1 | #include "particles.h" |
---|
2 | |
---|
3 | namespace bdm { |
---|
4 | |
---|
5 | using std::endl; |
---|
6 | |
---|
7 | void PF::bayes_gensmp ( const vec &ut ) { |
---|
8 | if ( ut.length() > 0 ) { |
---|
9 | vec cond ( par->dimensionc() ); |
---|
10 | cond.set_subvector ( par->dimension(), ut ); |
---|
11 | #pragma parallel for |
---|
12 | for ( int i = 0; i < n; i++ ) { |
---|
13 | cond.set_subvector ( 0, _samples ( i ) ); |
---|
14 | _samples ( i ) = par->samplecond ( cond ); |
---|
15 | lls ( i ) = 0; |
---|
16 | } |
---|
17 | } else { |
---|
18 | #pragma parallel for |
---|
19 | for ( int i = 0; i < n; i++ ) { |
---|
20 | _samples ( i ) = par->samplecond ( _samples ( i ) ); |
---|
21 | lls ( i ) = 0; |
---|
22 | } |
---|
23 | } |
---|
24 | } |
---|
25 | |
---|
26 | void PF::bayes_weights() { |
---|
27 | // |
---|
28 | double mlls = max ( lls ); |
---|
29 | // compute weights |
---|
30 | for ( int i = 0; i < n; i++ ) { |
---|
31 | _w ( i ) *= exp ( lls ( i ) - mlls ); // multiply w by likelihood |
---|
32 | } |
---|
33 | |
---|
34 | //renormalize |
---|
35 | double sw = sum ( _w ); |
---|
36 | if ( !std::isfinite ( sw ) ) { |
---|
37 | for ( int i = 0; i < n; i++ ) { |
---|
38 | if ( !std::isfinite ( _w ( i ) ) ) { |
---|
39 | _w ( i ) = 0; |
---|
40 | } |
---|
41 | } |
---|
42 | sw = sum ( _w ); |
---|
43 | if ( !std::isfinite ( sw ) || sw == 0.0 ) { |
---|
44 | bdm_error ( "Particle filter is lost; no particle is good enough." ); |
---|
45 | } |
---|
46 | } |
---|
47 | _w /= sw; |
---|
48 | } |
---|
49 | |
---|
50 | void PF::bayes ( const vec &yt, const vec &cond ) { |
---|
51 | const vec &ut = cond; //todo |
---|
52 | |
---|
53 | int i; |
---|
54 | // generate samples - time step |
---|
55 | bayes_gensmp ( ut ); |
---|
56 | // weight them - data step |
---|
57 | for ( i = 0; i < n; i++ ) { |
---|
58 | lls ( i ) += obs->evallogcond ( yt, _samples ( i ) ); //+= because lls may have something from gensmp! |
---|
59 | } |
---|
60 | |
---|
61 | bayes_weights(); |
---|
62 | |
---|
63 | if ( do_resampling() ) { |
---|
64 | est.resample ( resmethod ); |
---|
65 | } |
---|
66 | |
---|
67 | } |
---|
68 | |
---|
69 | // void PF::set_est ( const epdf &epdf0 ) { |
---|
70 | // int i; |
---|
71 | // |
---|
72 | // for ( i=0;i<n;i++ ) { |
---|
73 | // _samples ( i ) = epdf0.sample(); |
---|
74 | // } |
---|
75 | // } |
---|
76 | |
---|
77 | vec MPF::mpfepdf::mean() const { |
---|
78 | const vec &w = pf->posterior()._w(); |
---|
79 | vec pom = zeros ( BMs ( 0 )->posterior ().dimension() ); |
---|
80 | //compute mean of BMs |
---|
81 | for ( int i = 0; i < w.length(); i++ ) { |
---|
82 | pom += BMs ( i )->posterior().mean() * w ( i ); |
---|
83 | } |
---|
84 | return concat ( pf->posterior().mean(), pom ); |
---|
85 | } |
---|
86 | |
---|
87 | vec MPF::mpfepdf::variance() const { |
---|
88 | const vec &w = pf->posterior()._w(); |
---|
89 | |
---|
90 | vec pom = zeros ( BMs ( 0 )->posterior ().dimension() ); |
---|
91 | vec pom2 = zeros ( BMs ( 0 )->posterior ().dimension() ); |
---|
92 | vec mea; |
---|
93 | |
---|
94 | for ( int i = 0; i < w.length(); i++ ) { |
---|
95 | // save current mean |
---|
96 | mea = BMs ( i )->posterior().mean(); |
---|
97 | pom += mea * w ( i ); |
---|
98 | //compute variance |
---|
99 | pom2 += ( BMs ( i )->posterior().variance() + pow ( mea, 2 ) ) * w ( i ); |
---|
100 | } |
---|
101 | return concat ( pf->posterior().variance(), pom2 - pow ( pom, 2 ) ); |
---|
102 | } |
---|
103 | |
---|
104 | void MPF::mpfepdf::qbounds ( vec &lb, vec &ub, double perc ) const { |
---|
105 | //bounds on particles |
---|
106 | vec lbp; |
---|
107 | vec ubp; |
---|
108 | pf->posterior().qbounds ( lbp, ubp ); |
---|
109 | |
---|
110 | //bounds on Components |
---|
111 | int dimC = BMs ( 0 )->posterior().dimension(); |
---|
112 | int j; |
---|
113 | // temporary |
---|
114 | vec lbc ( dimC ); |
---|
115 | vec ubc ( dimC ); |
---|
116 | // minima and maxima |
---|
117 | vec Lbc ( dimC ); |
---|
118 | vec Ubc ( dimC ); |
---|
119 | Lbc = std::numeric_limits<double>::infinity(); |
---|
120 | Ubc = -std::numeric_limits<double>::infinity(); |
---|
121 | |
---|
122 | for ( int i = 0; i < BMs.length(); i++ ) { |
---|
123 | // check Coms |
---|
124 | BMs ( i )->posterior().qbounds ( lbc, ubc ); |
---|
125 | //save either minima or maxima |
---|
126 | for ( j = 0; j < dimC; j++ ) { |
---|
127 | if ( lbc ( j ) < Lbc ( j ) ) { |
---|
128 | Lbc ( j ) = lbc ( j ); |
---|
129 | } |
---|
130 | if ( ubc ( j ) > Ubc ( j ) ) { |
---|
131 | Ubc ( j ) = ubc ( j ); |
---|
132 | } |
---|
133 | } |
---|
134 | } |
---|
135 | lb = concat ( lbp, Lbc ); |
---|
136 | ub = concat ( ubp, Ubc ); |
---|
137 | } |
---|
138 | |
---|
139 | |
---|
140 | |
---|
141 | void MPF::bayes ( const vec &yt, const vec &cond ) { |
---|
142 | // follows PF::bayes in most places!!! |
---|
143 | int i; |
---|
144 | int n = pf->__w().length(); |
---|
145 | vec &lls = pf->_lls(); |
---|
146 | Array<vec> &samples = pf->__samples(); |
---|
147 | |
---|
148 | // generate samples - time step |
---|
149 | pf->bayes_gensmp ( vec ( 0 ) ); |
---|
150 | // weight them - data step |
---|
151 | #pragma parallel for |
---|
152 | for ( i = 0; i < n; i++ ) { |
---|
153 | vec bm_cond ( BMs ( i )->dimensionc() ); |
---|
154 | this2bm.fill_cond ( yt, cond, bm_cond ); |
---|
155 | pf2bm.filldown ( samples ( i ), bm_cond ); |
---|
156 | BMs ( i ) -> bayes ( this2bm.pushdown ( yt ), bm_cond ); |
---|
157 | lls ( i ) += BMs ( i )->_ll(); |
---|
158 | } |
---|
159 | |
---|
160 | pf->bayes_weights(); |
---|
161 | |
---|
162 | ivec ind; |
---|
163 | if ( pf->do_resampling() ) { |
---|
164 | pf->resample ( ind ); |
---|
165 | |
---|
166 | #pragma omp parallel for |
---|
167 | for ( i = 0; i < n; i++ ) { |
---|
168 | if ( ind ( i ) != i ) {//replace the current Bm by a new one |
---|
169 | delete BMs ( i ); |
---|
170 | BMs ( i ) = (BM*) BMs ( ind ( i ) )->_copy(); //copy constructor |
---|
171 | } |
---|
172 | }; |
---|
173 | } |
---|
174 | }; |
---|
175 | |
---|
176 | |
---|
177 | } |
---|
178 | //MPF::MPF:{} |
---|