce0b500636a44ecd7e5210de999" title="Set A and R.">set_parameters</a> ( <span class="keyword">const</span>  mat &amp;A, <span class="keyword">const</span> vec &amp;mu0, <span class="keyword">const</span> sq_T &amp;R );
-<a name="l00398"></a>00398 <span class="comment">//      //!Generate one sample of the posterior</span>
-<a name="l00399"></a>00399 <span class="comment">//      vec samplecond (const vec &amp;cond, double &amp;lik );</span>
-<a name="l00400"></a>00400 <span class="comment">//      //!Generate matrix of samples of the posterior</span>
-<a name="l00401"></a>00401 <span class="comment">//      mat samplecond (const vec &amp;cond, vec &amp;lik, int n );</span>
-<a name="l00403"></a>00403 <span class="comment"></span>        <span class="keywordtype">void</span> <a class="code" href="classmlnorm.html#d41126455ac64b888a38f677886e1b40" title="Set value of rvc . Result of this operation is stored in epdf use function _ep to...">condition</a> ( <span class="keyword">const</span> vec &amp;cond );
-<a name="l00404"></a>00404 
-<a name="l00406"></a><a class="code" href="classmlnorm.html#2732ae47835dd25d5784bf08fde0a546">00406</a>         vec&amp; <a class="code" href="classmlnorm.html#2732ae47835dd25d5784bf08fde0a546" title="access function">_mu_const</a>() {<span class="keywordflow">return</span> mu_const;}
-<a name="l00408"></a><a class="code" href="classmlnorm.html#65ec3840c21b21102896bfd2282b47b3">00408</a>         mat&amp; <a class="code" href="classmlnorm.html#65ec3840c21b21102896bfd2282b47b3" title="access function">_A</a>() {<span class="keywordflow">return</span> A;}
-<a name="l00410"></a><a class="code" href="classmlnorm.html#3ff2b03fbb5e1133a5fe1bf831939f63">00410</a>         mat <a class="code" href="classmlnorm.html#3ff2b03fbb5e1133a5fe1bf831939f63" title="access function">_R</a>() {<span class="keywordflow">return</span> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>._R().to_mat();}
-<a name="l00411"></a>00411 
-<a name="l00412"></a>00412         <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_M&gt;
-<a name="l00413"></a>00413         <span class="keyword">friend</span> std::ostream &amp;operator&lt;&lt; ( std::ostream &amp;os,  mlnorm&lt;sq_M&gt; &amp;ml );
-<a name="l00414"></a>00414 };
-<a name="l00415"></a>00415 
-<a name="l00418"></a><a class="code" href="classmlstudent.html">00418</a> <span class="keyword">class </span><a class="code" href="classmlstudent.html">mlstudent</a> : <span class="keyword">public</span> <a class="code" href="classmlnorm.html" title="Normal distributed linear function with linear function of mean value;.">mlnorm</a>&lt;ldmat&gt; {
-<a name="l00419"></a>00419 <span class="keyword">protected</span>:
-<a name="l00420"></a>00420         <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a> Lambda;
-<a name="l00421"></a>00421         <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a> &amp;<a class="code" href="classmlnorm.html#3ff2b03fbb5e1133a5fe1bf831939f63" title="access function">_R</a>;
-<a name="l00422"></a>00422         <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a> Re;
-<a name="l00423"></a>00423 <span class="keyword">public</span>:
-<a name="l00424"></a>00424         <a class="code" href="classmlstudent.html">mlstudent</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rv0, <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rvc0 ) :<a class="code" href="classmlnorm.html" title="Normal distributed linear function with linear function of mean value;.">mlnorm&lt;ldmat&gt;</a> ( rv0,rvc0 ),
-<a name="l00425"></a>00425                         Lambda ( rv0.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ),
-<a name="l00426"></a>00426                         <a class="code" href="classmlnorm.html#3ff2b03fbb5e1133a5fe1bf831939f63" title="access function">_R</a> ( <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>._R() ) {}
-<a name="l00427"></a>00427         <span class="keywordtype">void</span> set_parameters ( <span class="keyword">const</span> mat &amp;A0, <span class="keyword">const</span> vec &amp;mu0, <span class="keyword">const</span> <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a> &amp;R0, <span class="keyword">const</span> <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a>&amp; Lambda0) {
-<a name="l00428"></a>00428                 <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>.set_parameters ( zeros ( <a class="code" href="classmpdf.html#f6687c07ff07d47812dd565368ca59eb" title="modeled random variable">rv</a>.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ),Lambda );
-<a name="l00429"></a>00429                 A = A0;
-<a name="l00430"></a>00430                 mu_const = mu0;
-<a name="l00431"></a>00431                 Re=R0;
-<a name="l00432"></a>00432                 Lambda = Lambda0;
-<a name="l00433"></a>00433         }
-<a name="l00434"></a><a class="code" href="classmlstudent.html#d153460ae0180f4bc7f28301f5cde876">00434</a>         <span class="keywordtype">void</span> <a class="code" href="classmlstudent.html#d153460ae0180f4bc7f28301f5cde876" title="Set value of rvc . Result of this operation is stored in epdf use function _ep to...">condition</a> ( <span class="keyword">const</span> vec &amp;cond ) {
-<a name="l00435"></a>00435                 _mu = A*cond + mu_const;
-<a name="l00436"></a>00436                 <span class="keywordtype">double</span> zeta;
-<a name="l00437"></a>00437                 <span class="comment">//ugly hack!</span>
-<a name="l00438"></a>00438                 <span class="keywordflow">if</span> ((cond.length()+1)==Lambda.<a class="code" href="classldmat.html#96dfb21865db4f5bd36fa70f9b0b1163" title="access function">rows</a>()){
-<a name="l00439"></a>00439                         zeta = Lambda.<a class="code" href="classldmat.html#d876c5f83e02b3e809b35c9de5068f14" title="Evaluates quadratic form ;.">invqform</a> ( concat(cond, vec_1(1.0)) );
-<a name="l00440"></a>00440                 } <span class="keywordflow">else</span> {
-<a name="l00441"></a>00441                         zeta = Lambda.<a class="code" href="classldmat.html#d876c5f83e02b3e809b35c9de5068f14" title="Evaluates quadratic form ;.">invqform</a> ( cond );
-<a name="l00442"></a>00442                 }
-<a name="l00443"></a>00443                 <a class="code" href="classmlnorm.html#3ff2b03fbb5e1133a5fe1bf831939f63" title="access function">_R</a> = Re;
-<a name="l00444"></a>00444                 <a class="code" href="classmlnorm.html#3ff2b03fbb5e1133a5fe1bf831939f63" title="access function">_R</a>*=( 1+zeta );<span class="comment">// / ( nu ); &lt;&lt; nu is in Re!!!!!!</span>
-<a name="l00445"></a>00445         };
-<a name="l00446"></a>00446 
-<a name="l00447"></a>00447 };
-<a name="l00457"></a><a class="code" href="classmgamma.html">00457</a> <span class="keyword">class </span><a class="code" href="classmgamma.html" title="Gamma random walk.">mgamma</a> : <span class="keyword">public</span> <a class="code" href="classmEF.html" title="Exponential family model.">mEF</a> {
-<a name="l00458"></a>00458 <span class="keyword">protected</span>:
-<a name="l00460"></a><a class="code" href="classmgamma.html#612dbf35c770a780027619aaac2c443e">00460</a>         <a class="code" href="classegamma.html" title="Gamma posterior density.">egamma</a> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>;
-<a name="l00462"></a><a class="code" href="classmgamma.html#43f733cce0245a52363d566099add687">00462</a>         <span class="keywordtype">double</span> <a class="code" href="classmgamma.html#43f733cce0245a52363d566099add687" title="Constant .">k</a>;
-<a name="l00464"></a><a class="code" href="classmgamma.html#5e90652837448bcc29707e7412f99691">00464</a>         vec* <a class="code" href="classmgamma.html#5e90652837448bcc29707e7412f99691" title="cache of epdf.beta">_beta</a>;
-<a name="l00465"></a>00465 
-<a name="l00466"></a>00466 <span class="keyword">public</span>:
-<a name="l00468"></a>00468         <a class="code" href="classmgamma.html#af43e61b86900c0398d5c0ffc83b94e6" title="Constructor.">mgamma</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classmpdf.html#f6687c07ff07d47812dd565368ca59eb" title="modeled random variable">rv</a>,<span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classmpdf.html#acb7dda792b3cd5576f39fa3129abbab" title="random variable in condition">rvc</a> );
-<a name="l00470"></a>00470         <span class="keywordtype">void</span> <a class="code" href="classmgamma.html#a9d646cf758a70126dde7c48790b6e94" title="Set value of k.">set_parameters</a> ( <span class="keywordtype">double</span> <a class="code" href="classmgamma.html#43f733cce0245a52363d566099add687" title="Constant .">k</a> );
-<a name="l00471"></a><a class="code" href="classmgamma.html#a61094c9f7a2d64ea77b130cbc031f97">00471</a>         <span class="keywordtype">void</span> <a class="code" href="classmgamma.html#a61094c9f7a2d64ea77b130cbc031f97" title="Update ep so that it represents this mpdf conditioned on rvc = cond.">condition</a> ( <span class="keyword">const</span> vec &amp;val ) {*<a class="code" href="classmgamma.html#5e90652837448bcc29707e7412f99691" title="cache of epdf.beta">_beta</a>=<a class="code" href="classmgamma.html#43f733cce0245a52363d566099add687" title="Constant .">k</a>/val;};
-<a name="l00472"></a>00472 };
-<a name="l00473"></a>00473 
-<a name="l00485"></a><a class="code" href="classmgamma__fix.html">00485</a> <span class="keyword">class </span><a class="code" href="classmgamma__fix.html" title="Gamma random walk around a fixed point.">mgamma_fix</a> : <span class="keyword">public</span> <a class="code" href="classmgamma.html" title="Gamma random walk.">mgamma</a> {
-<a name="l00486"></a>00486 <span class="keyword">protected</span>:
-<a name="l00488"></a><a class="code" href="classmgamma__fix.html#3f48c09caddc298901ad75fe7c0529f6">00488</a>         <span class="keywordtype">double</span> <a class="code" href="classmgamma__fix.html#3f48c09caddc298901ad75fe7c0529f6" title="parameter l">l</a>;
-<a name="l00490"></a><a class="code" href="classmgamma__fix.html#81ce49029ecc385418619b200dcafeb0">00490</a>         vec <a class="code" href="classmgamma__fix.html#81ce49029ecc385418619b200dcafeb0" title="reference vector">refl</a>;
-<a name="l00491"></a>00491 <span class="keyword">public</span>:
-<a name="l00493"></a><a class="code" href="classmgamma__fix.html#b92c3d2e5fd0381033a072e5ef3bcf80">00493</a>         <a class="code" href="classmgamma__fix.html#b92c3d2e5fd0381033a072e5ef3bcf80" title="Constructor.">mgamma_fix</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classmpdf.html#f6687c07ff07d47812dd565368ca59eb" title="modeled random variable">rv</a>,<span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classmpdf.html#acb7dda792b3cd5576f39fa3129abbab" title="random variable in condition">rvc</a> ) : <a class="code" href="classmgamma.html" title="Gamma random walk.">mgamma</a> ( rv,rvc ),<a class="code" href="classmgamma__fix.html#81ce49029ecc385418619b200dcafeb0" title="reference vector">refl</a> ( rv.count() ) {};
-<a name="l00495"></a><a class="code" href="classmgamma__fix.html#ec6f846896749e27cb7be9fa48dd1cb1">00495</a>         <span class="keywordtype">void</span> <a class="code" href="classmgamma__fix.html#ec6f846896749e27cb7be9fa48dd1cb1" title="Set value of k.">set_parameters</a> ( <span class="keywordtype">double</span> k0 , vec ref0, <span class="keywordtype">double</span> l0 ) {
-<a name="l00496"></a>00496                 <a class="code" href="classmgamma.html#a9d646cf758a70126dde7c48790b6e94" title="Set value of k.">mgamma::set_parameters</a> ( k0 );
-<a name="l00497"></a>00497                 <a class="code" href="classmgamma__fix.html#81ce49029ecc385418619b200dcafeb0" title="reference vector">refl</a>=pow ( ref0,1.0-l0 );<a class="code" href="classmgamma__fix.html#3f48c09caddc298901ad75fe7c0529f6" title="parameter l">l</a>=l0;
-<a name="l00498"></a>00498         };
-<a name="l00499"></a>00499 
-<a name="l00500"></a><a class="code" href="classmgamma__fix.html#6ea3931eec7b7da7b693e45981052460">00500</a>         <span class="keywordtype">void</span> <a class="code" href="classmgamma__fix.html#6ea3931eec7b7da7b693e45981052460" title="Update ep so that it represents this mpdf conditioned on rvc = cond.">condition</a> ( <span class="keyword">const</span> vec &amp;val ) {vec mean=elem_mult ( <a class="code" href="classmgamma__fix.html#81ce49029ecc385418619b200dcafeb0" title="reference vector">refl</a>,pow ( val,<a class="code" href="classmgamma__fix.html#3f48c09caddc298901ad75fe7c0529f6" title="parameter l">l</a> ) ); *<a class="code" href="classmgamma.html#5e90652837448bcc29707e7412f99691" title="cache of epdf.beta">_beta</a>=<a class="code" href="classmgamma.html#43f733cce0245a52363d566099add687" title="Constant .">k</a>/mean;};
-<a name="l00501"></a>00501 };
-<a name="l00502"></a>00502 
-<a name="l00504"></a><a class="code" href="libEF_8h.html#99497a3ff630f761cf6bff7babd23212">00504</a> <span class="keyword">enum</span> <a class="code" href="libEF_8h.html#99497a3ff630f761cf6bff7babd23212" title="Switch between various resampling methods.">RESAMPLING_METHOD</a> { MULTINOMIAL = 0, STRATIFIED = 1, SYSTEMATIC = 3 };
-<a name="l00510"></a><a class="code" href="classeEmp.html">00510</a> <span class="keyword">class </span><a class="code" href="classeEmp.html" title="Weighted empirical density.">eEmp</a>: <span class="keyword">public</span> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a> {
-<a name="l00511"></a>00511 <span class="keyword">protected</span> :
-<a name="l00513"></a><a class="code" href="classeEmp.html#8c33034de0e35f03f8bb85d3d67438fd">00513</a>         <span class="keywordtype">int</span> <a class="code" href="classeEmp.html#8c33034de0e35f03f8bb85d3d67438fd" title="Number of particles.">n</a>;
-<a name="l00515"></a><a class="code" href="classeEmp.html#ae78d144404ddba843c93b171b215de8">00515</a>         vec <a class="code" href="classeEmp.html#ae78d144404ddba843c93b171b215de8" title="Sample weights .">w</a>;
-<a name="l00517"></a><a class="code" href="classeEmp.html#a4d6f4bbd6a6824fc39f14676701279a">00517</a>         Array&lt;vec&gt; <a class="code" href="classeEmp.html#a4d6f4bbd6a6824fc39f14676701279a" title="Samples .">samples</a>;
-<a name="l00518"></a>00518 <span class="keyword">public</span>:
-<a name="l00520"></a><a class="code" href="classeEmp.html#0c04b073ecd0dae3d498e680ae27e9e4">00520</a>         <a class="code" href="classeEmp.html#0c04b073ecd0dae3d498e680ae27e9e4" title="Default constructor.">eEmp</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rv0 ,<span class="keywordtype">int</span> n0 ) :<a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a> ( rv0 ),<a class="code" href="classeEmp.html#8c33034de0e35f03f8bb85d3d67438fd" title="Number of particles.">n</a> ( n0 ),<a class="code" href="classeEmp.html#ae78d144404ddba843c93b171b215de8" title="Sample weights .">w</a> ( <a class="code" href="classeEmp.html#8c33034de0e35f03f8bb85d3d67438fd" title="Number of particles.">n</a> ),<a class="code" href="classeEmp.html#a4d6f4bbd6a6824fc39f14676701279a" title="Samples .">samples</a> ( <a class="code" href="classeEmp.html#8c33034de0e35f03f8bb85d3d67438fd" title="Number of particles.">n</a> ) {};
-<a name="l00522"></a>00522         <span class="keywordtype">void</span> <a class="code" href="classeEmp.html#eab03bd3381aaea11ce34d5a26556353" title="Set samples and weights.">set_parameters</a> ( <span class="keyword">const</span> vec &amp;w0, <span class="keyword">const</span> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>* pdf0 );
-<a name="l00524"></a>00524         <span class="keywordtype">void</span> <a class="code" href="classeEmp.html#e31bc9e6196173c3480b06a761a3e716" title="Set sample.">set_samples</a> ( <span class="keyword">const</span> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>* pdf0 );
-<a name="l00526"></a><a class="code" href="classeEmp.html#a4215f6a5a04d07b43f7ebaa942e15f1">00526</a>         <span class="keywordtype">void</span> <a class="code" href="classeEmp.html#a4215f6a5a04d07b43f7ebaa942e15f1" title="Set sample.">set_n</a> ( <span class="keywordtype">int</span> n0, <span class="keywordtype">bool</span> copy=<span class="keyword">true</span> ){<a class="code" href="classeEmp.html#ae78d144404ddba843c93b171b215de8" title="Sample weights .">w</a>.set_size(n0,copy);<a class="code" href="classeEmp.html#a4d6f4bbd6a6824fc39f14676701279a" title="Samples .">samples</a>.set_size(n0,copy);};
-<a name="l00528"></a><a class="code" href="classeEmp.html#31b2bfb73b72486a5c89f2ab850c7a9b">00528</a>         vec&amp; <a class="code" href="classeEmp.html#31b2bfb73b72486a5c89f2ab850c7a9b" title="Potentially dangerous, use with care.">_w</a>()  {<span class="keywordflow">return</span> <a class="code" href="classeEmp.html#ae78d144404ddba843c93b171b215de8" title="Sample weights .">w</a>;};
-<a name="l00530"></a><a class="code" href="classeEmp.html#d6f4ae1a67ecd2bff8b9f176ee261afc">00530</a>         <span class="keyword">const</span> vec&amp; <a class="code" href="classeEmp.html#31b2bfb73b72486a5c89f2ab850c7a9b" title="Potentially dangerous, use with care.">_w</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> <a class="code" href="classeEmp.html#ae78d144404ddba843c93b171b215de8" title="Sample weights .">w</a>;};
-<a name="l00532"></a><a class="code" href="classeEmp.html#31b747eca73b16f30370827ba4cc3575">00532</a>         Array&lt;vec&gt;&amp; <a class="code" href="classeEmp.html#31b747eca73b16f30370827ba4cc3575" title="access function">_samples</a>() {<span class="keywordflow">return</span> <a class="code" href="classeEmp.html#a4d6f4bbd6a6824fc39f14676701279a" title="Samples .">samples</a>;};
-<a name="l00534"></a><a class="code" href="classeEmp.html#bd48c1c36e2e9e78dbcea7df66dcbf25">00534</a>         <span class="keyword">const</span> Array&lt;vec&gt;&amp; <a class="code" href="classeEmp.html#31b747eca73b16f30370827ba4cc3575" title="access function">_samples</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> <a class="code" href="classeEmp.html#a4d6f4bbd6a6824fc39f14676701279a" title="Samples .">samples</a>;};
-<a name="l00536"></a>00536         ivec <a class="code" href="classeEmp.html#77268292fc4465cb73ddbfb1f2932a59" title="Function performs resampling, i.e. removal of low-weight samples and duplication...">resample</a> ( <a class="code" href="libEF_8h.html#99497a3ff630f761cf6bff7babd23212" title="Switch between various resampling methods.">RESAMPLING_METHOD</a> method = SYSTEMATIC );
-<a name="l00538"></a><a class="code" href="classeEmp.html#83f9283f92b805508d896479dc1ccf12">00538</a>         vec <a class="code" href="classeEmp.html#83f9283f92b805508d896479dc1ccf12" title="inherited operation : NOT implemneted">sample</a>()<span class="keyword"> const </span>{it_error ( <span class="stringliteral">"Not implemented"</span> );<span class="keywordflow">return</span> 0;}
-<a name="l00540"></a><a class="code" href="classeEmp.html#23e7358995400865ad2e278945922fb3">00540</a>         <span class="keywordtype">double</span> <a class="code" href="classeEmp.html#23e7358995400865ad2e278945922fb3" title="inherited operation : NOT implemneted">evalpdflog</a> ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const </span>{it_error ( <span class="stringliteral">"Not implemented"</span> );<span class="keywordflow">return</span> 0.0;}
-<a name="l00541"></a><a class="code" href="classeEmp.html#ba055c19038cc72628d98e25197e982d">00541</a>         vec <a class="code" href="classeEmp.html#ba055c19038cc72628d98e25197e982d" title="return expected value">mean</a>()<span class="keyword"> const </span>{
-<a name="l00542"></a>00542                 vec pom=zeros ( <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() );
-<a name="l00543"></a>00543                 <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i=0;i&lt;<a class="code" href="classeEmp.html#8c33034de0e35f03f8bb85d3d67438fd" title="Number of particles.">n</a>;i++ ) {pom+=<a class="code" href="classeEmp.html#a4d6f4bbd6a6824fc39f14676701279a" title="Samples .">samples</a> ( i ) *<a class="code" href="classeEmp.html#ae78d144404ddba843c93b171b215de8" title="Sample weights .">w</a> ( i );}
-<a name="l00544"></a>00544                 <span class="keywordflow">return</span> pom;
-<a name="l00545"></a>00545         }
-<a name="l00546"></a>00546 };
-<a name="l00547"></a>00547 
-<a name="l00548"></a>00548 
-<a name="l00550"></a>00550 
-<a name="l00551"></a>00551 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00552"></a><a class="code" href="classenorm.html#0caf54fed9e48f9fe28b534b2027df2f">00552</a> <a class="code" href="classenorm.html#0caf54fed9e48f9fe28b534b2027df2f" title="Default constructor.">enorm&lt;sq_T&gt;::enorm</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rv ) :<a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> ( rv ), mu ( rv.count() ),R ( rv.count() ),dim ( rv.count() ) {};
+<a name="l00147"></a>00147         <span class="keyword">const</span> sq_T&amp; <a class="code" href="classenorm.html#7a5034b25771a84450a990d10fc40ac9" title="returns pointers to the internal variance and its inverse. Use with Care!">_R</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>;}
+<a name="l00148"></a>00148 
+<a name="l00150"></a>00150 <span class="comment">//      mat getR () {return R.to_mat();}</span>
+<a name="l00151"></a>00151 };
+<a name="l00152"></a>00152 
+<a name="l00159"></a><a class="code" href="classegiw.html">00159</a> <span class="keyword">class </span><a class="code" href="classegiw.html" title="Gauss-inverse-Wishart density stored in LD form.">egiw</a> : <span class="keyword">public</span> <a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> {
+<a name="l00160"></a>00160 <span class="keyword">protected</span>:
+<a name="l00162"></a><a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442">00162</a>         <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a> <a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>;
+<a name="l00164"></a><a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453">00164</a>         <span class="keywordtype">double</span> <a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a>;
+<a name="l00166"></a><a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e">00166</a>         <span class="keywordtype">int</span> <a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a>;
+<a name="l00168"></a><a class="code" href="classegiw.html#c70d13d86e0d9f0acede3e1dc0368812">00168</a>         <span class="keywordtype">int</span> <a class="code" href="classegiw.html#c70d13d86e0d9f0acede3e1dc0368812" title="Dimension of the regressor.">nPsi</a>;
+<a name="l00169"></a>00169 <span class="keyword">public</span>:
+<a name="l00171"></a><a class="code" href="classegiw.html#056c094f01ca1cc308d72162f47617c9">00171</a>         <a class="code" href="classegiw.html#056c094f01ca1cc308d72162f47617c9" title="Default constructor, if nu0&amp;lt;0 a minimal nu0 will be computed.">egiw</a> ( <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>, mat V0, <span class="keywordtype">double</span> nu0=-1.0 ) : <a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> ( rv ), <a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a> ( V0 ), <a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a> ( nu0 ) {
+<a name="l00172"></a>00172                 <a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a> = rv.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() /<a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>.<a class="code" href="group__math.html#g96dfb21865db4f5bd36fa70f9b0b1163" title="access function">rows</a>();
+<a name="l00173"></a>00173                 it_assert_debug ( rv.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ==<a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a>*<a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>.<a class="code" href="group__math.html#g96dfb21865db4f5bd36fa70f9b0b1163" title="access function">rows</a>(),<span class="stringliteral">"Incompatible V0."</span> );
+<a name="l00174"></a>00174                 <a class="code" href="classegiw.html#c70d13d86e0d9f0acede3e1dc0368812" title="Dimension of the regressor.">nPsi</a> = <a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>.<a class="code" href="group__math.html#g96dfb21865db4f5bd36fa70f9b0b1163" title="access function">rows</a>()-<a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a>;
+<a name="l00175"></a>00175                 <span class="comment">//set mu to have proper normalization and </span>
+<a name="l00176"></a>00176                 <span class="keywordflow">if</span> (nu0&lt;0){
+<a name="l00177"></a>00177                         <a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a> = 0.1 +nPsi +2*<a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a> +2; <span class="comment">// +2 assures finite expected value of R</span>
+<a name="l00178"></a>00178                         <span class="comment">// terms before that are sufficient for finite normalization</span>
+<a name="l00179"></a>00179                 }
+<a name="l00180"></a>00180         }
+<a name="l00182"></a><a class="code" href="classegiw.html#18c1bf6125652a6dcbca68dd02dddd8d">00182</a>         <a class="code" href="classegiw.html#056c094f01ca1cc308d72162f47617c9" title="Default constructor, if nu0&amp;lt;0 a minimal nu0 will be computed.">egiw</a> ( <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>, <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a> V0, <span class="keywordtype">double</span> nu0=-1.0 ) : <a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> ( rv ), <a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a> ( V0 ), <a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a> ( nu0 ) {
+<a name="l00183"></a>00183                 <a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a> = rv.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() /<a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>.<a class="code" href="group__math.html#g96dfb21865db4f5bd36fa70f9b0b1163" title="access function">rows</a>();
+<a name="l00184"></a>00184                 it_assert_debug ( rv.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ==<a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a>*<a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>.<a class="code" href="group__math.html#g96dfb21865db4f5bd36fa70f9b0b1163" title="access function">rows</a>(),<span class="stringliteral">"Incompatible V0."</span> );
+<a name="l00185"></a>00185                 <a class="code" href="classegiw.html#c70d13d86e0d9f0acede3e1dc0368812" title="Dimension of the regressor.">nPsi</a> = <a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>.<a class="code" href="group__math.html#g96dfb21865db4f5bd36fa70f9b0b1163" title="access function">rows</a>()-<a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a>;
+<a name="l00186"></a>00186                 <span class="keywordflow">if</span> (nu0&lt;0){
+<a name="l00187"></a>00187                         <a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a> = 0.1 +nPsi +2*<a class="code" href="classegiw.html#3d5c719f15a5527a6c62c2a53160148e" title="Dimension of the output.">xdim</a> +2; <span class="comment">// +2 assures finite expected value of R</span>
+<a name="l00188"></a>00188                         <span class="comment">// terms before that are sufficient for finite normalization</span>
+<a name="l00189"></a>00189                 }
+<a name="l00190"></a>00190         }
+<a name="l00191"></a>00191 
+<a name="l00192"></a>00192         vec <a class="code" href="classegiw.html#3d2c1f2ba0f9966781f1e0ae695e8a6f" title="Returns a sample,  from density .">sample</a>() <span class="keyword">const</span>;
+<a name="l00193"></a>00193         vec <a class="code" href="classegiw.html#6deb0ff2859f41ef7cbdf6a842cabb29" title="return expected value">mean</a>() <span class="keyword">const</span>;
+<a name="l00194"></a>00194         <span class="keywordtype">void</span> mean_mat ( mat &amp;M, mat&amp;R ) <span class="keyword">const</span>;
+<a name="l00196"></a>00196         <span class="keywordtype">double</span> <a class="code" href="classegiw.html#2d94daac10d66bb743e4ddc8c1ba7268" title="In this instance, val= [theta, r]. For multivariate instances, it is stored columnwise...">evallog_nn</a> ( <span class="keyword">const</span> vec &amp;val ) <span class="keyword">const</span>;
+<a name="l00197"></a>00197         <span class="keywordtype">double</span> <a class="code" href="classegiw.html#70eb1a0b88459b227f919b425b0d3359" title="logarithm of the normalizing constant, ">lognc</a> () <span class="keyword">const</span>;
+<a name="l00198"></a>00198 
+<a name="l00199"></a>00199         <span class="comment">//Access</span>
+<a name="l00201"></a><a class="code" href="classegiw.html#533e792e1175bfa06d5d595dc5d080d5">00201</a> <span class="comment"></span>        <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a>&amp; <a class="code" href="classegiw.html#533e792e1175bfa06d5d595dc5d080d5" title="returns a pointer to the internal statistics. Use with Care!">_V</a>() {<span class="keywordflow">return</span> <a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>;}
+<a name="l00203"></a><a class="code" href="classegiw.html#a46c8a206edf80b357a138d7491780c1">00203</a>         <span class="keyword">const</span> <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a>&amp; <a class="code" href="classegiw.html#533e792e1175bfa06d5d595dc5d080d5" title="returns a pointer to the internal statistics. Use with Care!">_V</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> <a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>;}
+<a name="l00205"></a><a class="code" href="classegiw.html#08029c481ff95d24f093df0573879afe">00205</a>         <span class="keywordtype">double</span>&amp; <a class="code" href="classegiw.html#08029c481ff95d24f093df0573879afe" title="returns a pointer to the internal statistics. Use with Care!">_nu</a>()  {<span class="keywordflow">return</span> <a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a>;}
+<a name="l00206"></a>00206         <span class="keyword">const</span> <span class="keywordtype">double</span>&amp; <a class="code" href="classegiw.html#08029c481ff95d24f093df0573879afe" title="returns a pointer to the internal statistics. Use with Care!">_nu</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> <a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a>;}
+<a name="l00207"></a><a class="code" href="classegiw.html#036306322a90a9977834baac07460816">00207</a>         <span class="keywordtype">void</span> <a class="code" href="classegiw.html#036306322a90a9977834baac07460816" title="Power of the density, used e.g. to flatten the density.">pow</a> ( <span class="keywordtype">double</span> p ) {<a class="code" href="classegiw.html#f343d03ede89db820edf44a6297fa442" title="Extended information matrix of sufficient statistics.">V</a>*=p;<a class="code" href="classegiw.html#4a2f130b91afe84f6d62fed289d5d453" title="Number of data records (degrees of freedom) of sufficient statistics.">nu</a>*=p;};
+<a name="l00208"></a>00208 };
+<a name="l00209"></a>00209 
+<a name="l00218"></a><a class="code" href="classeDirich.html">00218</a> <span class="keyword">class </span><a class="code" href="classeDirich.html" title="Dirichlet posterior density.">eDirich</a>: <span class="keyword">public</span> <a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> {
+<a name="l00219"></a>00219 <span class="keyword">protected</span>:
+<a name="l00221"></a><a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7">00221</a>         vec <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>;
+<a name="l00222"></a>00222 <span class="keyword">public</span>:
+<a name="l00224"></a><a class="code" href="classeDirich.html#ac7e6116f3575c3860d07355e96cd4af">00224</a>         <a class="code" href="classeDirich.html#ac7e6116f3575c3860d07355e96cd4af" title="Default constructor.">eDirich</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>, <span class="keyword">const</span> vec &amp;beta0 ) : <a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> ( rv ),<a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a> ( beta0 ) {it_assert_debug ( rv.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ==<a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>.length(),<span class="stringliteral">"Incompatible statistics"</span> ); };
+<a name="l00226"></a><a class="code" href="classeDirich.html#55cccbc5eb44764dce722567acf5fd58">00226</a>         <a class="code" href="classeDirich.html#ac7e6116f3575c3860d07355e96cd4af" title="Default constructor.">eDirich</a> ( <span class="keyword">const</span> <a class="code" href="classeDirich.html" title="Dirichlet posterior density.">eDirich</a> &amp;D0 ) : <a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> ( D0.<a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a> ),<a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a> ( D0.<a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a> ) {};
+<a name="l00227"></a><a class="code" href="classeDirich.html#23dff79110822e9639343fe8e177fd80">00227</a>         vec <a class="code" href="classeDirich.html#23dff79110822e9639343fe8e177fd80" title="Returns a sample,  from density .">sample</a>()<span class="keyword"> const </span>{it_error ( <span class="stringliteral">"Not implemented"</span> );<span class="keywordflow">return</span> vec_1 ( 0.0 );};
+<a name="l00228"></a><a class="code" href="classeDirich.html#4206e1da149d51ff3b663c9241096b73">00228</a>         vec <a class="code" href="classeDirich.html#4206e1da149d51ff3b663c9241096b73" title="return expected value">mean</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>/sum ( <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a> );};
+<a name="l00230"></a><a class="code" href="classeDirich.html#bb4b14ed7794777386de10608a83d142">00230</a>         <span class="keywordtype">double</span> <a class="code" href="classeDirich.html#bb4b14ed7794777386de10608a83d142" title="In this instance, val is ...">evallog_nn</a> ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const </span>{<span class="keywordtype">double</span> tmp; tmp=( <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>-1 ) *log ( val );            it_assert_debug(std::isfinite(tmp),<span class="stringliteral">"Infinite value"</span>);
+<a name="l00231"></a>00231         <span class="keywordflow">return</span> tmp;};
+<a name="l00232"></a><a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77">00232</a>         <span class="keywordtype">double</span> <a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a> ()<span class="keyword"> const </span>{
+<a name="l00233"></a>00233                 <span class="keywordtype">double</span> tmp;
+<a name="l00234"></a>00234                 <span class="keywordtype">double</span> gam=sum ( <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a> );
+<a name="l00235"></a>00235                 <span class="keywordtype">double</span> lgb=0.0;
+<a name="l00236"></a>00236                 <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i=0;i&lt;<a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>.length();i++ ) {lgb+=lgamma ( <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a> ( i ) );}
+<a name="l00237"></a>00237                 tmp= lgb-lgamma ( gam );
+<a name="l00238"></a>00238                 it_assert_debug(std::isfinite(tmp),<span class="stringliteral">"Infinite value"</span>);
+<a name="l00239"></a>00239                 <span class="keywordflow">return</span> tmp;
+<a name="l00240"></a>00240         };
+<a name="l00242"></a><a class="code" href="classeDirich.html#6409d0362143a23976b43641ff19e53a">00242</a>         vec&amp; <a class="code" href="classeDirich.html#6409d0362143a23976b43641ff19e53a" title="access function">_beta</a>()  {<span class="keywordflow">return</span> <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>;}
+<a name="l00244"></a><a class="code" href="classeDirich.html#c842acb2e1cce5cc9000769ff06c086d">00244</a>         <span class="keywordtype">void</span> <a class="code" href="classeDirich.html#c842acb2e1cce5cc9000769ff06c086d" title="Set internal parameters.">set_parameters</a> ( <span class="keyword">const</span> vec &amp;beta0 ) {
+<a name="l00245"></a>00245                 <span class="keywordflow">if</span> ( beta0.length() !=<a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>.length() ) {
+<a name="l00246"></a>00246                         it_assert_debug ( <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>.<a class="code" href="classRV.html#c114a6f3ff06796cc2f4dacba74291eb" title="Return length (number of entries) of the RV.">length</a>() ==1,<span class="stringliteral">"Undefined"</span> );
+<a name="l00247"></a>00247                         <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>.<a class="code" href="classRV.html#70b24c39c5130b1e4753fa2eef495433" title="access function">set_size</a> ( 0,beta0.length() );
+<a name="l00248"></a>00248                 }
+<a name="l00249"></a>00249                 <a class="code" href="classeDirich.html#15e6b65e9595eedc8a1286c6cecd36d7" title="sufficient statistics">beta</a>= beta0;
+<a name="l00250"></a>00250         }
+<a name="l00251"></a>00251 };
+<a name="l00252"></a>00252 
+<a name="l00254"></a><a class="code" href="classmultiBM.html">00254</a> <span class="keyword">class </span><a class="code" href="classmultiBM.html" title="Estimator for Multinomial density.">multiBM</a> : <span class="keyword">public</span> <a class="code" href="classBMEF.html" title="Estimator for Exponential family.">BMEF</a> {
+<a name="l00255"></a>00255 <span class="keyword">protected</span>:
+<a name="l00257"></a><a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5">00257</a>         <a class="code" href="classeDirich.html" title="Dirichlet posterior density.">eDirich</a> <a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>;
+<a name="l00259"></a><a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6">00259</a>         vec &amp;<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>;
+<a name="l00260"></a>00260 <span class="keyword">public</span>:
+<a name="l00262"></a><a class="code" href="classmultiBM.html#7d7d7e78c129602bcde96078359dc6e5">00262</a>         <a class="code" href="classmultiBM.html#7d7d7e78c129602bcde96078359dc6e5" title="Default constructor.">multiBM</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classBM.html#af00f0612fabe66241dd507188cdbf88" title="Random variable of the posterior.">rv</a>, <span class="keyword">const</span> vec beta0 ) : <a class="code" href="classBMEF.html" title="Estimator for Exponential family.">BMEF</a> ( rv ),<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a> ( rv,beta0 ),<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a> ( <a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>._beta() ) {<span class="keywordflow">if</span>(<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>.length()&gt;0){<a class="code" href="classBMEF.html#308cf5d4133cd471fdf1ecd5dfa09d02" title="cached value of lognc() in the previous step (used in evaluation of ll )">last_lognc</a>=<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>();}<span class="keywordflow">else</span>{<a class="code" href="classBMEF.html#308cf5d4133cd471fdf1ecd5dfa09d02" title="cached value of lognc() in the previous step (used in evaluation of ll )">last_lognc</a>=0.0;}}
+<a name="l00264"></a><a class="code" href="classmultiBM.html#b92751adbfb9f259ca8c95232cfd9c09">00264</a>         <a class="code" href="classmultiBM.html#7d7d7e78c129602bcde96078359dc6e5" title="Default constructor.">multiBM</a> ( <span class="keyword">const</span> <a class="code" href="classmultiBM.html" title="Estimator for Multinomial density.">multiBM</a> &amp;B ) : <a class="code" href="classBMEF.html" title="Estimator for Exponential family.">BMEF</a> ( B ),<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a> ( <a class="code" href="classBM.html#af00f0612fabe66241dd507188cdbf88" title="Random variable of the posterior.">rv</a>,B.<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a> ),<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a> ( <a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>._beta() ) {}
+<a name="l00266"></a><a class="code" href="classmultiBM.html#42e36804041e551d3ceea6c75abc0562">00266</a>         <span class="keywordtype">void</span> <a class="code" href="classmultiBM.html#42e36804041e551d3ceea6c75abc0562" title="Sets sufficient statistics to match that of givefrom mB0.">set_statistics</a> ( <span class="keyword">const</span> <a class="code" href="classBM.html" title="Bayesian Model of the world, i.e. all uncertainty is modeled by probabilities.">BM</a>* mB0 ) {<span class="keyword">const</span> <a class="code" href="classmultiBM.html" title="Estimator for Multinomial density.">multiBM</a>* mB=<span class="keyword">dynamic_cast&lt;</span><span class="keyword">const </span><a class="code" href="classmultiBM.html" title="Estimator for Multinomial density.">multiBM</a>*<span class="keyword">&gt;</span> ( mB0 ); <a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>=mB-&gt;<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>;}
+<a name="l00267"></a><a class="code" href="classmultiBM.html#11eeba7e97954e316e959116f90d80e2">00267</a>         <span class="keywordtype">void</span> <a class="code" href="classmultiBM.html#11eeba7e97954e316e959116f90d80e2" title="Incremental Bayes rule.">bayes</a> ( <span class="keyword">const</span> vec &amp;dt ) {
+<a name="l00268"></a>00268                 <span class="keywordflow">if</span> ( <a class="code" href="classBMEF.html#538d632e59f9afa8daa1de74da12ce71" title="forgetting factor">frg</a>&lt;1.0 ) {<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>*=<a class="code" href="classBMEF.html#538d632e59f9afa8daa1de74da12ce71" title="forgetting factor">frg</a>;<a class="code" href="classBMEF.html#308cf5d4133cd471fdf1ecd5dfa09d02" title="cached value of lognc() in the previous step (used in evaluation of ll )">last_lognc</a>=<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>();}
+<a name="l00269"></a>00269                 <a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>+=dt;
+<a name="l00270"></a>00270                 <span class="keywordflow">if</span> ( <a class="code" href="classBM.html#bf6fb59b30141074f8ee1e2f43d03129" title="If true, the filter will compute likelihood of the data record and store it in ll...">evalll</a> ) {<a class="code" href="classBM.html#5623fef6572a08c2b53b8c87b82dc979" title="Logarithm of marginalized data likelihood.">ll</a>=<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>()-<a class="code" href="classBMEF.html#308cf5d4133cd471fdf1ecd5dfa09d02" title="cached value of lognc() in the previous step (used in evaluation of ll )">last_lognc</a>;}
+<a name="l00271"></a>00271         }
+<a name="l00272"></a><a class="code" href="classmultiBM.html#13e26a61757278981fd8cac9a7ef91eb">00272</a>         <span class="keywordtype">double</span> <a class="code" href="classmultiBM.html#13e26a61757278981fd8cac9a7ef91eb">logpred</a> ( <span class="keyword">const</span> vec &amp;dt )<span class="keyword"> const </span>{
+<a name="l00273"></a>00273                 <a class="code" href="classeDirich.html" title="Dirichlet posterior density.">eDirich</a> pred ( <a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a> );
+<a name="l00274"></a>00274                 vec &amp;<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a> = pred.<a class="code" href="classeDirich.html#6409d0362143a23976b43641ff19e53a" title="access function">_beta</a>();
+<a name="l00275"></a>00275 
+<a name="l00276"></a>00276                 <span class="keywordtype">double</span> lll;
+<a name="l00277"></a>00277                 <span class="keywordflow">if</span> ( <a class="code" href="classBMEF.html#538d632e59f9afa8daa1de74da12ce71" title="forgetting factor">frg</a>&lt;1.0 )
+<a name="l00278"></a>00278                         {beta*=<a class="code" href="classBMEF.html#538d632e59f9afa8daa1de74da12ce71" title="forgetting factor">frg</a>;lll=pred.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>();}
+<a name="l00279"></a>00279                 <span class="keywordflow">else</span>
+<a name="l00280"></a>00280                         <span class="keywordflow">if</span> ( <a class="code" href="classBM.html#bf6fb59b30141074f8ee1e2f43d03129" title="If true, the filter will compute likelihood of the data record and store it in ll...">evalll</a> ) {lll=<a class="code" href="classBMEF.html#308cf5d4133cd471fdf1ecd5dfa09d02" title="cached value of lognc() in the previous step (used in evaluation of ll )">last_lognc</a>;}
+<a name="l00281"></a>00281                         <span class="keywordflow">else</span>{lll=pred.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>();}
+<a name="l00282"></a>00282 
+<a name="l00283"></a>00283                 beta+=dt;
+<a name="l00284"></a>00284                 <span class="keywordflow">return</span> pred.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>()-lll;
+<a name="l00285"></a>00285         }
+<a name="l00286"></a><a class="code" href="classmultiBM.html#3988322f8f51b153622036f461f62a67">00286</a>         <span class="keywordtype">void</span> <a class="code" href="classmultiBM.html#3988322f8f51b153622036f461f62a67" title="Flatten the posterior according to the given BMEF (of the same type!).">flatten</a> ( <span class="keyword">const</span> <a class="code" href="classBMEF.html" title="Estimator for Exponential family.">BMEF</a>* B ) {
+<a name="l00287"></a>00287                 <span class="keyword">const</span> <a class="code" href="classmultiBM.html" title="Estimator for Multinomial density.">multiBM</a>* E=<span class="keyword">dynamic_cast&lt;</span><span class="keyword">const </span><a class="code" href="classmultiBM.html" title="Estimator for Multinomial density.">multiBM</a>*<span class="keyword">&gt;</span> ( B );
+<a name="l00288"></a>00288                 <span class="comment">// sum(beta) should be equal to sum(B.beta)</span>
+<a name="l00289"></a>00289                 <span class="keyword">const</span> vec &amp;Eb=E-&gt;<a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>;<span class="comment">//const_cast&lt;multiBM*&gt; ( E )-&gt;_beta();</span>
+<a name="l00290"></a>00290                 <a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a>*= ( sum ( Eb ) /sum ( <a class="code" href="classmultiBM.html#7b606116aed7e8834a339cbb0424b1d6" title="Pointer inside est to sufficient statistics.">beta</a> ) );
+<a name="l00291"></a>00291                 <span class="keywordflow">if</span> ( <a class="code" href="classBM.html#bf6fb59b30141074f8ee1e2f43d03129" title="If true, the filter will compute likelihood of the data record and store it in ll...">evalll</a> ) {<a class="code" href="classBMEF.html#308cf5d4133cd471fdf1ecd5dfa09d02" title="cached value of lognc() in the previous step (used in evaluation of ll )">last_lognc</a>=<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>();}
+<a name="l00292"></a>00292         }
+<a name="l00293"></a><a class="code" href="classmultiBM.html#66cdfd83a70bc281840ab0646b941684">00293</a>         <span class="keyword">const</span> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>&amp; <a class="code" href="classmultiBM.html#66cdfd83a70bc281840ab0646b941684" title="Returns a reference to the epdf representing posterior density on parameters.">_epdf</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> <a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>;};
+<a name="l00294"></a><a class="code" href="classmultiBM.html#66a0fa6966e40bb6c3e7ba22d26e9d35">00294</a>         <span class="keyword">const</span> <a class="code" href="classeDirich.html" title="Dirichlet posterior density.">eDirich</a>* <a class="code" href="classmultiBM.html#66a0fa6966e40bb6c3e7ba22d26e9d35" title="Returns a pointer to the epdf representing posterior density on parameters. Use with...">_e</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> &amp;<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>;};
+<a name="l00295"></a>00295         <span class="keywordtype">void</span> set_parameters ( <span class="keyword">const</span> vec &amp;beta0 ) {
+<a name="l00296"></a>00296                 <a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>.<a class="code" href="classeDirich.html#c842acb2e1cce5cc9000769ff06c086d" title="Set internal parameters.">set_parameters</a> ( beta0 );
+<a name="l00297"></a>00297                 <a class="code" href="classBM.html#af00f0612fabe66241dd507188cdbf88" title="Random variable of the posterior.">rv</a> = <a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>.<a class="code" href="classepdf.html#ca0d32aabb4cbba347e0c37fe8607562" title="access function, possibly dangerous!">_rv</a>();
+<a name="l00298"></a>00298                 <span class="keywordflow">if</span> ( <a class="code" href="classBM.html#bf6fb59b30141074f8ee1e2f43d03129" title="If true, the filter will compute likelihood of the data record and store it in ll...">evalll</a> ) {<a class="code" href="classBMEF.html#308cf5d4133cd471fdf1ecd5dfa09d02" title="cached value of lognc() in the previous step (used in evaluation of ll )">last_lognc</a>=<a class="code" href="classmultiBM.html#eddee08a724170de63f36e40c57b27b5" title="Conjugate prior and posterior.">est</a>.<a class="code" href="classeDirich.html#7ce60be7119ffc639ede4e583c1f6e77" title="logarithm of the normalizing constant, ">lognc</a>();}
+<a name="l00299"></a>00299         }
+<a name="l00300"></a>00300 };
+<a name="l00301"></a>00301 
+<a name="l00311"></a><a class="code" href="classegamma.html">00311</a> <span class="keyword">class </span><a class="code" href="classegamma.html" title="Gamma posterior density.">egamma</a> : <span class="keyword">public</span> <a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> {
+<a name="l00312"></a>00312 <span class="keyword">protected</span>:
+<a name="l00314"></a><a class="code" href="classegamma.html#376cebd8932546c440f21b182910b01b">00314</a>         vec <a class="code" href="classegamma.html#376cebd8932546c440f21b182910b01b" title="Vector .">alpha</a>;
+<a name="l00316"></a><a class="code" href="classegamma.html#cfc5f136467488a421ab22f886323790">00316</a>         vec <a class="code" href="classegamma.html#cfc5f136467488a421ab22f886323790" title="Vector .">beta</a>;
+<a name="l00317"></a>00317 <span class="keyword">public</span> :
+<a name="l00319"></a><a class="code" href="classegamma.html#4b1d34f3b244ea51a58ec10c468788c1">00319</a>         <a class="code" href="classegamma.html#4b1d34f3b244ea51a58ec10c468788c1" title="Default constructor.">egamma</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a> ) :<a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> ( rv ) {};
+<a name="l00321"></a><a class="code" href="classegamma.html#8e348b89be82b70471fe8c5630f61339">00321</a>         <span class="keywordtype">void</span> <a class="code" href="classegamma.html#8e348b89be82b70471fe8c5630f61339" title="Sets parameters.">set_parameters</a> ( <span class="keyword">const</span> vec &amp;a, <span class="keyword">const</span> vec &amp;b ) {<a class="code" href="classegamma.html#376cebd8932546c440f21b182910b01b" title="Vector .">alpha</a>=a,<a class="code" href="classegamma.html#cfc5f136467488a421ab22f886323790" title="Vector .">beta</a>=b;};
+<a name="l00322"></a>00322         vec <a class="code" href="classegamma.html#8e10c0021b5dfdd9cb62c6959b5ef425" title="Returns a sample,  from density .">sample</a>() <span class="keyword">const</span>;
+<a name="l00324"></a>00324 <span class="comment">//      mat sample ( int N ) const;</span>
+<a name="l00325"></a>00325         <span class="keywordtype">double</span> <a class="code" href="classegamma.html#74a49a4c696f44e54bb6b0515e155a9b" title="TODO: is it used anywhere?">evallog</a> ( <span class="keyword">const</span> vec &amp;val ) <span class="keyword">const</span>;
+<a name="l00326"></a>00326         <span class="keywordtype">double</span> <a class="code" href="classegamma.html#d6dbbdb72360f9e54d64501f80318bb6" title="logarithm of the normalizing constant, ">lognc</a> () <span class="keyword">const</span>;
+<a name="l00328"></a><a class="code" href="classegamma.html#44445c56e60b91b377f207f8d5089790">00328</a>         <span class="keywordtype">void</span> <a class="code" href="classegamma.html#44445c56e60b91b377f207f8d5089790" title="Returns poiter to alpha and beta. Potentially dengerous: use with care!">_param</a> ( vec* &amp;a, vec* &amp;b ) {a=&amp;<a class="code" href="classegamma.html#376cebd8932546c440f21b182910b01b" title="Vector .">alpha</a>;b=&amp;<a class="code" href="classegamma.html#cfc5f136467488a421ab22f886323790" title="Vector .">beta</a>;};
+<a name="l00329"></a><a class="code" href="classegamma.html#6ab5ba56f7cdb2e5921c3e77524fa50a">00329</a>         vec <a class="code" href="classegamma.html#6ab5ba56f7cdb2e5921c3e77524fa50a" title="return expected value">mean</a>()<span class="keyword"> const </span>{vec pom ( <a class="code" href="classegamma.html#376cebd8932546c440f21b182910b01b" title="Vector .">alpha</a> ); pom/=<a class="code" href="classegamma.html#cfc5f136467488a421ab22f886323790" title="Vector .">beta</a>; <span class="keywordflow">return</span> pom;}
+<a name="l00330"></a>00330 };
+<a name="l00331"></a>00331 <span class="comment">/*</span>
+<a name="l00333"></a>00333 <span class="comment">class emix : public epdf {</span>
+<a name="l00334"></a>00334 <span class="comment">protected:</span>
+<a name="l00335"></a>00335 <span class="comment">        int n;</span>
+<a name="l00336"></a>00336 <span class="comment">        vec &amp;w;</span>
+<a name="l00337"></a>00337 <span class="comment">        Array&lt;epdf*&gt; Coms;</span>
+<a name="l00338"></a>00338 <span class="comment">public:</span>
+<a name="l00340"></a>00340 <span class="comment">        emix ( const RV &amp;rv, vec &amp;w0): epdf(rv), n(w0.length()), w(w0), Coms(n) {};</span>
+<a name="l00341"></a>00341 <span class="comment">        void set_parameters( int &amp;i, double wi, epdf* ep){w(i)=wi;Coms(i)=ep;}</span>
+<a name="l00342"></a>00342 <span class="comment">        vec mean(){vec pom; for(int i=0;i&lt;n;i++){pom+=Coms(i)-&gt;mean()*w(i);} return pom;};</span>
+<a name="l00343"></a>00343 <span class="comment">        vec sample() {it_error ( "Not implemented" );return 0;}</span>
+<a name="l00344"></a>00344 <span class="comment">};</span>
+<a name="l00345"></a>00345 <span class="comment">*/</span>
+<a name="l00346"></a>00346 
+<a name="l00348"></a>00348 
+<a name="l00349"></a><a class="code" href="classeuni.html">00349</a> <span class="keyword">class </span><a class="code" href="classeuni.html" title="Uniform distributed density on a rectangular support.">euni</a>: <span class="keyword">public</span> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a> {
+<a name="l00350"></a>00350 <span class="keyword">protected</span>:
+<a name="l00352"></a><a class="code" href="classeuni.html#ef42cd8d7645422048d46c46ec5cdac1">00352</a>         vec <a class="code" href="classeuni.html#ef42cd8d7645422048d46c46ec5cdac1" title="lower bound on support">low</a>;
+<a name="l00354"></a><a class="code" href="classeuni.html#71b6d6b41aeb61a7f76f682b72119231">00354</a>         vec <a class="code" href="classeuni.html#71b6d6b41aeb61a7f76f682b72119231" title="upper bound on support">high</a>;
+<a name="l00356"></a><a class="code" href="classeuni.html#52a6ff4a54010f88a6a19fca605c64a4">00356</a>         vec <a class="code" href="classeuni.html#52a6ff4a54010f88a6a19fca605c64a4" title="internal">distance</a>;
+<a name="l00358"></a><a class="code" href="classeuni.html#63105490e946e43372d6187ad1bafdda">00358</a>         <span class="keywordtype">double</span> <a class="code" href="classeuni.html#63105490e946e43372d6187ad1bafdda" title="normalizing coefficients">nk</a>;
+<a name="l00360"></a><a class="code" href="classeuni.html#f445a0ce24f39d14c1a4eed53fc8e2c3">00360</a>         <span class="keywordtype">double</span> <a class="code" href="classeuni.html#f445a0ce24f39d14c1a4eed53fc8e2c3" title="cache of log( nk )">lnk</a>;
+<a name="l00361"></a>00361 <span class="keyword">public</span>:
+<a name="l00363"></a><a class="code" href="classeuni.html#2537a6c239cff52e3ba814851a1116cd">00363</a>         <a class="code" href="classeuni.html#2537a6c239cff52e3ba814851a1116cd" title="Defualt constructor.">euni</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a> ) :<a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a> ( rv ) {}
+<a name="l00364"></a>00364         <span class="keywordtype">double</span> eval ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const  </span>{<span class="keywordflow">return</span> <a class="code" href="classeuni.html#63105490e946e43372d6187ad1bafdda" title="normalizing coefficients">nk</a>;}
+<a name="l00365"></a><a class="code" href="classeuni.html#357b36417ef4c9211d12e7a4a602fd6a">00365</a>         <span class="keywordtype">double</span> <a class="code" href="classeuni.html#357b36417ef4c9211d12e7a4a602fd6a" title="Compute log-probability of argument val.">evallog</a> ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const  </span>{<span class="keywordflow">return</span> <a class="code" href="classeuni.html#f445a0ce24f39d14c1a4eed53fc8e2c3" title="cache of log( nk )">lnk</a>;}
+<a name="l00366"></a><a class="code" href="classeuni.html#4a0e09392be17beaee120ba98fc038cd">00366</a>         vec <a class="code" href="classeuni.html#4a0e09392be17beaee120ba98fc038cd" title="Returns a sample,  from density .">sample</a>()<span class="keyword"> const </span>{
+<a name="l00367"></a>00367                 vec smp ( <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() );
+<a name="l00368"></a>00368 <span class="preprocessor">#pragma omp critical</span>
+<a name="l00369"></a>00369 <span class="preprocessor"></span>                UniRNG.sample_vector ( <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>(),smp );
+<a name="l00370"></a>00370                 <span class="keywordflow">return</span> <a class="code" href="classeuni.html#ef42cd8d7645422048d46c46ec5cdac1" title="lower bound on support">low</a>+elem_mult ( <a class="code" href="classeuni.html#52a6ff4a54010f88a6a19fca605c64a4" title="internal">distance</a>,smp );
+<a name="l00371"></a>00371         }
+<a name="l00373"></a><a class="code" href="classeuni.html#4fd7c6a05100616ad16ece405cad7bf2">00373</a>         <span class="keywordtype">void</span> <a class="code" href="classeuni.html#4fd7c6a05100616ad16ece405cad7bf2" title="set values of low and high ">set_parameters</a> ( <span class="keyword">const</span> vec &amp;low0, <span class="keyword">const</span> vec &amp;high0 ) {
+<a name="l00374"></a>00374                 <a class="code" href="classeuni.html#52a6ff4a54010f88a6a19fca605c64a4" title="internal">distance</a> = high0-low0;
+<a name="l00375"></a>00375                 it_assert_debug ( min ( <a class="code" href="classeuni.html#52a6ff4a54010f88a6a19fca605c64a4" title="internal">distance</a> ) &gt;0.0,<span class="stringliteral">"bad support"</span> );
+<a name="l00376"></a>00376                 <a class="code" href="classeuni.html#ef42cd8d7645422048d46c46ec5cdac1" title="lower bound on support">low</a> = low0;
+<a name="l00377"></a>00377                 <a class="code" href="classeuni.html#71b6d6b41aeb61a7f76f682b72119231" title="upper bound on support">high</a> = high0;
+<a name="l00378"></a>00378                 <a class="code" href="classeuni.html#63105490e946e43372d6187ad1bafdda" title="normalizing coefficients">nk</a> = prod ( 1.0/<a class="code" href="classeuni.html#52a6ff4a54010f88a6a19fca605c64a4" title="internal">distance</a> );
+<a name="l00379"></a>00379                 <a class="code" href="classeuni.html#f445a0ce24f39d14c1a4eed53fc8e2c3" title="cache of log( nk )">lnk</a> = log ( <a class="code" href="classeuni.html#63105490e946e43372d6187ad1bafdda" title="normalizing coefficients">nk</a> );
+<a name="l00380"></a>00380         }
+<a name="l00381"></a><a class="code" href="classeuni.html#8050087e421a9cfd1b4b1f8bd33b1cc1">00381</a>         vec <a class="code" href="classeuni.html#8050087e421a9cfd1b4b1f8bd33b1cc1" title="return expected value">mean</a>()<span class="keyword"> const </span>{vec pom=<a class="code" href="classeuni.html#71b6d6b41aeb61a7f76f682b72119231" title="upper bound on support">high</a>; pom-=<a class="code" href="classeuni.html#ef42cd8d7645422048d46c46ec5cdac1" title="lower bound on support">low</a>; pom/=2.0; <span class="keywordflow">return</span> pom;}
+<a name="l00382"></a>00382 };
+<a name="l00383"></a>00383 
+<a name="l00384"></a>00384 
+<a name="l00390"></a>00390 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
+<a name="l00391"></a><a class="code" href="classmlnorm.html">00391</a> <span class="keyword">class </span><a class="code" href="classmlnorm.html" title="Normal distributed linear function with linear function of mean value;.">mlnorm</a> : <span class="keyword">public</span> <a class="code" href="classmEF.html" title="Exponential family model.">mEF</a> {
+<a name="l00392"></a>00392 <span class="keyword">protected</span>:
+<a name="l00394"></a><a class="code" href="classmlnorm.html#b76ee2171ace4fb3ff95a131ae8fc421">00394</a>         <a class="code" href="classenorm.html" title="Gaussian density with positive definite (decomposed) covariance matrix.">enorm&lt;sq_T&gt;</a> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>;
+<a name="l00395"></a>00395         mat A;
+<a name="l00396"></a>00396         vec mu_const;
+<a name="l00397"></a>00397         vec&amp; _mu; <span class="comment">//cached epdf.mu;</span>
+<a name="l00398"></a>00398 <span class="keyword">public</span>:
+<a name="l00400"></a>00400         <a class="code" href="classmlnorm.html#3a5ad4798d8a3878c5e93b8e796c8837" title="Constructor.">mlnorm</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classmpdf.html#f6687c07ff07d47812dd565368ca59eb" title="modeled random variable">rv</a>, <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classmpdf.html#acb7dda792b3cd5576f39fa3129abbab" title="random variable in condition">rvc</a> );
+<a name="l00402"></a>00402         <span class="keywordtype">void</span> <a class="code" href="classmlnorm.html#f95dfce0b500636a44ecd7e5210de999" title="Set A and R.">set_parameters</a> ( <span class="keyword">const</span>  mat &amp;A, <span class="keyword">const</span> vec &amp;mu0, <span class="keyword">const</span> sq_T &amp;R );
+<a name="l00403"></a>00403 <span class="comment">//      //!Generate one sample of the posterior</span>
+<a name="l00404"></a>00404 <span class="comment">//      vec samplecond (const vec &amp;cond, double &amp;lik );</span>
+<a name="l00405"></a>00405 <span class="comment">//      //!Generate matrix of samples of the posterior</span>
+<a name="l00406"></a>00406 <span class="comment">//      mat samplecond (const vec &amp;cond, vec &amp;lik, int n );</span>
+<a name="l00408"></a>00408 <span class="comment"></span>        <span class="keywordtype">void</span> <a class="code" href="classmlnorm.html#d41126455ac64b888a38f677886e1b40" title="Set value of rvc . Result of this operation is stored in epdf use function _ep to...">condition</a> ( <span class="keyword">const</span> vec &amp;cond );
+<a name="l00409"></a>00409 
+<a name="l00411"></a><a class="code" href="classmlnorm.html#2732ae47835dd25d5784bf08fde0a546">00411</a>         vec&amp; <a class="code" href="classmlnorm.html#2732ae47835dd25d5784bf08fde0a546" title="access function">_mu_const</a>() {<span class="keywordflow">return</span> mu_const;}
+<a name="l00413"></a><a class="code" href="classmlnorm.html#65ec3840c21b21102896bfd2282b47b3">00413</a>         mat&amp; <a class="code" href="classmlnorm.html#65ec3840c21b21102896bfd2282b47b3" title="access function">_A</a>() {<span class="keywordflow">return</span> A;}
+<a name="l00415"></a><a class="code" href="classmlnorm.html#3ff2b03fbb5e1133a5fe1bf831939f63">00415</a>         mat <a class="code" href="classmlnorm.html#3ff2b03fbb5e1133a5fe1bf831939f63" title="access function">_R</a>() {<span class="keywordflow">return</span> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>._R().to_mat();}
+<a name="l00416"></a>00416 
+<a name="l00417"></a>00417         <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_M&gt;
+<a name="l00418"></a>00418         <span class="keyword">friend</span> std::ostream &amp;operator&lt;&lt; ( std::ostream &amp;os,  mlnorm&lt;sq_M&gt; &amp;ml );
+<a name="l00419"></a>00419 };
+<a name="l00420"></a>00420 
+<a name="l00423"></a><a class="code" href="classmlstudent.html">00423</a> <span class="keyword">class </span><a class="code" href="classmlstudent.html">mlstudent</a> : <span class="keyword">public</span> <a class="code" href="classmlnorm.html" title="Normal distributed linear function with linear function of mean value;.">mlnorm</a>&lt;ldmat&gt; {
+<a name="l00424"></a>00424 <span class="keyword">protected</span>:
+<a name="l00425"></a>00425         <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a> Lambda;
+<a name="l00426"></a>00426         <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a> &amp;<a class="code" href="classmlnorm.html#3ff2b03fbb5e1133a5fe1bf831939f63" title="access function">_R</a>;
+<a name="l00427"></a>00427         <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a> Re;
+<a name="l00428"></a>00428 <span class="keyword">public</span>:
+<a name="l00429"></a>00429         <a class="code" href="classmlstudent.html">mlstudent</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rv0, <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rvc0 ) :<a class="code" href="classmlnorm.html" title="Normal distributed linear function with linear function of mean value;.">mlnorm&lt;ldmat&gt;</a> ( rv0,rvc0 ),
+<a name="l00430"></a>00430                         Lambda ( rv0.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ),
+<a name="l00431"></a>00431                         <a class="code" href="classmlnorm.html#3ff2b03fbb5e1133a5fe1bf831939f63" title="access function">_R</a> ( <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>._R() ) {}
+<a name="l00432"></a>00432         <span class="keywordtype">void</span> set_parameters ( <span class="keyword">const</span> mat &amp;A0, <span class="keyword">const</span> vec &amp;mu0, <span class="keyword">const</span> <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a> &amp;R0, <span class="keyword">const</span> <a class="code" href="classldmat.html" title="Matrix stored in LD form, (commonly known as UD).">ldmat</a>&amp; Lambda0) {
+<a name="l00433"></a>00433                 <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>.set_parameters ( zeros ( <a class="code" href="classmpdf.html#f6687c07ff07d47812dd565368ca59eb" title="modeled random variable">rv</a>.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ),Lambda );
+<a name="l00434"></a>00434                 A = A0;
+<a name="l00435"></a>00435                 mu_const = mu0;
+<a name="l00436"></a>00436                 Re=R0;
+<a name="l00437"></a>00437                 Lambda = Lambda0;
+<a name="l00438"></a>00438         }
+<a name="l00439"></a><a class="code" href="classmlstudent.html#d153460ae0180f4bc7f28301f5cde876">00439</a>         <span class="keywordtype">void</span> <a class="code" href="classmlstudent.html#d153460ae0180f4bc7f28301f5cde876" title="Set value of rvc . Result of this operation is stored in epdf use function _ep to...">condition</a> ( <span class="keyword">const</span> vec &amp;cond ) {
+<a name="l00440"></a>00440                 _mu = A*cond + mu_const;
+<a name="l00441"></a>00441                 <span class="keywordtype">double</span> zeta;
+<a name="l00442"></a>00442                 <span class="comment">//ugly hack!</span>
+<a name="l00443"></a>00443                 <span class="keywordflow">if</span> ((cond.length()+1)==Lambda.<a class="code" href="group__math.html#g96dfb21865db4f5bd36fa70f9b0b1163" title="access function">rows</a>()){
+<a name="l00444"></a>00444                         zeta = Lambda.<a class="code" href="classldmat.html#d876c5f83e02b3e809b35c9de5068f14" title="Evaluates quadratic form ;.">invqform</a> ( <a class="code" href="group__core.html#g33c114e83980d883c5b211c47d5322a4" title="Concat two random variables.">concat</a>(cond, vec_1(1.0)) );
+<a name="l00445"></a>00445                 } <span class="keywordflow">else</span> {
+<a name="l00446"></a>00446                         zeta = Lambda.<a class="code" href="classldmat.html#d876c5f83e02b3e809b35c9de5068f14" title="Evaluates quadratic form ;.">invqform</a> ( cond );
+<a name="l00447"></a>00447                 }
+<a name="l00448"></a>00448                 <a class="code" href="classmlnorm.html#3ff2b03fbb5e1133a5fe1bf831939f63" title="access function">_R</a> = Re;
+<a name="l00449"></a>00449                 <a class="code" href="classmlnorm.html#3ff2b03fbb5e1133a5fe1bf831939f63" title="access function">_R</a>*=( 1+zeta );<span class="comment">// / ( nu ); &lt;&lt; nu is in Re!!!!!!</span>
+<a name="l00450"></a>00450         };
+<a name="l00451"></a>00451 
+<a name="l00452"></a>00452 };
+<a name="l00462"></a><a class="code" href="classmgamma.html">00462</a> <span class="keyword">class </span><a class="code" href="classmgamma.html" title="Gamma random walk.">mgamma</a> : <span class="keyword">public</span> <a class="code" href="classmEF.html" title="Exponential family model.">mEF</a> {
+<a name="l00463"></a>00463 <span class="keyword">protected</span>:
+<a name="l00465"></a><a class="code" href="classmgamma.html#612dbf35c770a780027619aaac2c443e">00465</a>         <a class="code" href="classegamma.html" title="Gamma posterior density.">egamma</a> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>;
+<a name="l00467"></a><a class="code" href="classmgamma.html#43f733cce0245a52363d566099add687">00467</a>         <span class="keywordtype">double</span> <a class="code" href="classmgamma.html#43f733cce0245a52363d566099add687" title="Constant .">k</a>;
+<a name="l00469"></a><a class="code" href="classmgamma.html#5e90652837448bcc29707e7412f99691">00469</a>         vec* <a class="code" href="classmgamma.html#5e90652837448bcc29707e7412f99691" title="cache of epdf.beta">_beta</a>;
+<a name="l00470"></a>00470 
+<a name="l00471"></a>00471 <span class="keyword">public</span>:
+<a name="l00473"></a>00473         <a class="code" href="classmgamma.html#af43e61b86900c0398d5c0ffc83b94e6" title="Constructor.">mgamma</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classmpdf.html#f6687c07ff07d47812dd565368ca59eb" title="modeled random variable">rv</a>,<span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classmpdf.html#acb7dda792b3cd5576f39fa3129abbab" title="random variable in condition">rvc</a> );
+<a name="l00475"></a>00475         <span class="keywordtype">void</span> <a class="code" href="classmgamma.html#a9d646cf758a70126dde7c48790b6e94" title="Set value of k.">set_parameters</a> ( <span class="keywordtype">double</span> <a class="code" href="classmgamma.html#43f733cce0245a52363d566099add687" title="Constant .">k</a> );
+<a name="l00476"></a><a class="code" href="classmgamma.html#a61094c9f7a2d64ea77b130cbc031f97">00476</a>         <span class="keywordtype">void</span> <a class="code" href="classmgamma.html#a61094c9f7a2d64ea77b130cbc031f97" title="Update ep so that it represents this mpdf conditioned on rvc = cond.">condition</a> ( <span class="keyword">const</span> vec &amp;val ) {*<a class="code" href="classmgamma.html#5e90652837448bcc29707e7412f99691" title="cache of epdf.beta">_beta</a>=<a class="code" href="classmgamma.html#43f733cce0245a52363d566099add687" title="Constant .">k</a>/val;};
+<a name="l00477"></a>00477 };
+<a name="l00478"></a>00478 
+<a name="l00490"></a><a class="code" href="classmgamma__fix.html">00490</a> <span class="keyword">class </span><a class="code" href="classmgamma__fix.html" title="Gamma random walk around a fixed point.">mgamma_fix</a> : <span class="keyword">public</span> <a class="code" href="classmgamma.html" title="Gamma random walk.">mgamma</a> {
+<a name="l00491"></a>00491 <span class="keyword">protected</span>:
+<a name="l00493"></a><a class="code" href="classmgamma__fix.html#3f48c09caddc298901ad75fe7c0529f6">00493</a>         <span class="keywordtype">double</span> <a class="code" href="classmgamma__fix.html#3f48c09caddc298901ad75fe7c0529f6" title="parameter l">l</a>;
+<a name="l00495"></a><a class="code" href="classmgamma__fix.html#81ce49029ecc385418619b200dcafeb0">00495</a>         vec <a class="code" href="classmgamma__fix.html#81ce49029ecc385418619b200dcafeb0" title="reference vector">refl</a>;
+<a name="l00496"></a>00496 <span class="keyword">public</span>:
+<a name="l00498"></a><a class="code" href="classmgamma__fix.html#b92c3d2e5fd0381033a072e5ef3bcf80">00498</a>         <a class="code" href="classmgamma__fix.html#b92c3d2e5fd0381033a072e5ef3bcf80" title="Constructor.">mgamma_fix</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classmpdf.html#f6687c07ff07d47812dd565368ca59eb" title="modeled random variable">rv</a>,<span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;<a class="code" href="classmpdf.html#acb7dda792b3cd5576f39fa3129abbab" title="random variable in condition">rvc</a> ) : <a class="code" href="classmgamma.html" title="Gamma random walk.">mgamma</a> ( rv,rvc ),<a class="code" href="classmgamma__fix.html#81ce49029ecc385418619b200dcafeb0" title="reference vector">refl</a> ( rv.count() ) {};
+<a name="l00500"></a><a class="code" href="classmgamma__fix.html#ec6f846896749e27cb7be9fa48dd1cb1">00500</a>         <span class="keywordtype">void</span> <a class="code" href="classmgamma__fix.html#ec6f846896749e27cb7be9fa48dd1cb1" title="Set value of k.">set_parameters</a> ( <span class="keywordtype">double</span> k0 , vec ref0, <span class="keywordtype">double</span> l0 ) {
+<a name="l00501"></a>00501                 <a class="code" href="classmgamma.html#a9d646cf758a70126dde7c48790b6e94" title="Set value of k.">mgamma::set_parameters</a> ( k0 );
+<a name="l00502"></a>00502                 <a class="code" href="classmgamma__fix.html#81ce49029ecc385418619b200dcafeb0" title="reference vector">refl</a>=pow ( ref0,1.0-l0 );<a class="code" href="classmgamma__fix.html#3f48c09caddc298901ad75fe7c0529f6" title="parameter l">l</a>=l0;
+<a name="l00503"></a>00503         };
+<a name="l00504"></a>00504 
+<a name="l00505"></a><a class="code" href="classmgamma__fix.html#6ea3931eec7b7da7b693e45981052460">00505</a>         <span class="keywordtype">void</span> <a class="code" href="classmgamma__fix.html#6ea3931eec7b7da7b693e45981052460" title="Update ep so that it represents this mpdf conditioned on rvc = cond.">condition</a> ( <span class="keyword">const</span> vec &amp;val ) {vec mean=elem_mult ( <a class="code" href="classmgamma__fix.html#81ce49029ecc385418619b200dcafeb0" title="reference vector">refl</a>,pow ( val,<a class="code" href="classmgamma__fix.html#3f48c09caddc298901ad75fe7c0529f6" title="parameter l">l</a> ) ); *<a class="code" href="classmgamma.html#5e90652837448bcc29707e7412f99691" title="cache of epdf.beta">_beta</a>=<a class="code" href="classmgamma.html#43f733cce0245a52363d566099add687" title="Constant .">k</a>/mean;};
+<a name="l00506"></a>00506 };
+<a name="l00507"></a>00507 
+<a name="l00509"></a><a class="code" href="libEF_8h.html#99497a3ff630f761cf6bff7babd23212">00509</a> <span class="keyword">enum</span> <a class="code" href="libEF_8h.html#99497a3ff630f761cf6bff7babd23212" title="Switch between various resampling methods.">RESAMPLING_METHOD</a> { MULTINOMIAL = 0, STRATIFIED = 1, SYSTEMATIC = 3 };
+<a name="l00515"></a><a class="code" href="classeEmp.html">00515</a> <span class="keyword">class </span><a class="code" href="classeEmp.html" title="Weighted empirical density.">eEmp</a>: <span class="keyword">public</span> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a> {
+<a name="l00516"></a>00516 <span class="keyword">protected</span> :
+<a name="l00518"></a><a class="code" href="classeEmp.html#8c33034de0e35f03f8bb85d3d67438fd">00518</a>         <span class="keywordtype">int</span> <a class="code" href="classeEmp.html#8c33034de0e35f03f8bb85d3d67438fd" title="Number of particles.">n</a>;
+<a name="l00520"></a><a class="code" href="classeEmp.html#ae78d144404ddba843c93b171b215de8">00520</a>         vec <a class="code" href="classeEmp.html#ae78d144404ddba843c93b171b215de8" title="Sample weights .">w</a>;
+<a name="l00522"></a><a class="code" href="classeEmp.html#a4d6f4bbd6a6824fc39f14676701279a">00522</a>         Array&lt;vec&gt; <a class="code" href="classeEmp.html#a4d6f4bbd6a6824fc39f14676701279a" title="Samples .">samples</a>;
+<a name="l00523"></a>00523 <span class="keyword">public</span>:
+<a name="l00525"></a><a class="code" href="classeEmp.html#0c04b073ecd0dae3d498e680ae27e9e4">00525</a>         <a class="code" href="classeEmp.html#0c04b073ecd0dae3d498e680ae27e9e4" title="Default constructor.">eEmp</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rv0 ,<span class="keywordtype">int</span> n0 ) :<a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a> ( rv0 ),<a class="code" href="classeEmp.html#8c33034de0e35f03f8bb85d3d67438fd" title="Number of particles.">n</a> ( n0 ),<a class="code" href="classeEmp.html#ae78d144404ddba843c93b171b215de8" title="Sample weights .">w</a> ( <a class="code" href="classeEmp.html#8c33034de0e35f03f8bb85d3d67438fd" title="Number of particles.">n</a> ),<a class="code" href="classeEmp.html#a4d6f4bbd6a6824fc39f14676701279a" title="Samples .">samples</a> ( <a class="code" href="classeEmp.html#8c33034de0e35f03f8bb85d3d67438fd" title="Number of particles.">n</a> ) {};
+<a name="l00527"></a>00527         <span class="keywordtype">void</span> <a class="code" href="classeEmp.html#eab03bd3381aaea11ce34d5a26556353" title="Set samples and weights.">set_parameters</a> ( <span class="keyword">const</span> vec &amp;w0, <span class="keyword">const</span> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>* pdf0 );
+<a name="l00529"></a>00529         <span class="keywordtype">void</span> <a class="code" href="classeEmp.html#e31bc9e6196173c3480b06a761a3e716" title="Set sample.">set_samples</a> ( <span class="keyword">const</span> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>* pdf0 );
+<a name="l00531"></a><a class="code" href="classeEmp.html#a4215f6a5a04d07b43f7ebaa942e15f1">00531</a>         <span class="keywordtype">void</span> <a class="code" href="classeEmp.html#a4215f6a5a04d07b43f7ebaa942e15f1" title="Set sample.">set_n</a> ( <span class="keywordtype">int</span> n0, <span class="keywordtype">bool</span> copy=<span class="keyword">true</span> ){<a class="code" href="classeEmp.html#ae78d144404ddba843c93b171b215de8" title="Sample weights .">w</a>.set_size(n0,copy);<a class="code" href="classeEmp.html#a4d6f4bbd6a6824fc39f14676701279a" title="Samples .">samples</a>.set_size(n0,copy);};
+<a name="l00533"></a><a class="code" href="classeEmp.html#31b2bfb73b72486a5c89f2ab850c7a9b">00533</a>         vec&amp; <a class="code" href="classeEmp.html#31b2bfb73b72486a5c89f2ab850c7a9b" title="Potentially dangerous, use with care.">_w</a>()  {<span class="keywordflow">return</span> <a class="code" href="classeEmp.html#ae78d144404ddba843c93b171b215de8" title="Sample weights .">w</a>;};
+<a name="l00535"></a><a class="code" href="classeEmp.html#d6f4ae1a67ecd2bff8b9f176ee261afc">00535</a>         <span class="keyword">const</span> vec&amp; <a class="code" href="classeEmp.html#31b2bfb73b72486a5c89f2ab850c7a9b" title="Potentially dangerous, use with care.">_w</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> <a class="code" href="classeEmp.html#ae78d144404ddba843c93b171b215de8" title="Sample weights .">w</a>;};
+<a name="l00537"></a><a class="code" href="classeEmp.html#31b747eca73b16f30370827ba4cc3575">00537</a>         Array&lt;vec&gt;&amp; <a class="code" href="classeEmp.html#31b747eca73b16f30370827ba4cc3575" title="access function">_samples</a>() {<span class="keywordflow">return</span> <a class="code" href="classeEmp.html#a4d6f4bbd6a6824fc39f14676701279a" title="Samples .">samples</a>;};
+<a name="l00539"></a><a class="code" href="classeEmp.html#bd48c1c36e2e9e78dbcea7df66dcbf25">00539</a>         <span class="keyword">const</span> Array&lt;vec&gt;&amp; <a class="code" href="classeEmp.html#31b747eca73b16f30370827ba4cc3575" title="access function">_samples</a>()<span class="keyword"> const </span>{<span class="keywordflow">return</span> <a class="code" href="classeEmp.html#a4d6f4bbd6a6824fc39f14676701279a" title="Samples .">samples</a>;};
+<a name="l00541"></a>00541         ivec <a class="code" href="classeEmp.html#77268292fc4465cb73ddbfb1f2932a59" title="Function performs resampling, i.e. removal of low-weight samples and duplication...">resample</a> ( <a class="code" href="libEF_8h.html#99497a3ff630f761cf6bff7babd23212" title="Switch between various resampling methods.">RESAMPLING_METHOD</a> method = SYSTEMATIC );
+<a name="l00543"></a><a class="code" href="classeEmp.html#83f9283f92b805508d896479dc1ccf12">00543</a>         vec <a class="code" href="classeEmp.html#83f9283f92b805508d896479dc1ccf12" title="inherited operation : NOT implemneted">sample</a>()<span class="keyword"> const </span>{it_error ( <span class="stringliteral">"Not implemented"</span> );<span class="keywordflow">return</span> 0;}
+<a name="l00545"></a><a class="code" href="classeEmp.html#884f16c9fc1f888408686a660a95dacd">00545</a>         <span class="keywordtype">double</span> <a class="code" href="classeEmp.html#884f16c9fc1f888408686a660a95dacd" title="inherited operation : NOT implemneted">evallog</a> ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const </span>{it_error ( <span class="stringliteral">"Not implemented"</span> );<span class="keywordflow">return</span> 0.0;}
+<a name="l00546"></a><a class="code" href="classeEmp.html#ba055c19038cc72628d98e25197e982d">00546</a>         vec <a class="code" href="classeEmp.html#ba055c19038cc72628d98e25197e982d" title="return expected value">mean</a>()<span class="keyword"> const </span>{
+<a name="l00547"></a>00547                 vec pom=zeros ( <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() );
+<a name="l00548"></a>00548                 <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i=0;i&lt;<a class="code" href="classeEmp.html#8c33034de0e35f03f8bb85d3d67438fd" title="Number of particles.">n</a>;i++ ) {pom+=<a class="code" href="classeEmp.html#a4d6f4bbd6a6824fc39f14676701279a" title="Samples .">samples</a> ( i ) *<a class="code" href="classeEmp.html#ae78d144404ddba843c93b171b215de8" title="Sample weights .">w</a> ( i );}
+<a name="l00549"></a>00549                 <span class="keywordflow">return</span> pom;
+<a name="l00550"></a>00550         }
+<a name="l00551"></a>00551 };
+<a name="l00552"></a>00552 
 <a name="l00553"></a>00553 
-<a name="l00554"></a>00554 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00555"></a><a class="code" href="classenorm.html#1394a65caa6e00d42e00cc99b12227af">00555</a> <span class="keywordtype">void</span> <a class="code" href="classenorm.html#1394a65caa6e00d42e00cc99b12227af" title="Set mean value mu and covariance R.">enorm&lt;sq_T&gt;::set_parameters</a> ( <span class="keyword">const</span> vec &amp;mu0, <span class="keyword">const</span> sq_T &amp;R0 ) {
-<a name="l00556"></a>00556 <span class="comment">//Fixme test dimensions of mu0 and R0;</span>
-<a name="l00557"></a>00557         <a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a> = mu0;
-<a name="l00558"></a>00558         <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a> = R0;
-<a name="l00559"></a>00559 };
-<a name="l00560"></a>00560 
-<a name="l00561"></a>00561 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00562"></a><a class="code" href="classenorm.html#5bf185e31e5954fceb90ada3debd2ff2">00562</a> <span class="keywordtype">void</span> <a class="code" href="classenorm.html#5bf185e31e5954fceb90ada3debd2ff2" title="dupdate in exponential form (not really handy)">enorm&lt;sq_T&gt;::dupdate</a> ( mat &amp;v, <span class="keywordtype">double</span> nu ) {
-<a name="l00563"></a>00563         <span class="comment">//</span>
+<a name="l00555"></a>00555 
+<a name="l00556"></a>00556 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
+<a name="l00557"></a><a class="code" href="classenorm.html#0caf54fed9e48f9fe28b534b2027df2f">00557</a> <a class="code" href="classenorm.html#0caf54fed9e48f9fe28b534b2027df2f" title="Default constructor.">enorm&lt;sq_T&gt;::enorm</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rv ) :<a class="code" href="classeEF.html" title="General conjugate exponential family posterior density.">eEF</a> ( rv ), mu ( rv.count() ),R ( rv.count() ),dim ( rv.count() ) {};
+<a name="l00558"></a>00558 
+<a name="l00559"></a>00559 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
+<a name="l00560"></a><a class="code" href="classenorm.html#1394a65caa6e00d42e00cc99b12227af">00560</a> <span class="keywordtype">void</span> <a class="code" href="classenorm.html#1394a65caa6e00d42e00cc99b12227af" title="Set mean value mu and covariance R.">enorm&lt;sq_T&gt;::set_parameters</a> ( <span class="keyword">const</span> vec &amp;mu0, <span class="keyword">const</span> sq_T &amp;R0 ) {
+<a name="l00561"></a>00561 <span class="comment">//Fixme test dimensions of mu0 and R0;</span>
+<a name="l00562"></a>00562         <a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a> = mu0;
+<a name="l00563"></a>00563         <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a> = R0;
 <a name="l00564"></a>00564 };
 <a name="l00565"></a>00565 
-<a name="l00566"></a>00566 <span class="comment">// template&lt;class sq_T&gt;</span>
-<a name="l00567"></a>00567 <span class="comment">// void enorm&lt;sq_T&gt;::tupdate ( double phi, mat &amp;vbar, double nubar ) {</span>
-<a name="l00568"></a>00568 <span class="comment">//      //</span>
-<a name="l00569"></a>00569 <span class="comment">// };</span>
+<a name="l00566"></a>00566 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
+<a name="l00567"></a><a class="code" href="classenorm.html#5bf185e31e5954fceb90ada3debd2ff2">00567</a> <span class="keywordtype">void</span> <a class="code" href="classenorm.html#5bf185e31e5954fceb90ada3debd2ff2" title="dupdate in exponential form (not really handy)">enorm&lt;sq_T&gt;::dupdate</a> ( mat &amp;v, <span class="keywordtype">double</span> nu ) {
+<a name="l00568"></a>00568         <span class="comment">//</span>
+<a name="l00569"></a>00569 };
 <a name="l00570"></a>00570 
-<a name="l00571"></a>00571 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00572"></a><a class="code" href="classenorm.html#60b47544f6181ffd4530d3e415ce12c5">00572</a> vec <a class="code" href="classenorm.html#60b47544f6181ffd4530d3e415ce12c5" title="Returns a sample,  from density .">enorm&lt;sq_T&gt;::sample</a>()<span class="keyword"> const </span>{
-<a name="l00573"></a>00573         vec x ( <a class="code" href="classenorm.html#6938fc390a19cdaf6ad4503fcbaada4e" title="dimension (redundant from rv.count() for easier coding )">dim</a> );
-<a name="l00574"></a>00574         NorRNG.sample_vector ( <a class="code" href="classenorm.html#6938fc390a19cdaf6ad4503fcbaada4e" title="dimension (redundant from rv.count() for easier coding )">dim</a>,x );
-<a name="l00575"></a>00575         vec smp = <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>.sqrt_mult ( x );
-<a name="l00576"></a>00576 
-<a name="l00577"></a>00577         smp += <a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a>;
-<a name="l00578"></a>00578         <span class="keywordflow">return</span> smp;
-<a name="l00579"></a>00579 };
-<a name="l00580"></a>00580 
-<a name="l00581"></a>00581 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00582"></a><a class="code" href="classenorm.html#60f0f3bfa53d6e65843eea9532b16d36">00582</a> mat <a class="code" href="classenorm.html#60b47544f6181ffd4530d3e415ce12c5" title="Returns a sample,  from density .">enorm&lt;sq_T&gt;::sample</a> ( <span class="keywordtype">int</span> N )<span class="keyword"> const </span>{
-<a name="l00583"></a>00583         mat X ( <a class="code" href="classenorm.html#6938fc390a19cdaf6ad4503fcbaada4e" title="dimension (redundant from rv.count() for easier coding )">dim</a>,N );
-<a name="l00584"></a>00584         vec x ( <a class="code" href="classenorm.html#6938fc390a19cdaf6ad4503fcbaada4e" title="dimension (redundant from rv.count() for easier coding )">dim</a> );
-<a name="l00585"></a>00585         vec pom;
-<a name="l00586"></a>00586         <span class="keywordtype">int</span> i;
-<a name="l00587"></a>00587 
-<a name="l00588"></a>00588         <span class="keywordflow">for</span> ( i=0;i&lt;N;i++ ) {
-<a name="l00589"></a>00589                 NorRNG.sample_vector ( <a class="code" href="classenorm.html#6938fc390a19cdaf6ad4503fcbaada4e" title="dimension (redundant from rv.count() for easier coding )">dim</a>,x );
-<a name="l00590"></a>00590                 pom = <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>.sqrt_mult ( x );
-<a name="l00591"></a>00591                 pom +=<a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a>;
-<a name="l00592"></a>00592                 X.set_col ( i, pom );
-<a name="l00593"></a>00593         }
-<a name="l00594"></a>00594 
-<a name="l00595"></a>00595         <span class="keywordflow">return</span> X;
-<a name="l00596"></a>00596 };
-<a name="l00597"></a>00597 
-<a name="l00598"></a>00598 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00599"></a>00599 <span class="keywordtype">double</span> <a class="code" href="classenorm.html" title="Gaussian density with positive definite (decomposed) covariance matrix.">enorm&lt;sq_T&gt;::eval</a> ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const </span>{
-<a name="l00600"></a>00600         <span class="keywordtype">double</span> pdfl,e;
-<a name="l00601"></a>00601         pdfl = <a class="code" href="classeEF.html#6466e8d4aa9dd64698ed288cbb1afc03" title="Evaluate normalized log-probability.">evalpdflog</a> ( val );
-<a name="l00602"></a>00602         e = exp ( pdfl );
-<a name="l00603"></a>00603         <span class="keywordflow">return</span> e;
-<a name="l00604"></a>00604 };
-<a name="l00605"></a>00605 
-<a name="l00606"></a>00606 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00607"></a><a class="code" href="classenorm.html#c1e3dcba256b0153cfdb286120e110be">00607</a> <span class="keywordtype">double</span> <a class="code" href="classenorm.html#c1e3dcba256b0153cfdb286120e110be" title="Evaluate normalized log-probability.">enorm&lt;sq_T&gt;::evalpdflog_nn</a> ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const </span>{
-<a name="l00608"></a>00608         <span class="comment">// 1.83787706640935 = log(2pi)</span>
-<a name="l00609"></a>00609         <span class="keywordtype">double</span> tmp=-0.5* ( <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>.invqform ( <a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a>-val ) );<span class="comment">// - lognc();</span>
-<a name="l00610"></a>00610         <span class="keywordflow">return</span>  tmp;
-<a name="l00611"></a>00611 };
-<a name="l00612"></a>00612 
-<a name="l00613"></a>00613 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00614"></a><a class="code" href="classenorm.html#b289a36a69db59d182bb6eba9c05d4a8">00614</a> <span class="keyword">inline</span> <span class="keywordtype">double</span> <a class="code" href="classenorm.html#b289a36a69db59d182bb6eba9c05d4a8" title="logarithm of the normalizing constant, ">enorm&lt;sq_T&gt;::lognc</a> ()<span class="keyword"> const </span>{
-<a name="l00615"></a>00615         <span class="comment">// 1.83787706640935 = log(2pi)</span>
-<a name="l00616"></a>00616         <span class="keywordtype">double</span> tmp=0.5* ( <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>.cols() * 1.83787706640935 +<a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>.logdet() );
-<a name="l00617"></a>00617         <span class="keywordflow">return</span> tmp;
-<a name="l00618"></a>00618 };
-<a name="l00619"></a>00619 
-<a name="l00620"></a>00620 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00621"></a><a class="code" href="classmlnorm.html#3a5ad4798d8a3878c5e93b8e796c8837">00621</a> <a class="code" href="classmlnorm.html#3a5ad4798d8a3878c5e93b8e796c8837" title="Constructor.">mlnorm&lt;sq_T&gt;::mlnorm</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rv0, <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rvc0 ) :<a class="code" href="classmEF.html" title="Exponential family model.">mEF</a> ( rv0,rvc0 ),<a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a> ( rv0 ),A ( rv0.count(),rv0.count() ),<a class="code" href="classenorm.html#0b8cb284e5af920a1b64a21d057ec5ac" title="returns a pointer to the internal mean value. Use with Care!">_mu</a> ( <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>.<a class="code" href="classenorm.html#0b8cb284e5af920a1b64a21d057ec5ac" title="returns a pointer to the internal mean value. Use with Care!">_mu</a>() ) {
-<a name="l00622"></a>00622         <a class="code" href="classmpdf.html#7aa894208a32f3487827df6d5054424c" title="pointer to internal epdf">ep</a> =&amp;<a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>;
-<a name="l00623"></a>00623 }
+<a name="l00571"></a>00571 <span class="comment">// template&lt;class sq_T&gt;</span>
+<a name="l00572"></a>00572 <span class="comment">// void enorm&lt;sq_T&gt;::tupdate ( double phi, mat &amp;vbar, double nubar ) {</span>
+<a name="l00573"></a>00573 <span class="comment">//      //</span>
+<a name="l00574"></a>00574 <span class="comment">// };</span>
+<a name="l00575"></a>00575 
+<a name="l00576"></a>00576 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
+<a name="l00577"></a><a class="code" href="classenorm.html#60b47544f6181ffd4530d3e415ce12c5">00577</a> vec <a class="code" href="classenorm.html#60b47544f6181ffd4530d3e415ce12c5" title="Returns a sample,  from density .">enorm&lt;sq_T&gt;::sample</a>()<span class="keyword"> const </span>{
+<a name="l00578"></a>00578         vec x ( <a class="code" href="classenorm.html#6938fc390a19cdaf6ad4503fcbaada4e" title="dimension (redundant from rv.count() for easier coding )">dim</a> );
+<a name="l00579"></a>00579         NorRNG.sample_vector ( <a class="code" href="classenorm.html#6938fc390a19cdaf6ad4503fcbaada4e" title="dimension (redundant from rv.count() for easier coding )">dim</a>,x );
+<a name="l00580"></a>00580         vec smp = <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>.sqrt_mult ( x );
+<a name="l00581"></a>00581 
+<a name="l00582"></a>00582         smp += <a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a>;
+<a name="l00583"></a>00583         <span class="keywordflow">return</span> smp;
+<a name="l00584"></a>00584 };
+<a name="l00585"></a>00585 
+<a name="l00586"></a>00586 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
+<a name="l00587"></a><a class="code" href="classenorm.html#60f0f3bfa53d6e65843eea9532b16d36">00587</a> mat <a class="code" href="classenorm.html#60b47544f6181ffd4530d3e415ce12c5" title="Returns a sample,  from density .">enorm&lt;sq_T&gt;::sample</a> ( <span class="keywordtype">int</span> N )<span class="keyword"> const </span>{
+<a name="l00588"></a>00588         mat X ( <a class="code" href="classenorm.html#6938fc390a19cdaf6ad4503fcbaada4e" title="dimension (redundant from rv.count() for easier coding )">dim</a>,N );
+<a name="l00589"></a>00589         vec x ( <a class="code" href="classenorm.html#6938fc390a19cdaf6ad4503fcbaada4e" title="dimension (redundant from rv.count() for easier coding )">dim</a> );
+<a name="l00590"></a>00590         vec pom;
+<a name="l00591"></a>00591         <span class="keywordtype">int</span> i;
+<a name="l00592"></a>00592 
+<a name="l00593"></a>00593         <span class="keywordflow">for</span> ( i=0;i&lt;N;i++ ) {
+<a name="l00594"></a>00594                 NorRNG.sample_vector ( <a class="code" href="classenorm.html#6938fc390a19cdaf6ad4503fcbaada4e" title="dimension (redundant from rv.count() for easier coding )">dim</a>,x );
+<a name="l00595"></a>00595                 pom = <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>.sqrt_mult ( x );
+<a name="l00596"></a>00596                 pom +=<a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a>;
+<a name="l00597"></a>00597                 X.set_col ( i, pom );
+<a name="l00598"></a>00598         }
+<a name="l00599"></a>00599 
+<a name="l00600"></a>00600         <span class="keywordflow">return</span> X;
+<a name="l00601"></a>00601 };
+<a name="l00602"></a>00602 
+<a name="l00603"></a>00603 <span class="comment">// template&lt;class sq_T&gt;</span>
+<a name="l00604"></a>00604 <span class="comment">// double enorm&lt;sq_T&gt;::eval ( const vec &amp;val ) const {</span>
+<a name="l00605"></a>00605 <span class="comment">//      double pdfl,e;</span>
+<a name="l00606"></a>00606 <span class="comment">//      pdfl = evallog ( val );</span>
+<a name="l00607"></a>00607 <span class="comment">//      e = exp ( pdfl );</span>
+<a name="l00608"></a>00608 <span class="comment">//      return e;</span>
+<a name="l00609"></a>00609 <span class="comment">// };</span>
+<a name="l00610"></a>00610 
+<a name="l00611"></a>00611 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
+<a name="l00612"></a><a class="code" href="classenorm.html#50cb0a083d97a7adbbd97c92e712c46c">00612</a> <span class="keywordtype">double</span> <a class="code" href="classenorm.html#50cb0a083d97a7adbbd97c92e712c46c" title="Evaluate normalized log-probability.">enorm&lt;sq_T&gt;::evallog_nn</a> ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const </span>{
+<a name="l00613"></a>00613         <span class="comment">// 1.83787706640935 = log(2pi)</span>
+<a name="l00614"></a>00614         <span class="keywordtype">double</span> tmp=-0.5* ( <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>.invqform ( <a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a>-val ) );<span class="comment">// - lognc();</span>
+<a name="l00615"></a>00615         <span class="keywordflow">return</span>  tmp;
+<a name="l00616"></a>00616 };
+<a name="l00617"></a>00617 
+<a name="l00618"></a>00618 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
+<a name="l00619"></a><a class="code" href="classenorm.html#b289a36a69db59d182bb6eba9c05d4a8">00619</a> <span class="keyword">inline</span> <span class="keywordtype">double</span> <a class="code" href="classenorm.html#b289a36a69db59d182bb6eba9c05d4a8" title="logarithm of the normalizing constant, ">enorm&lt;sq_T&gt;::lognc</a> ()<span class="keyword"> const </span>{
+<a name="l00620"></a>00620         <span class="comment">// 1.83787706640935 = log(2pi)</span>
+<a name="l00621"></a>00621         <span class="keywordtype">double</span> tmp=0.5* ( <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>.cols() * 1.83787706640935 +<a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>.logdet() );
+<a name="l00622"></a>00622         <span class="keywordflow">return</span> tmp;
+<a name="l00623"></a>00623 };
 <a name="l00624"></a>00624 
 <a name="l00625"></a>00625 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00626"></a><a class="code" href="classmlnorm.html#f95dfce0b500636a44ecd7e5210de999">00626</a> <span class="keywordtype">void</span> <a class="code" href="classmlnorm.html#f95dfce0b500636a44ecd7e5210de999" title="Set A and R.">mlnorm&lt;sq_T&gt;::set_parameters</a> ( <span class="keyword">const</span> mat &amp;A0, <span class="keyword">const</span> vec &amp;mu0, <span class="keyword">const</span> sq_T &amp;R0 ) {
-<a name="l00627"></a>00627         <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>.set_parameters ( zeros ( <a class="code" href="classmpdf.html#f6687c07ff07d47812dd565368ca59eb" title="modeled random variable">rv</a>.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ),R0 );
-<a name="l00628"></a>00628         A = A0;
-<a name="l00629"></a>00629         mu_const = mu0;
-<a name="l00630"></a>00630 }
-<a name="l00631"></a>00631 
-<a name="l00632"></a>00632 <span class="comment">// template&lt;class sq_T&gt;</span>
-<a name="l00633"></a>00633 <span class="comment">// vec mlnorm&lt;sq_T&gt;::samplecond (const  vec &amp;cond, double &amp;lik ) {</span>
-<a name="l00634"></a>00634 <span class="comment">//      this-&gt;condition ( cond );</span>
-<a name="l00635"></a>00635 <span class="comment">//      vec smp = epdf.sample();</span>
-<a name="l00636"></a>00636 <span class="comment">//      lik = epdf.eval ( smp );</span>
-<a name="l00637"></a>00637 <span class="comment">//      return smp;</span>
-<a name="l00638"></a>00638 <span class="comment">// }</span>
-<a name="l00639"></a>00639 
-<a name="l00640"></a>00640 <span class="comment">// template&lt;class sq_T&gt;</span>
-<a name="l00641"></a>00641 <span class="comment">// mat mlnorm&lt;sq_T&gt;::samplecond (const vec &amp;cond, vec &amp;lik, int n ) {</span>
-<a name="l00642"></a>00642 <span class="comment">//      int i;</span>
-<a name="l00643"></a>00643 <span class="comment">//      int dim = rv.count();</span>
-<a name="l00644"></a>00644 <span class="comment">//      mat Smp ( dim,n );</span>
-<a name="l00645"></a>00645 <span class="comment">//      vec smp ( dim );</span>
-<a name="l00646"></a>00646 <span class="comment">//      this-&gt;condition ( cond );</span>
-<a name="l00647"></a>00647 <span class="comment">//</span>
-<a name="l00648"></a>00648 <span class="comment">//      for ( i=0; i&lt;n; i++ ) {</span>
-<a name="l00649"></a>00649 <span class="comment">//              smp = epdf.sample();</span>
-<a name="l00650"></a>00650 <span class="comment">//              lik ( i ) = epdf.eval ( smp );</span>
-<a name="l00651"></a>00651 <span class="comment">//              Smp.set_col ( i ,smp );</span>
-<a name="l00652"></a>00652 <span class="comment">//      }</span>
-<a name="l00653"></a>00653 <span class="comment">//</span>
-<a name="l00654"></a>00654 <span class="comment">//      return Smp;</span>
-<a name="l00655"></a>00655 <span class="comment">// }</span>
-<a name="l00656"></a>00656 
-<a name="l00657"></a>00657 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00658"></a><a class="code" href="classmlnorm.html#d41126455ac64b888a38f677886e1b40">00658</a> <span class="keywordtype">void</span> <a class="code" href="classmlnorm.html#d41126455ac64b888a38f677886e1b40" title="Set value of rvc . Result of this operation is stored in epdf use function _ep to...">mlnorm&lt;sq_T&gt;::condition</a> ( <span class="keyword">const</span> vec &amp;cond ) {
-<a name="l00659"></a>00659         _mu = A*cond + mu_const;
-<a name="l00660"></a>00660 <span class="comment">//R is already assigned;</span>
-<a name="l00661"></a>00661 }
-<a name="l00662"></a>00662 
-<a name="l00663"></a>00663 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00664"></a><a class="code" href="classenorm.html#af50a6102846060bcb23a670bf38117b">00664</a> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>* <a class="code" href="classenorm.html#af50a6102846060bcb23a670bf38117b" title="Return marginal density on the given RV, the remainig rvs are intergrated out.">enorm&lt;sq_T&gt;::marginal</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rvn )<span class="keyword"> const </span>{
-<a name="l00665"></a>00665         ivec irvn = rvn.<a class="code" href="classRV.html#bb724fa4e2d9ed7bfd0993b5975018a4">dataind</a> ( <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a> );
-<a name="l00666"></a>00666 
-<a name="l00667"></a>00667         sq_T Rn ( <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>,irvn );
-<a name="l00668"></a>00668         <a class="code" href="classenorm.html" title="Gaussian density with positive definite (decomposed) covariance matrix.">enorm&lt;sq_T&gt;</a>* tmp = <span class="keyword">new</span> <a class="code" href="classenorm.html" title="Gaussian density with positive definite (decomposed) covariance matrix.">enorm&lt;sq_T&gt;</a> ( rvn );
-<a name="l00669"></a>00669         tmp-&gt;<a class="code" href="classenorm.html#1394a65caa6e00d42e00cc99b12227af" title="Set mean value mu and covariance R.">set_parameters</a> ( <a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a> ( irvn ), Rn );
-<a name="l00670"></a>00670         <span class="keywordflow">return</span> tmp;
-<a name="l00671"></a>00671 }
-<a name="l00672"></a>00672 
-<a name="l00673"></a>00673 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00674"></a><a class="code" href="classenorm.html#921024bd6d5a0e65f2af2e39bf38dfca">00674</a> <a class="code" href="classmpdf.html" title="Conditional probability density, e.g. modeling some dependencies.">mpdf</a>* <a class="code" href="classenorm.html#921024bd6d5a0e65f2af2e39bf38dfca" title="Return conditional density on the given RV, the remaining rvs will be in conditioning...">enorm&lt;sq_T&gt;::condition</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rvn )<span class="keyword"> const </span>{
-<a name="l00675"></a>00675 
-<a name="l00676"></a>00676         <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> rvc = <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>.<a class="code" href="classRV.html#b9d175c327c21488b1e2fb756a84e149" title="Subtract another variable from the current one.">subt</a> ( rvn );
-<a name="l00677"></a>00677         it_assert_debug ( ( rvc.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() +rvn.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ==<a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ),<span class="stringliteral">"wrong rvn"</span> );
-<a name="l00678"></a>00678         <span class="comment">//Permutation vector of the new R</span>
-<a name="l00679"></a>00679         ivec irvn = rvn.<a class="code" href="classRV.html#bb724fa4e2d9ed7bfd0993b5975018a4">dataind</a> ( <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a> );
-<a name="l00680"></a>00680         ivec irvc = rvc.<a class="code" href="classRV.html#bb724fa4e2d9ed7bfd0993b5975018a4">dataind</a> ( <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a> );
-<a name="l00681"></a>00681         ivec perm=concat ( irvn , irvc );
-<a name="l00682"></a>00682         sq_T Rn ( <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>,perm );
-<a name="l00683"></a>00683 
-<a name="l00684"></a>00684         <span class="comment">//fixme - could this be done in general for all sq_T?</span>
-<a name="l00685"></a>00685         mat S=Rn.to_mat();
-<a name="l00686"></a>00686         <span class="comment">//fixme</span>
-<a name="l00687"></a>00687         <span class="keywordtype">int</span> n=rvn.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>()-1;
-<a name="l00688"></a>00688         <span class="keywordtype">int</span> end=<a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>.rows()-1;
-<a name="l00689"></a>00689         mat S11 = S.get ( 0,n, 0, n );
-<a name="l00690"></a>00690         mat S12 = S.get ( 0, n , rvn.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>(), end );
-<a name="l00691"></a>00691         mat S22 = S.get ( rvn.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>(), end, rvn.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>(), end );
-<a name="l00692"></a>00692 
-<a name="l00693"></a>00693         vec mu1 = <a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a> ( irvn );
-<a name="l00694"></a>00694         vec mu2 = <a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a> ( irvc );
-<a name="l00695"></a>00695         mat A=S12*inv ( S22 );
-<a name="l00696"></a>00696         sq_T R_n ( S11 - A *S12.T() );
+<a name="l00626"></a><a class="code" href="classmlnorm.html#3a5ad4798d8a3878c5e93b8e796c8837">00626</a> <a class="code" href="classmlnorm.html#3a5ad4798d8a3878c5e93b8e796c8837" title="Constructor.">mlnorm&lt;sq_T&gt;::mlnorm</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rv0, <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rvc0 ) :<a class="code" href="classmEF.html" title="Exponential family model.">mEF</a> ( rv0,rvc0 ),<a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a> ( rv0 ),A ( rv0.count(),rv0.count() ),<a class="code" href="classenorm.html#0b8cb284e5af920a1b64a21d057ec5ac" title="returns a pointer to the internal mean value. Use with Care!">_mu</a> ( <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>.<a class="code" href="classenorm.html#0b8cb284e5af920a1b64a21d057ec5ac" title="returns a pointer to the internal mean value. Use with Care!">_mu</a>() ) {
+<a name="l00627"></a>00627         <a class="code" href="classmpdf.html#7aa894208a32f3487827df6d5054424c" title="pointer to internal epdf">ep</a> =&amp;<a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>;
+<a name="l00628"></a>00628 }
+<a name="l00629"></a>00629 
+<a name="l00630"></a>00630 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
+<a name="l00631"></a><a class="code" href="classmlnorm.html#f95dfce0b500636a44ecd7e5210de999">00631</a> <span class="keywordtype">void</span> <a class="code" href="classmlnorm.html#f95dfce0b500636a44ecd7e5210de999" title="Set A and R.">mlnorm&lt;sq_T&gt;::set_parameters</a> ( <span class="keyword">const</span> mat &amp;A0, <span class="keyword">const</span> vec &amp;mu0, <span class="keyword">const</span> sq_T &amp;R0 ) {
+<a name="l00632"></a>00632         <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>.set_parameters ( zeros ( <a class="code" href="classmpdf.html#f6687c07ff07d47812dd565368ca59eb" title="modeled random variable">rv</a>.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ),R0 );
+<a name="l00633"></a>00633         A = A0;
+<a name="l00634"></a>00634         mu_const = mu0;
+<a name="l00635"></a>00635 }
+<a name="l00636"></a>00636 
+<a name="l00637"></a>00637 <span class="comment">// template&lt;class sq_T&gt;</span>
+<a name="l00638"></a>00638 <span class="comment">// vec mlnorm&lt;sq_T&gt;::samplecond (const  vec &amp;cond, double &amp;lik ) {</span>
+<a name="l00639"></a>00639 <span class="comment">//      this-&gt;condition ( cond );</span>
+<a name="l00640"></a>00640 <span class="comment">//      vec smp = epdf.sample();</span>
+<a name="l00641"></a>00641 <span class="comment">//      lik = epdf.eval ( smp );</span>
+<a name="l00642"></a>00642 <span class="comment">//      return smp;</span>
+<a name="l00643"></a>00643 <span class="comment">// }</span>
+<a name="l00644"></a>00644 
+<a name="l00645"></a>00645 <span class="comment">// template&lt;class sq_T&gt;</span>
+<a name="l00646"></a>00646 <span class="comment">// mat mlnorm&lt;sq_T&gt;::samplecond (const vec &amp;cond, vec &amp;lik, int n ) {</span>
+<a name="l00647"></a>00647 <span class="comment">//      int i;</span>
+<a name="l00648"></a>00648 <span class="comment">//      int dim = rv.count();</span>
+<a name="l00649"></a>00649 <span class="comment">//      mat Smp ( dim,n );</span>
+<a name="l00650"></a>00650 <span class="comment">//      vec smp ( dim );</span>
+<a name="l00651"></a>00651 <span class="comment">//      this-&gt;condition ( cond );</span>
+<a name="l00652"></a>00652 <span class="comment">//</span>
+<a name="l00653"></a>00653 <span class="comment">//      for ( i=0; i&lt;n; i++ ) {</span>
+<a name="l00654"></a>00654 <span class="comment">//              smp = epdf.sample();</span>
+<a name="l00655"></a>00655 <span class="comment">//              lik ( i ) = epdf.eval ( smp );</span>
+<a name="l00656"></a>00656 <span class="comment">//              Smp.set_col ( i ,smp );</span>
+<a name="l00657"></a>00657 <span class="comment">//      }</span>
+<a name="l00658"></a>00658 <span class="comment">//</span>
+<a name="l00659"></a>00659 <span class="comment">//      return Smp;</span>
+<a name="l00660"></a>00660 <span class="comment">// }</span>
+<a name="l00661"></a>00661 
+<a name="l00662"></a>00662 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
+<a name="l00663"></a><a class="code" href="classmlnorm.html#d41126455ac64b888a38f677886e1b40">00663</a> <span class="keywordtype">void</span> <a class="code" href="classmlnorm.html#d41126455ac64b888a38f677886e1b40" title="Set value of rvc . Result of this operation is stored in epdf use function _ep to...">mlnorm&lt;sq_T&gt;::condition</a> ( <span class="keyword">const</span> vec &amp;cond ) {
+<a name="l00664"></a>00664         _mu = A*cond + mu_const;
+<a name="l00665"></a>00665 <span class="comment">//R is already assigned;</span>
+<a name="l00666"></a>00666 }
+<a name="l00667"></a>00667 
+<a name="l00668"></a>00668 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
+<a name="l00669"></a><a class="code" href="classenorm.html#af50a6102846060bcb23a670bf38117b">00669</a> <a class="code" href="classepdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>* <a class="code" href="classenorm.html#af50a6102846060bcb23a670bf38117b" title="Return marginal density on the given RV, the remainig rvs are intergrated out.">enorm&lt;sq_T&gt;::marginal</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rvn )<span class="keyword"> const </span>{
+<a name="l00670"></a>00670         ivec irvn = rvn.<a class="code" href="classRV.html#bb724fa4e2d9ed7bfd0993b5975018a4">dataind</a> ( <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a> );
+<a name="l00671"></a>00671 
+<a name="l00672"></a>00672         sq_T Rn ( <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>,irvn );
+<a name="l00673"></a>00673         <a class="code" href="classenorm.html" title="Gaussian density with positive definite (decomposed) covariance matrix.">enorm&lt;sq_T&gt;</a>* tmp = <span class="keyword">new</span> <a class="code" href="classenorm.html" title="Gaussian density with positive definite (decomposed) covariance matrix.">enorm&lt;sq_T&gt;</a> ( rvn );
+<a name="l00674"></a>00674         tmp-&gt;<a class="code" href="classenorm.html#1394a65caa6e00d42e00cc99b12227af" title="Set mean value mu and covariance R.">set_parameters</a> ( <a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a> ( irvn ), Rn );
+<a name="l00675"></a>00675         <span class="keywordflow">return</span> tmp;
+<a name="l00676"></a>00676 }
+<a name="l00677"></a>00677 
+<a name="l00678"></a>00678 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
+<a name="l00679"></a><a class="code" href="classenorm.html#921024bd6d5a0e65f2af2e39bf38dfca">00679</a> <a class="code" href="classmpdf.html" title="Conditional probability density, e.g. modeling some dependencies.">mpdf</a>* <a class="code" href="classenorm.html#921024bd6d5a0e65f2af2e39bf38dfca" title="Return conditional density on the given RV, the remaining rvs will be in conditioning...">enorm&lt;sq_T&gt;::condition</a> ( <span class="keyword">const</span> <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> &amp;rvn )<span class="keyword"> const </span>{
+<a name="l00680"></a>00680 
+<a name="l00681"></a>00681         <a class="code" href="classRV.html" title="Class representing variables, most often random variables.">RV</a> rvc = <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>.<a class="code" href="classRV.html#b9d175c327c21488b1e2fb756a84e149" title="Subtract another variable from the current one.">subt</a> ( rvn );
+<a name="l00682"></a>00682         it_assert_debug ( ( rvc.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() +rvn.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ==<a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a>.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>() ),<span class="stringliteral">"wrong rvn"</span> );
+<a name="l00683"></a>00683         <span class="comment">//Permutation vector of the new R</span>
+<a name="l00684"></a>00684         ivec irvn = rvn.<a class="code" href="classRV.html#bb724fa4e2d9ed7bfd0993b5975018a4">dataind</a> ( <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a> );
+<a name="l00685"></a>00685         ivec irvc = rvc.<a class="code" href="classRV.html#bb724fa4e2d9ed7bfd0993b5975018a4">dataind</a> ( <a class="code" href="classepdf.html#74da992e3f5d598da8850b646b79b9d9" title="Identified of the random variable.">rv</a> );
+<a name="l00686"></a>00686         ivec perm=<a class="code" href="group__core.html#g33c114e83980d883c5b211c47d5322a4" title="Concat two random variables.">concat</a> ( irvn , irvc );
+<a name="l00687"></a>00687         sq_T Rn ( <a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>,perm );
+<a name="l00688"></a>00688 
+<a name="l00689"></a>00689         <span class="comment">//fixme - could this be done in general for all sq_T?</span>
+<a name="l00690"></a>00690         mat S=Rn.to_mat();
+<a name="l00691"></a>00691         <span class="comment">//fixme</span>
+<a name="l00692"></a>00692         <span class="keywordtype">int</span> n=rvn.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>()-1;
+<a name="l00693"></a>00693         <span class="keywordtype">int</span> end=<a class="code" href="classenorm.html#4ccc8d8514d644ef1c98d8ab023748a1" title="Covariance matrix in decomposed form.">R</a>.rows()-1;
+<a name="l00694"></a>00694         mat S11 = S.get ( 0,n, 0, n );
+<a name="l00695"></a>00695         mat S12 = S.get ( 0, n , rvn.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>(), end );
+<a name="l00696"></a>00696         mat S22 = S.get ( rvn.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>(), end, rvn.<a class="code" href="classRV.html#f5c7b8bd589eef09ccdf3329a0addea0" title="Return number of scalars in the RV.">count</a>(), end );
 <a name="l00697"></a>00697 
-<a name="l00698"></a>00698         <a class="code" href="classmlnorm.html" title="Normal distributed linear function with linear function of mean value;.">mlnorm&lt;sq_T&gt;</a>* tmp=<span class="keyword">new</span> <a class="code" href="classmlnorm.html" title="Normal distributed linear function with linear function of mean value;.">mlnorm&lt;sq_T&gt;</a> ( rvn,rvc );
-<a name="l00699"></a>00699 
-<a name="l00700"></a>00700         tmp-&gt;set_parameters ( A,mu1-A*mu2,R_n );
-<a name="l00701"></a>00701         <span class="keywordflow">return</span> tmp;
-<a name="l00702"></a>00702 }
-<a name="l00703"></a>00703 
-<a name="l00705"></a>00705 
-<a name="l00706"></a>00706 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
-<a name="l00707"></a>00707 std::ostream &amp;operator&lt;&lt; ( std::ostream &amp;os,  mlnorm&lt;sq_T&gt; &amp;ml ) {
-<a name="l00708"></a>00708         os &lt;&lt; <span class="stringliteral">"A:"</span>&lt;&lt; ml.A&lt;&lt;endl;
-<a name="l00709"></a>00709         os &lt;&lt; <span class="stringliteral">"mu:"</span>&lt;&lt; ml.mu_const&lt;&lt;endl;
-<a name="l00710"></a>00710         os &lt;&lt; <span class="stringliteral">"R:"</span> &lt;&lt; ml.epdf._R().to_mat() &lt;&lt;endl;
-<a name="l00711"></a>00711         <span class="keywordflow">return</span> os;
-<a name="l00712"></a>00712 };
-<a name="l00713"></a>00713 
-<a name="l00714"></a>00714 <span class="preprocessor">#endif //EF_H</span>
+<a name="l00698"></a>00698         vec mu1 = <a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a> ( irvn );
+<a name="l00699"></a>00699         vec mu2 = <a class="code" href="classenorm.html#71fde0d54bba147e00f612577f95ad20" title="mean value">mu</a> ( irvc );
+<a name="l00700"></a>00700         mat A=S12*inv ( S22 );
+<a name="l00701"></a>00701         sq_T R_n ( S11 - A *S12.T() );
+<a name="l00702"></a>00702 
+<a name="l00703"></a>00703         <a class="code" href="classmlnorm.html" title="Normal distributed linear function with linear function of mean value;.">mlnorm&lt;sq_T&gt;</a>* tmp=<span class="keyword">new</span> <a class="code" href="classmlnorm.html" title="Normal distributed linear function with linear function of mean value;.">mlnorm&lt;sq_T&gt;</a> ( rvn,rvc );
+<a name="l00704"></a>00704 
+<a name="l00705"></a>00705         tmp-&gt;set_parameters ( A,mu1-A*mu2,R_n );
+<a name="l00706"></a>00706         <span class="keywordflow">return</span> tmp;
+<a name="l00707"></a>00707 }
+<a name="l00708"></a>00708 
+<a name="l00710"></a>00710 
+<a name="l00711"></a>00711 <span class="keyword">template</span>&lt;<span class="keyword">class</span> sq_T&gt;
+<a name="l00712"></a>00712 std::ostream &amp;operator&lt;&lt; ( std::ostream &amp;os,  mlnorm&lt;sq_T&gt; &amp;ml ) {
+<a name="l00713"></a>00713         os &lt;&lt; <span class="stringliteral">"A:"</span>&lt;&lt; ml.A&lt;&lt;endl;
+<a name="l00714"></a>00714         os &lt;&lt; <span class="stringliteral">"mu:"</span>&lt;&lt; ml.mu_const&lt;&lt;endl;
+<a name="l00715"></a>00715         os &lt;&lt; <span class="stringliteral">"R:"</span> &lt;&lt; ml.epdf._R().to_mat() &lt;&lt;endl;
+<a name="l00716"></a>00716         <span class="keywordflow">return</span> os;
+<a name="l00717"></a>00717 };
+<a name="l00718"></a>00718 
+<a name="l00719"></a>00719 <span class="preprocessor">#endif //EF_H</span>
 </pre></div></div>
-<hr size="1"><address style="text-align: right;"><small>Generated on Wed Nov 12 20:46:06 2008 for mixpp by&nbsp;
+<hr size="1"><address style="text-align: right;"><small>Generated on Thu Dec 4 14:42:13 2008 for mixpp by&nbsp;
 <a href="http://www.doxygen.org/index.html">
 <img src="doxygen.png" alt="doxygen" align="middle" border="0"></a> 1.5.6 </small></address>