[384] | 1 | #include "particles.h" |
---|
[8] | 2 | |
---|
[283] | 3 | namespace bdm { |
---|
[8] | 4 | |
---|
| 5 | using std::endl; |
---|
| 6 | |
---|
[665] | 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 | } |
---|
[11] | 23 | } |
---|
[638] | 24 | } |
---|
[11] | 25 | |
---|
[638] | 26 | void PF::bayes_weights(){ |
---|
| 27 | // |
---|
| 28 | double mlls=max(lls); |
---|
[32] | 29 | // compute weights |
---|
[638] | 30 | for (int i = 0; i < n; i++ ) { |
---|
[32] | 31 | _w ( i ) *= exp ( lls ( i ) - mlls ); // multiply w by likelihood |
---|
[11] | 32 | } |
---|
| 33 | |
---|
[32] | 34 | //renormalize |
---|
[638] | 35 | double sw=sum(_w); |
---|
[663] | 36 | if (!std::isfinite(sw)) { |
---|
[638] | 37 | for (int i=0;i<n;i++){ |
---|
[663] | 38 | if (!std::isfinite(_w(i))) {_w(i)=0;} |
---|
[638] | 39 | } |
---|
| 40 | sw = sum(_w); |
---|
[665] | 41 | if (!std::isfinite(sw) || sw==0.0) { |
---|
[638] | 42 | bdm_error("Particle filter is lost; no particle is good enough."); |
---|
| 43 | } |
---|
| 44 | } |
---|
| 45 | _w /= sw; |
---|
| 46 | } |
---|
| 47 | |
---|
| 48 | void PF::bayes ( const vec &dt ) { |
---|
[665] | 49 | vec yt=dt.left(obs->dimension()); |
---|
| 50 | vec ut=dt.get(obs->dimension(),dt.length()-1); |
---|
| 51 | |
---|
[638] | 52 | int i; |
---|
| 53 | // generate samples - time step |
---|
[665] | 54 | bayes_gensmp(ut); |
---|
[638] | 55 | // weight them - data step |
---|
[477] | 56 | for ( i = 0; i < n; i++ ) { |
---|
[665] | 57 | lls ( i ) += obs->evallogcond ( yt, _samples ( i ) ); //+= because lls may have something from gensmp! |
---|
[638] | 58 | } |
---|
[11] | 59 | |
---|
[638] | 60 | bayes_weights(); |
---|
[11] | 61 | |
---|
[638] | 62 | if (do_resampling()) { |
---|
| 63 | est.resample ( resmethod); |
---|
| 64 | } |
---|
[8] | 65 | |
---|
| 66 | } |
---|
| 67 | |
---|
[283] | 68 | // void PF::set_est ( const epdf &epdf0 ) { |
---|
| 69 | // int i; |
---|
[477] | 70 | // |
---|
[283] | 71 | // for ( i=0;i<n;i++ ) { |
---|
| 72 | // _samples ( i ) = epdf0.sample(); |
---|
| 73 | // } |
---|
| 74 | // } |
---|
[8] | 75 | |
---|
[638] | 76 | void MPF::bayes ( const vec &dt ) { |
---|
[665] | 77 | // follows PF::bayes in most places!!! |
---|
[638] | 78 | int i; |
---|
| 79 | int n=pf->__w().length(); |
---|
| 80 | vec &lls = pf->_lls(); |
---|
| 81 | |
---|
| 82 | // generate samples - time step |
---|
[665] | 83 | pf->bayes_gensmp(vec(0)); |
---|
[638] | 84 | // weight them - data step |
---|
| 85 | #pragma parallel for |
---|
| 86 | for ( i = 0; i < n; i++ ) { |
---|
| 87 | BMs(i) -> condition(pf->posterior()._sample(i)); |
---|
| 88 | BMs(i) -> bayes(dt); |
---|
| 89 | lls ( i ) += BMs(i)->_ll(); |
---|
| 90 | } |
---|
[32] | 91 | |
---|
[638] | 92 | pf->bayes_weights(); |
---|
| 93 | |
---|
| 94 | ivec ind; |
---|
| 95 | if (pf->do_resampling()) { |
---|
| 96 | pf->resample ( ind); |
---|
| 97 | |
---|
| 98 | #pragma omp parallel for |
---|
| 99 | for ( i = 0; i < n; i++ ) { |
---|
| 100 | if ( ind ( i ) != i ) {//replace the current Bm by a new one |
---|
| 101 | delete BMs ( i ); |
---|
| 102 | BMs ( i ) = BMs ( ind ( i ) )->_copy_(); //copy constructor |
---|
| 103 | } |
---|
| 104 | }; |
---|
| 105 | } |
---|
| 106 | }; |
---|
| 107 | |
---|
| 108 | |
---|
[254] | 109 | } |
---|
[32] | 110 | //MPF::MPF:{} |
---|