Show
Ignore:
Timestamp:
08/30/09 22:13:15 (16 years ago)
Author:
smidl
Message:

doc

Files:
1 modified

Legend:

Unmodified
Added
Removed

  • library/doc/html/particles_8h_source.html

    r472 r591  
    1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> 
    2 <html><head><meta http-equiv="Content-Type" content="text/html;charset=UTF-8"> 
     1<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"> 
     2<html xmlns="http://www.w3.org/1999/xhtml"> 
     3<head> 
     4<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/> 
    35<title>mixpp: particles.h Source File</title> 
    4 <link href="tabs.css" rel="stylesheet" type="text/css"> 
    5 <link href="doxygen.css" rel="stylesheet" type="text/css"> 
    6 </head><body> 
    7 <!-- Generated by Doxygen 1.5.9 --> 
     6<link href="tabs.css" rel="stylesheet" type="text/css"/> 
     7<link href="doxygen.css" rel="stylesheet" type="text/css"/> 
     8</head> 
     9<body> 
     10<!-- Generated by Doxygen 1.6.1 --> 
    811<script type="text/javascript"> 
    912<!-- 
    … …  
    6770<a name="l00015"></a>00015 <span class="preprocessor"></span> 
    6871<a name="l00016"></a>00016  
    69 <a name="l00017"></a>00017 <span class="preprocessor">#include "../stat/exp_family.h"</span> 
     72<a name="l00017"></a>00017 <span class="preprocessor">#include &quot;../stat/exp_family.h&quot;</span> 
    7073<a name="l00018"></a>00018  
    7174<a name="l00019"></a>00019 <span class="keyword">namespace </span>bdm { 
    … …  
    7376<a name="l00027"></a><a class="code" href="classbdm_1_1PF.html">00027</a> <span class="keyword">class </span><a class="code" href="classbdm_1_1PF.html" title="Trivial particle filter with proposal density equal to parameter evolution model...">PF</a> : <span class="keyword">public</span> <a class="code" href="classbdm_1_1BM.html" title="Bayesian Model of a system, i.e. all uncertainty is modeled by probabilities.">BM</a> { 
    7477<a name="l00028"></a>00028 <span class="keyword">protected</span>: 
    75 <a name="l00030"></a><a class="code" href="classbdm_1_1PF.html#eeafaf9b8ad75fe62ee9fd6369e3f7fe">00030</a>         <span class="keywordtype">int</span> <a class="code" href="classbdm_1_1PF.html#eeafaf9b8ad75fe62ee9fd6369e3f7fe" title="number of particles;">n</a>; 
    76 <a name="l00032"></a><a class="code" href="classbdm_1_1PF.html#dc049265b9086cad7071f98d00a2b9af">00032</a>         <a class="code" href="classbdm_1_1eEmp.html" title="Weighted empirical density.">eEmp</a> <a class="code" href="classbdm_1_1PF.html#dc049265b9086cad7071f98d00a2b9af" title="posterior density">est</a>; 
    77 <a name="l00034"></a><a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3">00034</a>         vec &amp;<a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>; 
    78 <a name="l00036"></a><a class="code" href="classbdm_1_1PF.html#914bd66025692c4018dbd482cb3c47c1">00036</a>         Array&lt;vec&gt; &amp;<a class="code" href="classbdm_1_1PF.html#914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a>; 
    79 <a name="l00038"></a><a class="code" href="classbdm_1_1PF.html#521e9621d3b5e1274275f323691afdaf">00038</a>         <a class="code" href="classbdm_1_1mpdf.html" title="Conditional probability density, e.g. modeling some dependencies.">mpdf</a> *<a class="code" href="classbdm_1_1PF.html#521e9621d3b5e1274275f323691afdaf" title="Parameter evolution model.">par</a>; 
    80 <a name="l00040"></a><a class="code" href="classbdm_1_1PF.html#d6e7a62fba1e0a0d73c9b87f4fb683ec">00040</a>         <a class="code" href="classbdm_1_1mpdf.html" title="Conditional probability density, e.g. modeling some dependencies.">mpdf</a> *<a class="code" href="classbdm_1_1PF.html#d6e7a62fba1e0a0d73c9b87f4fb683ec" title="Observation model.">obs</a>; 
     78<a name="l00030"></a><a class="code" href="classbdm_1_1PF.html#aeeafaf9b8ad75fe62ee9fd6369e3f7fe">00030</a>         <span class="keywordtype">int</span> <a class="code" href="classbdm_1_1PF.html#aeeafaf9b8ad75fe62ee9fd6369e3f7fe" title="number of particles;">n</a>; 
     79<a name="l00032"></a><a class="code" href="classbdm_1_1PF.html#adc049265b9086cad7071f98d00a2b9af">00032</a>         <a class="code" href="classbdm_1_1eEmp.html" title="Weighted empirical density.">eEmp</a> <a class="code" href="classbdm_1_1PF.html#adc049265b9086cad7071f98d00a2b9af" title="posterior density">est</a>; 
     80<a name="l00034"></a><a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3">00034</a>         vec &amp;<a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>; 
     81<a name="l00036"></a><a class="code" href="classbdm_1_1PF.html#a914bd66025692c4018dbd482cb3c47c1">00036</a>         Array&lt;vec&gt; &amp;<a class="code" href="classbdm_1_1PF.html#a914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a>; 
     82<a name="l00038"></a><a class="code" href="classbdm_1_1PF.html#a521e9621d3b5e1274275f323691afdaf">00038</a>         <a class="code" href="classbdm_1_1mpdf.html" title="Conditional probability density, e.g. modeling , where  is random variable, rv, and...">mpdf</a> *<a class="code" href="classbdm_1_1PF.html#a521e9621d3b5e1274275f323691afdaf" title="Parameter evolution model.">par</a>; 
     83<a name="l00040"></a><a class="code" href="classbdm_1_1PF.html#ad6e7a62fba1e0a0d73c9b87f4fb683ec">00040</a>         <a class="code" href="classbdm_1_1mpdf.html" title="Conditional probability density, e.g. modeling , where  is random variable, rv, and...">mpdf</a> *<a class="code" href="classbdm_1_1PF.html#ad6e7a62fba1e0a0d73c9b87f4fb683ec" title="Observation model.">obs</a>; 
    8184<a name="l00041"></a>00041  
    82 <a name="l00043"></a><a class="code" href="classbdm_1_1PF.html#9932c7c5865954ef9a438afcbe944e52">00043</a>         RESAMPLING_METHOD <a class="code" href="classbdm_1_1PF.html#9932c7c5865954ef9a438afcbe944e52" title="which resampling method will be used">resmethod</a>; 
     85<a name="l00043"></a><a class="code" href="classbdm_1_1PF.html#a9932c7c5865954ef9a438afcbe944e52">00043</a>         RESAMPLING_METHOD <a class="code" href="classbdm_1_1PF.html#a9932c7c5865954ef9a438afcbe944e52" title="which resampling method will be used">resmethod</a>; 
    8386<a name="l00044"></a>00044  
    8487<a name="l00047"></a>00047  
    85 <a name="l00049"></a><a class="code" href="classbdm_1_1PF.html#98ef9ff80c394fafd28680b7a3f831b1">00049</a>         <span class="keywordtype">bool</span> <a class="code" href="classbdm_1_1PF.html#98ef9ff80c394fafd28680b7a3f831b1" title="Log all samples.">opt_L_smp</a>; 
    86 <a name="l00051"></a><a class="code" href="classbdm_1_1PF.html#5a49463a88ee80771a464861df845ff6">00051</a>         <span class="keywordtype">bool</span> <a class="code" href="classbdm_1_1PF.html#5a49463a88ee80771a464861df845ff6" title="Log all samples.">opt_L_wei</a>; 
     88<a name="l00049"></a><a class="code" href="classbdm_1_1PF.html#a98ef9ff80c394fafd28680b7a3f831b1">00049</a>         <span class="keywordtype">bool</span> <a class="code" href="classbdm_1_1PF.html#a98ef9ff80c394fafd28680b7a3f831b1" title="Log all samples.">opt_L_smp</a>; 
     89<a name="l00051"></a><a class="code" href="classbdm_1_1PF.html#a5a49463a88ee80771a464861df845ff6">00051</a>         <span class="keywordtype">bool</span> <a class="code" href="classbdm_1_1PF.html#a5a49463a88ee80771a464861df845ff6" title="Log all samples.">opt_L_wei</a>; 
    8790<a name="l00053"></a>00053  
    8891<a name="l00054"></a>00054 <span class="keyword">public</span>: 
    89 <a name="l00057"></a>00057         <a class="code" href="classbdm_1_1PF.html" title="Trivial particle filter with proposal density equal to parameter evolution model...">PF</a> ( ) :<a class="code" href="classbdm_1_1PF.html#dc049265b9086cad7071f98d00a2b9af" title="posterior density">est</a>(), <a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( <a class="code" href="classbdm_1_1PF.html#dc049265b9086cad7071f98d00a2b9af" title="posterior density">est</a>.<a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>() ),<a class="code" href="classbdm_1_1PF.html#914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a> ( <a class="code" href="classbdm_1_1PF.html#dc049265b9086cad7071f98d00a2b9af" title="posterior density">est</a>.<a class="code" href="classbdm_1_1PF.html#914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a>() ), <a class="code" href="classbdm_1_1PF.html#98ef9ff80c394fafd28680b7a3f831b1" title="Log all samples.">opt_L_smp</a> ( false ), <a class="code" href="classbdm_1_1PF.html#5a49463a88ee80771a464861df845ff6" title="Log all samples.">opt_L_wei</a> ( false ) {<a class="code" href="classbdm_1_1BM.html#109c1a626a69031658e3a44e9e500cca" title="IDs of storages in loggers 4:[1=mean,2=lb,3=ub,4=ll].">LIDs</a>.set_size ( 5 );}; 
    90 <a name="l00058"></a>00058         <span class="comment">/*      PF ( mpdf *par0, mpdf *obs0, epdf *epdf0, int n0 ) :</span> 
    91 <a name="l00059"></a>00059 <span class="comment">                                est ( ),_w ( est._w() ),_samples ( est._samples() ),opt_L_smp(false), opt_L_wei(false)</span> 
    92 <a name="l00060"></a>00060 <span class="comment">                { set_parameters ( par0,obs0,n0 ); set_statistics ( ones ( n0 ),epdf0 ); };*/</span> 
    93 <a name="l00061"></a>00061         <span class="keywordtype">void</span> set_parameters ( <a class="code" href="classbdm_1_1mpdf.html" title="Conditional probability density, e.g. modeling some dependencies.">mpdf</a> *par0, <a class="code" href="classbdm_1_1mpdf.html" title="Conditional probability density, e.g. modeling some dependencies.">mpdf</a> *obs0, <span class="keywordtype">int</span> n0, RESAMPLING_METHOD rm=SYSTEMATIC ) 
    94 <a name="l00062"></a>00062         { <a class="code" href="classbdm_1_1PF.html#521e9621d3b5e1274275f323691afdaf" title="Parameter evolution model.">par</a> = par0; <a class="code" href="classbdm_1_1PF.html#d6e7a62fba1e0a0d73c9b87f4fb683ec" title="Observation model.">obs</a>=obs0; <a class="code" href="classbdm_1_1PF.html#eeafaf9b8ad75fe62ee9fd6369e3f7fe" title="number of particles;">n</a>=n0; <a class="code" href="classbdm_1_1PF.html#9932c7c5865954ef9a438afcbe944e52" title="which resampling method will be used">resmethod</a>= rm;}; 
    95 <a name="l00063"></a>00063         <span class="keywordtype">void</span> set_statistics ( <span class="keyword">const</span> vec w0, epdf *epdf0 ) {<a class="code" href="classbdm_1_1PF.html#dc049265b9086cad7071f98d00a2b9af" title="posterior density">est</a>.set_statistics ( w0,epdf0 );}; 
    96 <a name="l00066"></a>00066 <span class="comment">//      void set_est ( const epdf &amp;epdf0 );</span> 
    97 <a name="l00067"></a><a class="code" href="classbdm_1_1PF.html#bf104b869b5df8dd4a14bbe430d40488">00067</a>         <span class="keywordtype">void</span> <a class="code" href="classbdm_1_1PF.html#bf104b869b5df8dd4a14bbe430d40488">set_options</a> ( <span class="keyword">const</span> <span class="keywordtype">string</span> &amp;opt ) { 
    98 <a name="l00068"></a>00068                 <a class="code" href="classbdm_1_1PF.html#bf104b869b5df8dd4a14bbe430d40488">BM::set_options</a>(opt); 
    99 <a name="l00069"></a>00069                 <a class="code" href="classbdm_1_1PF.html#5a49463a88ee80771a464861df845ff6" title="Log all samples.">opt_L_wei</a>= ( opt.find ( <span class="stringliteral">"logweights"</span> ) !=string::npos ); 
    100 <a name="l00070"></a>00070                 <a class="code" href="classbdm_1_1PF.html#98ef9ff80c394fafd28680b7a3f831b1" title="Log all samples.">opt_L_smp</a>= ( opt.find ( <span class="stringliteral">"logsamples"</span> ) !=string::npos ); 
    101 <a name="l00071"></a>00071         } 
    102 <a name="l00072"></a>00072         <span class="keywordtype">void</span> <a class="code" href="classbdm_1_1PF.html#638946eea22d4964bf9350286bb4efd8" title="Incremental Bayes rule.">bayes</a> ( <span class="keyword">const</span> vec &amp;dt ); 
    103 <a name="l00074"></a><a class="code" href="classbdm_1_1PF.html#78a9f6809827be1d9bfe215d03b1c6ed">00074</a>         vec* <a class="code" href="classbdm_1_1PF.html#78a9f6809827be1d9bfe215d03b1c6ed" title="access function">__w</a>() {<span class="keywordflow">return</span> &amp;<a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>;} 
    104 <a name="l00075"></a>00075 }; 
    105 <a name="l00076"></a>00076  
    106 <a name="l00083"></a>00083 <span class="keyword">template</span>&lt;<span class="keyword">class</span> BM_T&gt; 
    107 <a name="l00084"></a><a class="code" href="classbdm_1_1MPF.html">00084</a> <span class="keyword">class </span><a class="code" href="classbdm_1_1MPF.html" title="Marginalized Particle filter.">MPF</a> : <span class="keyword">public</span> <a class="code" href="classbdm_1_1PF.html" title="Trivial particle filter with proposal density equal to parameter evolution model...">PF</a> { 
    108 <a name="l00085"></a>00085         Array&lt;BM_T*&gt; BMs; 
     92<a name="l00057"></a>00057         <a class="code" href="classbdm_1_1PF.html" title="Trivial particle filter with proposal density equal to parameter evolution model...">PF</a> ( ) : <a class="code" href="classbdm_1_1PF.html#adc049265b9086cad7071f98d00a2b9af" title="posterior density">est</a>(), <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( <a class="code" href="classbdm_1_1PF.html#adc049265b9086cad7071f98d00a2b9af" title="posterior density">est</a>.<a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>() ), <a class="code" href="classbdm_1_1PF.html#a914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a> ( <a class="code" href="classbdm_1_1PF.html#adc049265b9086cad7071f98d00a2b9af" title="posterior density">est</a>.<a class="code" href="classbdm_1_1PF.html#a914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a>() ), <a class="code" href="classbdm_1_1PF.html#a98ef9ff80c394fafd28680b7a3f831b1" title="Log all samples.">opt_L_smp</a> ( false ), <a class="code" href="classbdm_1_1PF.html#a5a49463a88ee80771a464861df845ff6" title="Log all samples.">opt_L_wei</a> ( false ) { 
     93<a name="l00058"></a>00058                 <a class="code" href="classbdm_1_1BM.html#a109c1a626a69031658e3a44e9e500cca" title="IDs of storages in loggers 4:[1=mean,2=lb,3=ub,4=ll].">LIDs</a>.set_size ( 5 ); 
     94<a name="l00059"></a>00059         }; 
     95<a name="l00060"></a>00060         <span class="comment">/*      PF ( mpdf *par0, mpdf *obs0, epdf *epdf0, int n0 ) :</span> 
     96<a name="l00061"></a>00061 <span class="comment">                                est ( ),_w ( est._w() ),_samples ( est._samples() ),opt_L_smp(false), opt_L_wei(false)</span> 
     97<a name="l00062"></a>00062 <span class="comment">                { set_parameters ( par0,obs0,n0 ); set_statistics ( ones ( n0 ),epdf0 ); };*/</span> 
     98<a name="l00063"></a>00063         <span class="keywordtype">void</span> set_parameters ( <a class="code" href="classbdm_1_1mpdf.html" title="Conditional probability density, e.g. modeling , where  is random variable, rv, and...">mpdf</a> *par0, <a class="code" href="classbdm_1_1mpdf.html" title="Conditional probability density, e.g. modeling , where  is random variable, rv, and...">mpdf</a> *obs0, <span class="keywordtype">int</span> n0, RESAMPLING_METHOD rm = SYSTEMATIC ) { 
     99<a name="l00064"></a>00064                 <a class="code" href="classbdm_1_1PF.html#a521e9621d3b5e1274275f323691afdaf" title="Parameter evolution model.">par</a> = par0; 
     100<a name="l00065"></a>00065                 <a class="code" href="classbdm_1_1PF.html#ad6e7a62fba1e0a0d73c9b87f4fb683ec" title="Observation model.">obs</a> = obs0; 
     101<a name="l00066"></a>00066                 <a class="code" href="classbdm_1_1PF.html#aeeafaf9b8ad75fe62ee9fd6369e3f7fe" title="number of particles;">n</a> = n0; 
     102<a name="l00067"></a>00067                 <a class="code" href="classbdm_1_1PF.html#a9932c7c5865954ef9a438afcbe944e52" title="which resampling method will be used">resmethod</a> = rm; 
     103<a name="l00068"></a>00068         }; 
     104<a name="l00069"></a>00069         <span class="keywordtype">void</span> set_statistics ( <span class="keyword">const</span> vec w0, <span class="keyword">const</span> epdf &amp;epdf0 ) { 
     105<a name="l00070"></a>00070                 <a class="code" href="classbdm_1_1PF.html#adc049265b9086cad7071f98d00a2b9af" title="posterior density">est</a>.set_statistics ( w0, epdf0 ); 
     106<a name="l00071"></a>00071         }; 
     107<a name="l00074"></a>00074 <span class="comment">//      void set_est ( const epdf &amp;epdf0 );</span> 
     108<a name="l00075"></a><a class="code" href="classbdm_1_1PF.html#abf104b869b5df8dd4a14bbe430d40488">00075</a>         <span class="keywordtype">void</span> <a class="code" href="classbdm_1_1PF.html#abf104b869b5df8dd4a14bbe430d40488">set_options</a> ( <span class="keyword">const</span> <span class="keywordtype">string</span> &amp;opt ) { 
     109<a name="l00076"></a>00076                 <a class="code" href="classbdm_1_1PF.html#abf104b869b5df8dd4a14bbe430d40488">BM::set_options</a> ( opt ); 
     110<a name="l00077"></a>00077                 <a class="code" href="classbdm_1_1PF.html#a5a49463a88ee80771a464861df845ff6" title="Log all samples.">opt_L_wei</a> = ( opt.find ( <span class="stringliteral">&quot;logweights&quot;</span> ) != string::npos ); 
     111<a name="l00078"></a>00078                 <a class="code" href="classbdm_1_1PF.html#a98ef9ff80c394fafd28680b7a3f831b1" title="Log all samples.">opt_L_smp</a> = ( opt.find ( <span class="stringliteral">&quot;logsamples&quot;</span> ) != string::npos ); 
     112<a name="l00079"></a>00079         } 
     113<a name="l00080"></a>00080         <span class="keywordtype">void</span> <a class="code" href="classbdm_1_1PF.html#a638946eea22d4964bf9350286bb4efd8" title="Incremental Bayes rule.">bayes</a> ( <span class="keyword">const</span> vec &amp;dt ); 
     114<a name="l00082"></a><a class="code" href="classbdm_1_1PF.html#a78a9f6809827be1d9bfe215d03b1c6ed">00082</a>         vec* <a class="code" href="classbdm_1_1PF.html#a78a9f6809827be1d9bfe215d03b1c6ed" title="access function">__w</a>() { 
     115<a name="l00083"></a>00083                 <span class="keywordflow">return</span> &amp;<a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>; 
     116<a name="l00084"></a>00084         } 
     117<a name="l00085"></a>00085 }; 
    109118<a name="l00086"></a>00086  
    110 <a name="l00088"></a>00088  
    111 <a name="l00089"></a>00089 <span class="keyword">class </span>mpfepdf : <span class="keyword">public</span> <a class="code" href="classbdm_1_1epdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>  { 
    112 <a name="l00090"></a>00090         <span class="keyword">protected</span>: 
    113 <a name="l00091"></a>00091                 <a class="code" href="classbdm_1_1eEmp.html" title="Weighted empirical density.">eEmp</a> &amp;E; 
    114 <a name="l00092"></a>00092                 vec &amp;<a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>; 
    115 <a name="l00093"></a>00093                 Array&lt;const epdf*&gt; Coms; 
    116 <a name="l00094"></a>00094         <span class="keyword">public</span>: 
    117 <a name="l00095"></a>00095                 mpfepdf ( <a class="code" href="classbdm_1_1eEmp.html" title="Weighted empirical density.">eEmp</a> &amp;E0 ) : 
    118 <a name="l00096"></a>00096                                 <a class="code" href="classbdm_1_1epdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a> ( ), E ( E0 ),  <a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( E._w() ), 
    119 <a name="l00097"></a>00097                                 Coms ( <a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>.length() ) { 
    120 <a name="l00098"></a>00098                 }; 
    121 <a name="l00100"></a>00100                 <span class="keywordtype">void</span> read_statistics ( Array&lt;BM_T*&gt; &amp;A ) { 
    122 <a name="l00101"></a>00101                         dim = E.dimension() +A ( 0 )-&gt;posterior().dimension(); 
    123 <a name="l00102"></a>00102                         <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i=0; i&lt;<a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>.length() ;i++ ) {Coms ( i ) = A ( i )-&gt;_e();} 
    124 <a name="l00103"></a>00103                 } 
    125 <a name="l00105"></a>00105                 <span class="keywordtype">void</span> set_elements ( <span class="keywordtype">int</span> &amp;i, <span class="keywordtype">double</span> wi, <span class="keyword">const</span> <a class="code" href="classbdm_1_1epdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>* ep ) 
    126 <a name="l00106"></a>00106                 {<a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( i ) =wi; Coms ( i ) =ep;}; 
    127 <a name="l00107"></a>00107  
    128 <a name="l00108"></a>00108                 <span class="keywordtype">void</span> set_parameters ( <span class="keywordtype">int</span> <a class="code" href="classbdm_1_1PF.html#eeafaf9b8ad75fe62ee9fd6369e3f7fe" title="number of particles;">n</a> ) { 
    129 <a name="l00109"></a>00109                         E.set_parameters ( n, <span class="keyword">false</span> ); 
    130 <a name="l00110"></a>00110                         Coms.set_length ( n ); 
    131 <a name="l00111"></a>00111                 } 
    132 <a name="l00112"></a>00112                 vec mean()<span class="keyword"> const </span>{ 
    133 <a name="l00113"></a>00113                         <span class="comment">// ugly</span> 
    134 <a name="l00114"></a>00114                         vec pom=zeros ( Coms ( 0 )-&gt;dimension() ); 
    135 <a name="l00115"></a>00115                         <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i=0; i&lt;<a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>.length(); i++ ) {pom += Coms ( i )-&gt;mean() * <a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( i );} 
    136 <a name="l00116"></a>00116                         <span class="keywordflow">return</span> concat ( E.mean(),pom ); 
    137 <a name="l00117"></a>00117                 } 
    138 <a name="l00118"></a>00118                 vec variance()<span class="keyword"> const </span>{ 
    139 <a name="l00119"></a>00119                         <span class="comment">// ugly</span> 
    140 <a name="l00120"></a>00120                         vec pom=zeros ( Coms ( 0 )-&gt;dimension() ); 
    141 <a name="l00121"></a>00121                         vec pom2=zeros ( Coms ( 0 )-&gt;dimension() ); 
    142 <a name="l00122"></a>00122                         <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i=0; i&lt;<a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>.length(); i++ ) { 
    143 <a name="l00123"></a>00123                                 pom += Coms ( i )-&gt;mean() * <a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( i ); 
    144 <a name="l00124"></a>00124                                 pom2 += ( Coms ( i )-&gt;variance() + pow ( Coms ( i )-&gt;mean(),2 ) ) * <a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( i ); 
    145 <a name="l00125"></a>00125                         } 
    146 <a name="l00126"></a>00126                         <span class="keywordflow">return</span> concat ( E.variance(),pom2-pow ( pom,2 ) ); 
    147 <a name="l00127"></a>00127                 } 
    148 <a name="l00128"></a>00128                 <span class="keywordtype">void</span> qbounds ( vec &amp;lb, vec &amp;ub, <span class="keywordtype">double</span> perc=0.95 )<span class="keyword"> const </span>{ 
    149 <a name="l00129"></a>00129                         <span class="comment">//bounds on particles</span> 
    150 <a name="l00130"></a>00130                         vec lbp; 
    151 <a name="l00131"></a>00131                         vec ubp; 
    152 <a name="l00132"></a>00132                         E.qbounds ( lbp,ubp ); 
    153 <a name="l00133"></a>00133  
    154 <a name="l00134"></a>00134                         <span class="comment">//bounds on Components</span> 
    155 <a name="l00135"></a>00135                         <span class="keywordtype">int</span> dimC=Coms ( 0 )-&gt;dimension(); 
    156 <a name="l00136"></a>00136                         <span class="keywordtype">int</span> j; 
    157 <a name="l00137"></a>00137                         <span class="comment">// temporary</span> 
    158 <a name="l00138"></a>00138                         vec lbc(dimC); 
    159 <a name="l00139"></a>00139                         vec ubc(dimC); 
    160 <a name="l00140"></a>00140                         <span class="comment">// minima and maxima</span> 
    161 <a name="l00141"></a>00141                         vec Lbc(dimC); 
    162 <a name="l00142"></a>00142                         vec Ubc(dimC); 
    163 <a name="l00143"></a>00143                         Lbc = std::numeric_limits&lt;double&gt;::infinity(); 
    164 <a name="l00144"></a>00144                         Ubc = -std::numeric_limits&lt;double&gt;::infinity(); 
    165 <a name="l00145"></a>00145  
    166 <a name="l00146"></a>00146                         <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i=0;i&lt;<a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>.length();i++ ) { 
    167 <a name="l00147"></a>00147                                 <span class="comment">// check Coms</span> 
    168 <a name="l00148"></a>00148                                 Coms ( i )-&gt;qbounds ( lbc,ubc ); 
    169 <a name="l00149"></a>00149                                 <span class="keywordflow">for</span> ( j=0;j&lt;dimC; j++ ) { 
    170 <a name="l00150"></a>00150                                         <span class="keywordflow">if</span> ( lbc ( j ) &lt;Lbc ( j ) ) {Lbc ( j ) =lbc ( j );} 
    171 <a name="l00151"></a>00151                                         <span class="keywordflow">if</span> ( ubc ( j ) &gt;Ubc ( j ) ) {Ubc ( j ) =ubc ( j );} 
    172 <a name="l00152"></a>00152                                 } 
    173 <a name="l00153"></a>00153                         } 
    174 <a name="l00154"></a>00154                         lb=concat(lbp,Lbc); 
    175 <a name="l00155"></a>00155                         ub=concat(ubp,Ubc); 
    176 <a name="l00156"></a>00156                 } 
    177 <a name="l00157"></a>00157  
    178 <a name="l00158"></a>00158                 vec sample()<span class="keyword"> const </span>{it_error ( <span class="stringliteral">"Not implemented"</span> );<span class="keywordflow">return</span> 0;} 
    179 <a name="l00159"></a>00159  
    180 <a name="l00160"></a>00160                 <span class="keywordtype">double</span> evallog ( <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;} 
    181 <a name="l00161"></a>00161         }; 
    182 <a name="l00162"></a>00162  
    183 <a name="l00164"></a>00164         mpfepdf jest; 
    184 <a name="l00165"></a>00165  
    185 <a name="l00167"></a>00167         <span class="keywordtype">bool</span> opt_L_mea; 
    186 <a name="l00168"></a>00168  
    187 <a name="l00169"></a>00169 <span class="keyword">public</span>: 
    188 <a name="l00171"></a><a class="code" href="classbdm_1_1MPF.html#0068e7ca53d90fa5911eb31a0d657f26">00171</a>         <a class="code" href="classbdm_1_1MPF.html#0068e7ca53d90fa5911eb31a0d657f26" title="Default constructor.">MPF</a> () : <a class="code" href="classbdm_1_1PF.html" title="Trivial particle filter with proposal density equal to parameter evolution model...">PF</a> (), jest ( <a class="code" href="classbdm_1_1PF.html#dc049265b9086cad7071f98d00a2b9af" title="posterior density">est</a> ) {}; 
    189 <a name="l00172"></a>00172         <span class="keywordtype">void</span> set_parameters ( <a class="code" href="classbdm_1_1mpdf.html" title="Conditional probability density, e.g. modeling some dependencies.">mpdf</a> *par0, <a class="code" href="classbdm_1_1mpdf.html" title="Conditional probability density, e.g. modeling some dependencies.">mpdf</a> *obs0, <span class="keywordtype">int</span> n0, RESAMPLING_METHOD rm=SYSTEMATIC ) { 
    190 <a name="l00173"></a>00173                 PF::set_parameters ( par0, obs0, n0, rm ); 
    191 <a name="l00174"></a>00174                 jest.set_parameters ( n0 );<span class="comment">//duplication of rm</span> 
    192 <a name="l00175"></a>00175                 BMs.set_length ( n0 ); 
    193 <a name="l00176"></a>00176         } 
    194 <a name="l00177"></a>00177         <span class="keywordtype">void</span> set_statistics ( epdf *epdf0, <span class="keyword">const</span> BM_T* BMcond0 ) { 
    195 <a name="l00178"></a>00178  
    196 <a name="l00179"></a>00179                 PF::set_statistics ( ones ( <a class="code" href="classbdm_1_1PF.html#eeafaf9b8ad75fe62ee9fd6369e3f7fe" title="number of particles;">n</a> ) /<a class="code" href="classbdm_1_1PF.html#eeafaf9b8ad75fe62ee9fd6369e3f7fe" title="number of particles;">n</a>, epdf0 ); 
    197 <a name="l00180"></a>00180                 <span class="comment">// copy</span> 
    198 <a name="l00181"></a>00181                 <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i=0;i&lt;<a class="code" href="classbdm_1_1PF.html#eeafaf9b8ad75fe62ee9fd6369e3f7fe" title="number of particles;">n</a>;i++ ) { BMs ( i ) = <span class="keyword">new</span> BM_T ( *BMcond0 ); BMs ( i )-&gt;condition ( <a class="code" href="classbdm_1_1PF.html#914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a> ( i ) );} 
     119<a name="l00093"></a>00093 <span class="keyword">template</span>&lt;<span class="keyword">class</span> BM_T&gt; 
     120<a name="l00094"></a><a class="code" href="classbdm_1_1MPF.html">00094</a> <span class="keyword">class </span><a class="code" href="classbdm_1_1MPF.html" title="Marginalized Particle filter.">MPF</a> : <span class="keyword">public</span> <a class="code" href="classbdm_1_1PF.html" title="Trivial particle filter with proposal density equal to parameter evolution model...">PF</a> { 
     121<a name="l00095"></a>00095         Array&lt;BM_T*&gt; BMs; 
     122<a name="l00096"></a>00096  
     123<a name="l00098"></a>00098  
     124<a name="l00099"></a>00099         <span class="keyword">class </span>mpfepdf : <span class="keyword">public</span> <a class="code" href="classbdm_1_1epdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>  { 
     125<a name="l00100"></a>00100         <span class="keyword">protected</span>: 
     126<a name="l00101"></a>00101                 <a class="code" href="classbdm_1_1eEmp.html" title="Weighted empirical density.">eEmp</a> &amp;E; 
     127<a name="l00102"></a>00102                 vec &amp;<a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>; 
     128<a name="l00103"></a>00103                 Array&lt;const epdf*&gt; Coms; 
     129<a name="l00104"></a>00104         <span class="keyword">public</span>: 
     130<a name="l00105"></a>00105                 mpfepdf ( <a class="code" href="classbdm_1_1eEmp.html" title="Weighted empirical density.">eEmp</a> &amp;E0 ) : 
     131<a name="l00106"></a>00106                                 <a class="code" href="classbdm_1_1epdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a> ( ), E ( E0 ),  <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( E._w() ), 
     132<a name="l00107"></a>00107                                 Coms ( <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>.length() ) { 
     133<a name="l00108"></a>00108                 }; 
     134<a name="l00110"></a>00110                 <span class="keywordtype">void</span> read_statistics ( Array&lt;BM_T*&gt; &amp;A ) { 
     135<a name="l00111"></a>00111                         dim = E.dimension() + A ( 0 )-&gt;posterior().dimension(); 
     136<a name="l00112"></a>00112                         <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i = 0; i &lt; <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>.length() ; i++ ) { 
     137<a name="l00113"></a>00113                                 Coms ( i ) = &amp;(A ( i )-&gt;posterior()); 
     138<a name="l00114"></a>00114                         } 
     139<a name="l00115"></a>00115                 } 
     140<a name="l00117"></a>00117                 <span class="keywordtype">void</span> set_elements ( <span class="keywordtype">int</span> &amp;i, <span class="keywordtype">double</span> wi, <span class="keyword">const</span> <a class="code" href="classbdm_1_1epdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>* ep ) { 
     141<a name="l00118"></a>00118                         <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( i ) = wi; 
     142<a name="l00119"></a>00119                         Coms ( i ) = ep; 
     143<a name="l00120"></a>00120                 }; 
     144<a name="l00121"></a>00121  
     145<a name="l00122"></a>00122                 <span class="keywordtype">void</span> set_parameters ( <span class="keywordtype">int</span> <a class="code" href="classbdm_1_1PF.html#aeeafaf9b8ad75fe62ee9fd6369e3f7fe" title="number of particles;">n</a> ) { 
     146<a name="l00123"></a>00123                         E.set_parameters ( n, <span class="keyword">false</span> ); 
     147<a name="l00124"></a>00124                         Coms.set_length ( n ); 
     148<a name="l00125"></a>00125                 } 
     149<a name="l00126"></a>00126                 vec mean()<span class="keyword"> const </span>{ 
     150<a name="l00127"></a>00127                         <span class="comment">// ugly</span> 
     151<a name="l00128"></a>00128                         vec pom = zeros ( Coms ( 0 )-&gt;dimension() ); 
     152<a name="l00129"></a>00129                         <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i = 0; i &lt; <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>.length(); i++ ) { 
     153<a name="l00130"></a>00130                                 pom += Coms ( i )-&gt;mean() * <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( i ); 
     154<a name="l00131"></a>00131                         } 
     155<a name="l00132"></a>00132                         <span class="keywordflow">return</span> concat ( E.mean(), pom ); 
     156<a name="l00133"></a>00133                 } 
     157<a name="l00134"></a>00134                 vec variance()<span class="keyword"> const </span>{ 
     158<a name="l00135"></a>00135                         <span class="comment">// ugly</span> 
     159<a name="l00136"></a>00136                         vec pom = zeros ( Coms ( 0 )-&gt;dimension() ); 
     160<a name="l00137"></a>00137                         vec pom2 = zeros ( Coms ( 0 )-&gt;dimension() ); 
     161<a name="l00138"></a>00138                         <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i = 0; i &lt; <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>.length(); i++ ) { 
     162<a name="l00139"></a>00139                                 pom += Coms ( i )-&gt;mean() * <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( i ); 
     163<a name="l00140"></a>00140                                 pom2 += ( Coms ( i )-&gt;variance() + pow ( Coms ( i )-&gt;mean(), 2 ) ) * <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( i ); 
     164<a name="l00141"></a>00141                         } 
     165<a name="l00142"></a>00142                         <span class="keywordflow">return</span> concat ( E.variance(), pom2 - pow ( pom, 2 ) ); 
     166<a name="l00143"></a>00143                 } 
     167<a name="l00144"></a>00144                 <span class="keywordtype">void</span> qbounds ( vec &amp;lb, vec &amp;ub, <span class="keywordtype">double</span> perc = 0.95 )<span class="keyword"> const </span>{ 
     168<a name="l00145"></a>00145                         <span class="comment">//bounds on particles</span> 
     169<a name="l00146"></a>00146                         vec lbp; 
     170<a name="l00147"></a>00147                         vec ubp; 
     171<a name="l00148"></a>00148                         E.qbounds ( lbp, ubp ); 
     172<a name="l00149"></a>00149  
     173<a name="l00150"></a>00150                         <span class="comment">//bounds on Components</span> 
     174<a name="l00151"></a>00151                         <span class="keywordtype">int</span> dimC = Coms ( 0 )-&gt;dimension(); 
     175<a name="l00152"></a>00152                         <span class="keywordtype">int</span> j; 
     176<a name="l00153"></a>00153                         <span class="comment">// temporary</span> 
     177<a name="l00154"></a>00154                         vec lbc ( dimC ); 
     178<a name="l00155"></a>00155                         vec ubc ( dimC ); 
     179<a name="l00156"></a>00156                         <span class="comment">// minima and maxima</span> 
     180<a name="l00157"></a>00157                         vec Lbc ( dimC ); 
     181<a name="l00158"></a>00158                         vec Ubc ( dimC ); 
     182<a name="l00159"></a>00159                         Lbc = std::numeric_limits&lt;double&gt;::infinity(); 
     183<a name="l00160"></a>00160                         Ubc = -std::numeric_limits&lt;double&gt;::infinity(); 
     184<a name="l00161"></a>00161  
     185<a name="l00162"></a>00162                         <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i = 0; i &lt; <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>.length(); i++ ) { 
     186<a name="l00163"></a>00163                                 <span class="comment">// check Coms</span> 
     187<a name="l00164"></a>00164                                 Coms ( i )-&gt;qbounds ( lbc, ubc ); 
     188<a name="l00165"></a>00165                                 <span class="keywordflow">for</span> ( j = 0; j &lt; dimC; j++ ) { 
     189<a name="l00166"></a>00166                                         <span class="keywordflow">if</span> ( lbc ( j ) &lt; Lbc ( j ) ) { 
     190<a name="l00167"></a>00167                                                 Lbc ( j ) = lbc ( j ); 
     191<a name="l00168"></a>00168                                         } 
     192<a name="l00169"></a>00169                                         <span class="keywordflow">if</span> ( ubc ( j ) &gt; Ubc ( j ) ) { 
     193<a name="l00170"></a>00170                                                 Ubc ( j ) = ubc ( j ); 
     194<a name="l00171"></a>00171                                         } 
     195<a name="l00172"></a>00172                                 } 
     196<a name="l00173"></a>00173                         } 
     197<a name="l00174"></a>00174                         lb = concat ( lbp, Lbc ); 
     198<a name="l00175"></a>00175                         ub = concat ( ubp, Ubc ); 
     199<a name="l00176"></a>00176                 } 
     200<a name="l00177"></a>00177  
     201<a name="l00178"></a>00178                 vec sample()<span class="keyword"> const </span>{ 
     202<a name="l00179"></a>00179                         <a class="code" href="bdmerror_8h.html#a7c43f3a72afe68ab0c85663a1bb3521a" title="Unconditionally throw std::runtime_error.">bdm_error</a> ( <span class="stringliteral">&quot;Not implemented&quot;</span> ); 
     203<a name="l00180"></a>00180                         <span class="keywordflow">return</span> vec(); 
     204<a name="l00181"></a>00181                 } 
    199205<a name="l00182"></a>00182  
    200 <a name="l00183"></a>00183                 jest.read_statistics ( BMs ); 
    201 <a name="l00184"></a>00184                 <span class="comment">//options</span> 
    202 <a name="l00185"></a>00185         }; 
    203 <a name="l00186"></a>00186  
    204 <a name="l00187"></a>00187         <span class="keywordtype">void</span> <a class="code" href="classbdm_1_1MPF.html#286d040770d08bd7ff416cea617b1b14" title="Incremental Bayes rule.">bayes</a> ( <span class="keyword">const</span> vec &amp;dt ); 
    205 <a name="l00188"></a>00188         <span class="keyword">const</span> epdf&amp; posterior()<span class="keyword"> const </span>{<span class="keywordflow">return</span> jest;} 
    206 <a name="l00189"></a>00189         <span class="keyword">const</span> epdf* _e()<span class="keyword"> const </span>{<span class="keywordflow">return</span> &amp;jest;} <span class="comment">//Fixme: is it useful?</span> 
    207 <a name="l00191"></a>00191 <span class="comment"></span>        <span class="comment">/*      void set_est ( const epdf&amp; epdf0 ) {</span> 
    208 <a name="l00192"></a>00192 <span class="comment">                        PF::set_est ( epdf0 );  // sample params in condition</span> 
    209 <a name="l00193"></a>00193 <span class="comment">                        // copy conditions to BMs</span> 
    210 <a name="l00194"></a>00194 <span class="comment"></span> 
    211 <a name="l00195"></a>00195 <span class="comment">                        for ( int i=0;i&lt;n;i++ ) {BMs(i)-&gt;condition ( _samples ( i ) );}</span> 
    212 <a name="l00196"></a>00196 <span class="comment">                }*/</span> 
    213 <a name="l00197"></a><a class="code" href="classbdm_1_1MPF.html#2e95498dec734088ab9f4878ff404144">00197</a>         <span class="keywordtype">void</span> <a class="code" href="classbdm_1_1MPF.html#2e95498dec734088ab9f4878ff404144" title="Set postrior of rvc to samples from epdf0. Statistics of BMs are not re-computed!...">set_options</a> ( <span class="keyword">const</span> <span class="keywordtype">string</span> &amp;opt ) { 
    214 <a name="l00198"></a>00198                 <a class="code" href="classbdm_1_1MPF.html#2e95498dec734088ab9f4878ff404144" title="Set postrior of rvc to samples from epdf0. Statistics of BMs are not re-computed!...">PF::set_options</a> ( opt ); 
    215 <a name="l00199"></a>00199                 opt_L_mea = ( opt.find ( <span class="stringliteral">"logmeans"</span> ) !=string::npos ); 
    216 <a name="l00200"></a>00200         } 
    217 <a name="l00201"></a>00201  
    218 <a name="l00203"></a><a class="code" href="classbdm_1_1MPF.html#82b5a34d9ed0e78452f98d2ecbf1e93c">00203</a>         <a class="code" href="classbdm_1_1BM.html" title="Bayesian Model of a system, i.e. all uncertainty is modeled by probabilities.">BM</a>* <a class="code" href="classbdm_1_1MPF.html#82b5a34d9ed0e78452f98d2ecbf1e93c" title="Access function.">_BM</a> ( <span class="keywordtype">int</span> i ) {<span class="keywordflow">return</span> BMs ( i );} 
    219 <a name="l00204"></a>00204 }; 
    220 <a name="l00205"></a>00205  
    221 <a name="l00206"></a>00206 <span class="keyword">template</span>&lt;<span class="keyword">class</span> BM_T&gt; 
    222 <a name="l00207"></a><a class="code" href="classbdm_1_1MPF.html#286d040770d08bd7ff416cea617b1b14">00207</a> <span class="keywordtype">void</span> <a class="code" href="classbdm_1_1MPF.html#286d040770d08bd7ff416cea617b1b14" title="Incremental Bayes rule.">MPF&lt;BM_T&gt;::bayes</a> ( <span class="keyword">const</span> vec &amp;dt ) { 
    223 <a name="l00208"></a>00208         <span class="keywordtype">int</span> i; 
    224 <a name="l00209"></a>00209         vec lls ( n ); 
    225 <a name="l00210"></a>00210         vec llsP ( n ); 
    226 <a name="l00211"></a>00211         ivec ind; 
    227 <a name="l00212"></a>00212         <span class="keywordtype">double</span> mlls=-std::numeric_limits&lt;double&gt;::infinity(); 
    228 <a name="l00213"></a>00213  
    229 <a name="l00214"></a>00214 <span class="preprocessor">#pragma omp parallel for</span> 
    230 <a name="l00215"></a>00215 <span class="preprocessor"></span>        <span class="keywordflow">for</span> ( i=0;i&lt;n;i++ ) { 
    231 <a name="l00216"></a>00216                 <span class="comment">//generate new samples from paramater evolution model;</span> 
    232 <a name="l00217"></a>00217                 <a class="code" href="classbdm_1_1PF.html#914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a> ( i ) = <a class="code" href="classbdm_1_1PF.html#521e9621d3b5e1274275f323691afdaf" title="Parameter evolution model.">par</a>-&gt;<a class="code" href="classbdm_1_1mpdf.html#b024f2e636e5507c755de4a6df1f1e99" title="Returns a sample from the density conditioned on cond, .">samplecond</a> ( <a class="code" href="classbdm_1_1PF.html#914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a> ( i ) ); 
    233 <a name="l00218"></a>00218                 llsP ( i ) = <a class="code" href="classbdm_1_1PF.html#521e9621d3b5e1274275f323691afdaf" title="Parameter evolution model.">par</a>-&gt;e()-&gt;<a class="code" href="classbdm_1_1epdf.html#deab266d63c236c277538867d5c3f249">evallog</a> ( <a class="code" href="classbdm_1_1PF.html#914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a> ( i ) ); 
    234 <a name="l00219"></a>00219                 BMs ( i )-&gt;condition ( <a class="code" href="classbdm_1_1PF.html#914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a> ( i ) ); 
    235 <a name="l00220"></a>00220                 BMs ( i )-&gt;bayes ( dt ); 
    236 <a name="l00221"></a>00221                 lls ( i ) = BMs ( i )-&gt;_ll(); <span class="comment">// lls above is also in proposal her must be lls(i) =, not +=!!</span> 
    237 <a name="l00222"></a>00222                 <span class="keywordflow">if</span> ( lls ( i ) &gt;mlls ) mlls=lls ( i ); <span class="comment">//find maximum likelihood (for numerical stability)</span> 
    238 <a name="l00223"></a>00223         } 
    239 <a name="l00224"></a>00224  
    240 <a name="l00225"></a>00225         <span class="keywordtype">double</span> sum_w=0.0; 
    241 <a name="l00226"></a>00226         <span class="comment">// compute weights</span> 
    242 <a name="l00227"></a>00227 <span class="preprocessor">#pragma omp parallel for</span> 
    243 <a name="l00228"></a>00228 <span class="preprocessor"></span>        <span class="keywordflow">for</span> ( i=0;i&lt;n;i++ ) { 
    244 <a name="l00229"></a>00229                 <a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( i ) *= exp ( lls ( i ) - mlls ); <span class="comment">// multiply w by likelihood</span> 
    245 <a name="l00230"></a>00230                 sum_w+=<a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( i ); 
    246 <a name="l00231"></a>00231         } 
    247 <a name="l00232"></a>00232  
    248 <a name="l00233"></a>00233         <span class="keywordflow">if</span> ( sum_w  &gt;0.0 ) { 
    249 <a name="l00234"></a>00234                 <a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> /=sum_w; <span class="comment">//?</span> 
     206<a name="l00183"></a>00183                 <span class="keywordtype">double</span> evallog ( <span class="keyword">const</span> vec &amp;val )<span class="keyword"> const </span>{ 
     207<a name="l00184"></a>00184                         <a class="code" href="bdmerror_8h.html#a7c43f3a72afe68ab0c85663a1bb3521a" title="Unconditionally throw std::runtime_error.">bdm_error</a> ( <span class="stringliteral">&quot;not implemented&quot;</span> ); 
     208<a name="l00185"></a>00185                         <span class="keywordflow">return</span> 0.0; 
     209<a name="l00186"></a>00186                 } 
     210<a name="l00187"></a>00187         }; 
     211<a name="l00188"></a>00188  
     212<a name="l00190"></a>00190         mpfepdf jest; 
     213<a name="l00191"></a>00191  
     214<a name="l00193"></a>00193         <span class="keywordtype">bool</span> opt_L_mea; 
     215<a name="l00194"></a>00194  
     216<a name="l00195"></a>00195 <span class="keyword">public</span>: 
     217<a name="l00197"></a><a class="code" href="classbdm_1_1MPF.html#a0068e7ca53d90fa5911eb31a0d657f26">00197</a>         <a class="code" href="classbdm_1_1MPF.html#a0068e7ca53d90fa5911eb31a0d657f26" title="Default constructor.">MPF</a> () : <a class="code" href="classbdm_1_1PF.html" title="Trivial particle filter with proposal density equal to parameter evolution model...">PF</a> (), jest ( <a class="code" href="classbdm_1_1PF.html#adc049265b9086cad7071f98d00a2b9af" title="posterior density">est</a> ) {}; 
     218<a name="l00198"></a>00198         <span class="keywordtype">void</span> set_parameters ( <a class="code" href="classbdm_1_1mpdf.html" title="Conditional probability density, e.g. modeling , where  is random variable, rv, and...">mpdf</a> *par0, <a class="code" href="classbdm_1_1mpdf.html" title="Conditional probability density, e.g. modeling , where  is random variable, rv, and...">mpdf</a> *obs0, <span class="keywordtype">int</span> n0, RESAMPLING_METHOD rm = SYSTEMATIC ) { 
     219<a name="l00199"></a>00199                 PF::set_parameters ( par0, obs0, n0, rm ); 
     220<a name="l00200"></a>00200                 jest.set_parameters ( n0 );<span class="comment">//duplication of rm</span> 
     221<a name="l00201"></a>00201                 BMs.set_length ( n0 ); 
     222<a name="l00202"></a>00202         } 
     223<a name="l00203"></a>00203         <span class="keywordtype">void</span> set_statistics ( <span class="keyword">const</span> epdf &amp;epdf0, <span class="keyword">const</span> BM_T* BMcond0 ) { 
     224<a name="l00204"></a>00204  
     225<a name="l00205"></a>00205                 PF::set_statistics ( ones ( <a class="code" href="classbdm_1_1PF.html#aeeafaf9b8ad75fe62ee9fd6369e3f7fe" title="number of particles;">n</a> ) / <a class="code" href="classbdm_1_1PF.html#aeeafaf9b8ad75fe62ee9fd6369e3f7fe" title="number of particles;">n</a>, epdf0 ); 
     226<a name="l00206"></a>00206                 <span class="comment">// copy</span> 
     227<a name="l00207"></a>00207                 <span class="keywordflow">for</span> ( <span class="keywordtype">int</span> i = 0; i &lt; <a class="code" href="classbdm_1_1PF.html#aeeafaf9b8ad75fe62ee9fd6369e3f7fe" title="number of particles;">n</a>; i++ ) { 
     228<a name="l00208"></a>00208                         BMs ( i ) = <span class="keyword">new</span> BM_T ( *BMcond0 ); 
     229<a name="l00209"></a>00209                         BMs ( i )-&gt;condition ( <a class="code" href="classbdm_1_1PF.html#a914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a> ( i ) ); 
     230<a name="l00210"></a>00210                 } 
     231<a name="l00211"></a>00211  
     232<a name="l00212"></a>00212                 jest.read_statistics ( BMs ); 
     233<a name="l00213"></a>00213                 <span class="comment">//options</span> 
     234<a name="l00214"></a>00214         }; 
     235<a name="l00215"></a>00215  
     236<a name="l00216"></a>00216         <span class="keywordtype">void</span> <a class="code" href="classbdm_1_1MPF.html#a286d040770d08bd7ff416cea617b1b14" title="Incremental Bayes rule.">bayes</a> ( <span class="keyword">const</span> vec &amp;dt ); 
     237<a name="l00217"></a>00217         <span class="keyword">const</span> epdf&amp; posterior()<span class="keyword"> const </span>{ 
     238<a name="l00218"></a>00218                 <span class="keywordflow">return</span> jest; 
     239<a name="l00219"></a>00219         } 
     240<a name="l00221"></a>00221         <span class="comment">/*      void set_est ( const epdf&amp; epdf0 ) {</span> 
     241<a name="l00222"></a>00222 <span class="comment">                        PF::set_est ( epdf0 );  // sample params in condition</span> 
     242<a name="l00223"></a>00223 <span class="comment">                        // copy conditions to BMs</span> 
     243<a name="l00224"></a>00224 <span class="comment"></span> 
     244<a name="l00225"></a>00225 <span class="comment">                        for ( int i=0;i&lt;n;i++ ) {BMs(i)-&gt;condition ( _samples ( i ) );}</span> 
     245<a name="l00226"></a>00226 <span class="comment">                }*/</span> 
     246<a name="l00227"></a><a class="code" href="classbdm_1_1MPF.html#a2e95498dec734088ab9f4878ff404144">00227</a>         <span class="keywordtype">void</span> <a class="code" href="classbdm_1_1MPF.html#a2e95498dec734088ab9f4878ff404144" title="Set postrior of rvc to samples from epdf0. Statistics of BMs are not re-computed!...">set_options</a> ( <span class="keyword">const</span> <span class="keywordtype">string</span> &amp;opt ) { 
     247<a name="l00228"></a>00228                 <a class="code" href="classbdm_1_1MPF.html#a2e95498dec734088ab9f4878ff404144" title="Set postrior of rvc to samples from epdf0. Statistics of BMs are not re-computed!...">PF::set_options</a> ( opt ); 
     248<a name="l00229"></a>00229                 opt_L_mea = ( opt.find ( <span class="stringliteral">&quot;logmeans&quot;</span> ) != string::npos ); 
     249<a name="l00230"></a>00230         } 
     250<a name="l00231"></a>00231  
     251<a name="l00233"></a><a class="code" href="classbdm_1_1MPF.html#ab6e7b094cbe32944b8ccde5df73cd839">00233</a>         <span class="keyword">const</span> <a class="code" href="classbdm_1_1BM.html" title="Bayesian Model of a system, i.e. all uncertainty is modeled by probabilities.">BM</a>* <a class="code" href="classbdm_1_1MPF.html#ab6e7b094cbe32944b8ccde5df73cd839" title="Access function.">_BM</a> ( <span class="keywordtype">int</span> i ) { 
     252<a name="l00234"></a>00234                 <span class="keywordflow">return</span> BMs ( i ); 
    250253<a name="l00235"></a>00235         } 
    251 <a name="l00236"></a>00236         <span class="keywordflow">else</span> { 
    252 <a name="l00237"></a>00237                 cout&lt;&lt;<span class="stringliteral">"sum(w)==0"</span>&lt;&lt;endl; 
    253 <a name="l00238"></a>00238         } 
    254 <a name="l00239"></a>00239  
    255 <a name="l00240"></a>00240  
    256 <a name="l00241"></a>00241         <span class="keywordtype">double</span> eff = 1.0/ ( <a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a>*<a class="code" href="classbdm_1_1PF.html#f5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ); 
    257 <a name="l00242"></a>00242         <span class="keywordflow">if</span> ( eff &lt; ( 0.3*n ) ) { 
    258 <a name="l00243"></a>00243                 ind = <a class="code" href="classbdm_1_1PF.html#dc049265b9086cad7071f98d00a2b9af" title="posterior density">est</a>.<a class="code" href="classbdm_1_1eEmp.html#f06ce255de5dbb2313f52ee51f82ba3d" title="Function performs resampling, i.e. removal of low-weight samples and duplication...">resample</a> ( <a class="code" href="classbdm_1_1PF.html#9932c7c5865954ef9a438afcbe944e52" title="which resampling method will be used">resmethod</a> ); 
    259 <a name="l00244"></a>00244                 <span class="comment">// Resample Bms!</span> 
     254<a name="l00236"></a>00236 }; 
     255<a name="l00237"></a>00237  
     256<a name="l00238"></a>00238 <span class="keyword">template</span>&lt;<span class="keyword">class</span> BM_T&gt; 
     257<a name="l00239"></a><a class="code" href="classbdm_1_1MPF.html#a286d040770d08bd7ff416cea617b1b14">00239</a> <span class="keywordtype">void</span> <a class="code" href="classbdm_1_1MPF.html#a286d040770d08bd7ff416cea617b1b14" title="Incremental Bayes rule.">MPF&lt;BM_T&gt;::bayes</a> ( <span class="keyword">const</span> vec &amp;dt ) { 
     258<a name="l00240"></a>00240         <span class="keywordtype">int</span> i; 
     259<a name="l00241"></a>00241         vec lls ( n ); 
     260<a name="l00242"></a>00242         vec llsP ( n ); 
     261<a name="l00243"></a>00243         ivec ind; 
     262<a name="l00244"></a>00244         <span class="keywordtype">double</span> mlls = -std::numeric_limits&lt;double&gt;::infinity(); 
    260263<a name="l00245"></a>00245  
    261264<a name="l00246"></a>00246 <span class="preprocessor">#pragma omp parallel for</span> 
    262 <a name="l00247"></a>00247 <span class="preprocessor"></span>                <span class="keywordflow">for</span> ( i=0;i&lt;n;i++ ) { 
    263 <a name="l00248"></a>00248                         <span class="keywordflow">if</span> ( ind ( i ) !=i ) {<span class="comment">//replace the current Bm by a new one</span> 
    264 <a name="l00249"></a>00249                                 <span class="comment">//fixme this would require new assignment operator</span> 
    265 <a name="l00250"></a>00250                                 <span class="comment">// *Bms[i] = *Bms[ind ( i ) ];</span> 
    266 <a name="l00251"></a>00251  
    267 <a name="l00252"></a>00252                                 <span class="comment">// poor-man's solution: replicate constructor here</span> 
    268 <a name="l00253"></a>00253                                 <span class="comment">// copied from MPF::MPF</span> 
    269 <a name="l00254"></a>00254                                 <span class="keyword">delete</span> BMs ( i ); 
    270 <a name="l00255"></a>00255                                 BMs ( i ) = <span class="keyword">new</span> BM_T ( *BMs ( ind ( i ) ) ); <span class="comment">//copy constructor</span> 
    271 <a name="l00256"></a>00256                                 <span class="keyword">const</span> <a class="code" href="classbdm_1_1epdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>&amp; pom=BMs ( i )-&gt;posterior(); 
    272 <a name="l00257"></a>00257                                 jest.set_elements ( i,1.0/n,&amp;pom ); 
    273 <a name="l00258"></a>00258                         } 
    274 <a name="l00259"></a>00259                 }; 
    275 <a name="l00260"></a>00260                 cout &lt;&lt; <span class="charliteral">'.'</span>; 
    276 <a name="l00261"></a>00261         } 
    277 <a name="l00262"></a>00262 } 
    278 <a name="l00263"></a>00263  
    279 <a name="l00264"></a>00264 } 
    280 <a name="l00265"></a>00265 <span class="preprocessor">#endif // KF_H</span> 
    281 <a name="l00266"></a>00266 <span class="preprocessor"></span> 
    282 <a name="l00267"></a>00267  
     265<a name="l00247"></a>00247 <span class="preprocessor"></span>        <span class="keywordflow">for</span> ( i = 0; i &lt; n; i++ ) { 
     266<a name="l00248"></a>00248                 <span class="comment">//generate new samples from paramater evolution model;</span> 
     267<a name="l00249"></a>00249                 vec old_smp=<a class="code" href="classbdm_1_1PF.html#a914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a> ( i ); 
     268<a name="l00250"></a>00250                 <a class="code" href="classbdm_1_1PF.html#a914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a> ( i ) = <a class="code" href="classbdm_1_1PF.html#a521e9621d3b5e1274275f323691afdaf" title="Parameter evolution model.">par</a>-&gt;<a class="code" href="classbdm_1_1mpdf.html#af0c1db6fcbb3aae2dd6123884457a367" title="Returns a sample from the density conditioned on cond, .">samplecond</a> ( old_smp ); 
     269<a name="l00251"></a>00251                 llsP ( i ) = <a class="code" href="classbdm_1_1PF.html#a521e9621d3b5e1274275f323691afdaf" title="Parameter evolution model.">par</a>-&gt;<a class="code" href="classbdm_1_1mpdf.html#a6336a8a72462e2a56a3989a220f18b1b" title="Shortcut for conditioning and evaluation of the internal epdf. In some cases, this...">evallogcond</a> ( <a class="code" href="classbdm_1_1PF.html#a914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a> ( i ), old_smp ); 
     270<a name="l00252"></a>00252                 BMs ( i )-&gt;condition ( <a class="code" href="classbdm_1_1PF.html#a914bd66025692c4018dbd482cb3c47c1" title="pointer into eEmp ">_samples</a> ( i ) ); 
     271<a name="l00253"></a>00253                 BMs ( i )-&gt;bayes ( dt ); 
     272<a name="l00254"></a>00254                 lls ( i ) = BMs ( i )-&gt;_ll(); <span class="comment">// lls above is also in proposal her must be lls(i) =, not +=!!</span> 
     273<a name="l00255"></a>00255                 <span class="keywordflow">if</span> ( lls ( i ) &gt; mlls ) mlls = lls ( i ); <span class="comment">//find maximum likelihood (for numerical stability)</span> 
     274<a name="l00256"></a>00256         } 
     275<a name="l00257"></a>00257  
     276<a name="l00258"></a>00258         <span class="keywordtype">double</span> sum_w = 0.0; 
     277<a name="l00259"></a>00259         <span class="comment">// compute weights</span> 
     278<a name="l00260"></a>00260 <span class="preprocessor">#pragma omp parallel for</span> 
     279<a name="l00261"></a>00261 <span class="preprocessor"></span>        <span class="keywordflow">for</span> ( i = 0; i &lt; n; i++ ) { 
     280<a name="l00262"></a>00262                 <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( i ) *= exp ( lls ( i ) - mlls ); <span class="comment">// multiply w by likelihood</span> 
     281<a name="l00263"></a>00263                 sum_w += <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ( i ); 
     282<a name="l00264"></a>00264         } 
     283<a name="l00265"></a>00265  
     284<a name="l00266"></a>00266         <span class="keywordflow">if</span> ( sum_w  &gt; 0.0 ) { 
     285<a name="l00267"></a>00267                 <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> /= sum_w; <span class="comment">//?</span> 
     286<a name="l00268"></a>00268         } <span class="keywordflow">else</span> { 
     287<a name="l00269"></a>00269                 cout &lt;&lt; <span class="stringliteral">&quot;sum(w)==0&quot;</span> &lt;&lt; endl; 
     288<a name="l00270"></a>00270         } 
     289<a name="l00271"></a>00271  
     290<a name="l00272"></a>00272  
     291<a name="l00273"></a>00273         <span class="keywordtype">double</span> eff = 1.0 / ( <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> * <a class="code" href="classbdm_1_1PF.html#af5149d5522d1095d39240c4c607f61a3" title="pointer into eEmp ">_w</a> ); 
     292<a name="l00274"></a>00274         <span class="keywordflow">if</span> ( eff &lt; ( 0.3*n ) ) { 
     293<a name="l00275"></a>00275                 ind = <a class="code" href="classbdm_1_1PF.html#adc049265b9086cad7071f98d00a2b9af" title="posterior density">est</a>.<a class="code" href="classbdm_1_1eEmp.html#af06ce255de5dbb2313f52ee51f82ba3d" title="Function performs resampling, i.e. removal of low-weight samples and duplication...">resample</a> ( <a class="code" href="classbdm_1_1PF.html#a9932c7c5865954ef9a438afcbe944e52" title="which resampling method will be used">resmethod</a> ); 
     294<a name="l00276"></a>00276                 <span class="comment">// Resample Bms!</span> 
     295<a name="l00277"></a>00277  
     296<a name="l00278"></a>00278 <span class="preprocessor">#pragma omp parallel for</span> 
     297<a name="l00279"></a>00279 <span class="preprocessor"></span>                <span class="keywordflow">for</span> ( i = 0; i &lt; n; i++ ) { 
     298<a name="l00280"></a>00280                         <span class="keywordflow">if</span> ( ind ( i ) != i ) {<span class="comment">//replace the current Bm by a new one</span> 
     299<a name="l00281"></a>00281                                 <span class="comment">//fixme this would require new assignment operator</span> 
     300<a name="l00282"></a>00282                                 <span class="comment">// *Bms[i] = *Bms[ind ( i ) ];</span> 
     301<a name="l00283"></a>00283  
     302<a name="l00284"></a>00284                                 <span class="comment">// poor-man&apos;s solution: replicate constructor here</span> 
     303<a name="l00285"></a>00285                                 <span class="comment">// copied from MPF::MPF</span> 
     304<a name="l00286"></a>00286                                 <span class="keyword">delete</span> BMs ( i ); 
     305<a name="l00287"></a>00287                                 BMs ( i ) = <span class="keyword">new</span> BM_T ( *BMs ( ind ( i ) ) ); <span class="comment">//copy constructor</span> 
     306<a name="l00288"></a>00288                                 <span class="keyword">const</span> <a class="code" href="classbdm_1_1epdf.html" title="Probability density function with numerical statistics, e.g. posterior density.">epdf</a>&amp; pom = BMs ( i )-&gt;posterior(); 
     307<a name="l00289"></a>00289                                 jest.set_elements ( i, 1.0 / n, &amp;pom ); 
     308<a name="l00290"></a>00290                         } 
     309<a name="l00291"></a>00291                 }; 
     310<a name="l00292"></a>00292                 cout &lt;&lt; <span class="charliteral">&apos;.&apos;</span>; 
     311<a name="l00293"></a>00293         } 
     312<a name="l00294"></a>00294 } 
     313<a name="l00295"></a>00295  
     314<a name="l00296"></a>00296 } 
     315<a name="l00297"></a>00297 <span class="preprocessor">#endif // KF_H</span> 
     316<a name="l00298"></a>00298 <span class="preprocessor"></span> 
     317<a name="l00299"></a>00299  
    283318</pre></div></div> 
    284 <hr size="1"><address style="text-align: right;"><small>Generated on Wed Aug 5 00:06:48 2009 for mixpp by&nbsp; 
     319<hr size="1"/><address style="text-align: right;"><small>Generated on Sun Aug 30 22:10:49 2009 for mixpp by&nbsp; 
    285320<a href="http://www.doxygen.org/index.html"> 
    286 <img src="doxygen.png" alt="doxygen" align="middle" border="0"></a> 1.5.9 </small></address> 
     321<img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.6.1 </small></address> 
    287322</body> 
    288323</html>