int main(void)
{
/* Local scalars */
char compz, compz_i;
lapack_int n, n_i;
lapack_int ldz, ldz_i;
lapack_int ldz_r;
lapack_int info, info_i;
lapack_int i;
int failed;
/* Local arrays */
float *d = NULL, *d_i = NULL;
float *e = NULL, *e_i = NULL;
lapack_complex_float *z = NULL, *z_i = NULL;
float *work = NULL, *work_i = NULL;
float *d_save = NULL;
float *e_save = NULL;
lapack_complex_float *z_save = NULL;
lapack_complex_float *z_r = NULL;
/* Iniitialize the scalar parameters */
init_scalars_csteqr( &compz, &n, &ldz );
ldz_r = n+2;
compz_i = compz;
n_i = n;
ldz_i = ldz;
/* Allocate memory for the LAPACK routine arrays */
d = (float *)LAPACKE_malloc( n * sizeof(float) );
e = (float *)LAPACKE_malloc( (n-1) * sizeof(float) );
z = (lapack_complex_float *)
LAPACKE_malloc( ldz*n * sizeof(lapack_complex_float) );
work = (float *)LAPACKE_malloc( ((MAX(1,2*n-2))) * sizeof(float) );
/* Allocate memory for the C interface function arrays */
d_i = (float *)LAPACKE_malloc( n * sizeof(float) );
e_i = (float *)LAPACKE_malloc( (n-1) * sizeof(float) );
z_i = (lapack_complex_float *)
LAPACKE_malloc( ldz*n * sizeof(lapack_complex_float) );
work_i = (float *)LAPACKE_malloc( ((MAX(1,2*n-2))) * sizeof(float) );
/* Allocate memory for the backup arrays */
d_save = (float *)LAPACKE_malloc( n * sizeof(float) );
e_save = (float *)LAPACKE_malloc( (n-1) * sizeof(float) );
z_save = (lapack_complex_float *)
LAPACKE_malloc( ldz*n * sizeof(lapack_complex_float) );
/* Allocate memory for the row-major arrays */
z_r = (lapack_complex_float *)
LAPACKE_malloc( n*(n+2) * sizeof(lapack_complex_float) );
/* Initialize input arrays */
init_d( n, d );
init_e( (n-1), e );
init_z( ldz*n, z );
init_work( (MAX(1,2*n-2)), work );
/* Backup the ouptut arrays */
for( i = 0; i < n; i++ ) {
d_save[i] = d[i];
}
for( i = 0; i < (n-1); i++ ) {
e_save[i] = e[i];
}
for( i = 0; i < ldz*n; i++ ) {
z_save[i] = z[i];
}
/* Call the LAPACK routine */
csteqr_( &compz, &n, d, e, z, &ldz, work, &info );
/* Initialize input data, call the column-major middle-level
* interface to LAPACK routine and check the results */
for( i = 0; i < n; i++ ) {
d_i[i] = d_save[i];
}
for( i = 0; i < (n-1); i++ ) {
e_i[i] = e_save[i];
}
for( i = 0; i < ldz*n; i++ ) {
z_i[i] = z_save[i];
}
for( i = 0; i < (MAX(1,2*n-2)); i++ ) {
work_i[i] = work[i];
}
info_i = LAPACKE_csteqr_work( LAPACK_COL_MAJOR, compz_i, n_i, d_i, e_i, z_i,
ldz_i, work_i );
failed = compare_csteqr( d, d_i, e, e_i, z, z_i, info, info_i, compz, ldz,
n );
if( failed == 0 ) {
printf( "PASSED: column-major middle-level interface to csteqr\n" );
} else {
printf( "FAILED: column-major middle-level interface to csteqr\n" );
}
/* Initialize input data, call the column-major high-level
* interface to LAPACK routine and check the results */
for( i = 0; i < n; i++ ) {
d_i[i] = d_save[i];
}
for( i = 0; i < (n-1); i++ ) {
e_i[i] = e_save[i];
}
for( i = 0; i < ldz*n; i++ ) {
z_i[i] = z_save[i];
}
for( i = 0; i < (MAX(1,2*n-2)); i++ ) {
work_i[i] = work[i];
}
info_i = LAPACKE_csteqr( LAPACK_COL_MAJOR, compz_i, n_i, d_i, e_i, z_i,
ldz_i );
failed = compare_csteqr( d, d_i, e, e_i, z, z_i, info, info_i, compz, ldz,
n );
if( failed == 0 ) {
printf( "PASSED: column-major high-level interface to csteqr\n" );
} else {
printf( "FAILED: column-major high-level interface to csteqr\n" );
}
/* Initialize input data, call the row-major middle-level
* interface to LAPACK routine and check the results */
for( i = 0; i < n; i++ ) {
d_i[i] = d_save[i];
}
for( i = 0; i < (n-1); i++ ) {
e_i[i] = e_save[i];
}
for( i = 0; i < ldz*n; i++ ) {
z_i[i] = z_save[i];
}
for( i = 0; i < (MAX(1,2*n-2)); i++ ) {
work_i[i] = work[i];
}
if( LAPACKE_lsame( compz, 'i' ) || LAPACKE_lsame( compz, 'v' ) ) {
LAPACKE_cge_trans( LAPACK_COL_MAJOR, n, n, z_i, ldz, z_r, n+2 );
}
info_i = LAPACKE_csteqr_work( LAPACK_ROW_MAJOR, compz_i, n_i, d_i, e_i, z_r,
ldz_r, work_i );
if( LAPACKE_lsame( compz, 'i' ) || LAPACKE_lsame( compz, 'v' ) ) {
LAPACKE_cge_trans( LAPACK_ROW_MAJOR, n, n, z_r, n+2, z_i, ldz );
}
failed = compare_csteqr( d, d_i, e, e_i, z, z_i, info, info_i, compz, ldz,
n );
if( failed == 0 ) {
printf( "PASSED: row-major middle-level interface to csteqr\n" );
} else {
printf( "FAILED: row-major middle-level interface to csteqr\n" );
}
/* Initialize input data, call the row-major high-level
* interface to LAPACK routine and check the results */
for( i = 0; i < n; i++ ) {
d_i[i] = d_save[i];
}
for( i = 0; i < (n-1); i++ ) {
e_i[i] = e_save[i];
}
for( i = 0; i < ldz*n; i++ ) {
z_i[i] = z_save[i];
}
for( i = 0; i < (MAX(1,2*n-2)); i++ ) {
work_i[i] = work[i];
}
/* Init row_major arrays */
if( LAPACKE_lsame( compz, 'i' ) || LAPACKE_lsame( compz, 'v' ) ) {
LAPACKE_cge_trans( LAPACK_COL_MAJOR, n, n, z_i, ldz, z_r, n+2 );
}
info_i = LAPACKE_csteqr( LAPACK_ROW_MAJOR, compz_i, n_i, d_i, e_i, z_r,
ldz_r );
if( LAPACKE_lsame( compz, 'i' ) || LAPACKE_lsame( compz, 'v' ) ) {
LAPACKE_cge_trans( LAPACK_ROW_MAJOR, n, n, z_r, n+2, z_i, ldz );
}
failed = compare_csteqr( d, d_i, e, e_i, z, z_i, info, info_i, compz, ldz,
n );
if( failed == 0 ) {
printf( "PASSED: row-major high-level interface to csteqr\n" );
} else {
printf( "FAILED: row-major high-level interface to csteqr\n" );
}
/* Release memory */
if( d != NULL ) {
LAPACKE_free( d );
}
if( d_i != NULL ) {
LAPACKE_free( d_i );
}
if( d_save != NULL ) {
LAPACKE_free( d_save );
}
if( e != NULL ) {
LAPACKE_free( e );
}
if( e_i != NULL ) {
LAPACKE_free( e_i );
}
if( e_save != NULL ) {
LAPACKE_free( e_save );
}
if( z != NULL ) {
LAPACKE_free( z );
}
if( z_i != NULL ) {
LAPACKE_free( z_i );
}
if( z_r != NULL ) {
LAPACKE_free( z_r );
}
if( z_save != NULL ) {
LAPACKE_free( z_save );
}
if( work != NULL ) {
LAPACKE_free( work );
}
if( work_i != NULL ) {
LAPACKE_free( work_i );
}
return 0;
}
int main(void)
{
/* Local scalars */
char job, job_i;
char compz, compz_i;
lapack_int n, n_i;
lapack_int ilo, ilo_i;
lapack_int ihi, ihi_i;
lapack_int ldh, ldh_i;
lapack_int ldh_r;
lapack_int ldz, ldz_i;
lapack_int ldz_r;
lapack_int lwork, lwork_i;
lapack_int info, info_i;
lapack_int i;
int failed;
/* Local arrays */
lapack_complex_float *h = NULL, *h_i = NULL;
lapack_complex_float *w = NULL, *w_i = NULL;
lapack_complex_float *z = NULL, *z_i = NULL;
lapack_complex_float *work = NULL, *work_i = NULL;
lapack_complex_float *h_save = NULL;
lapack_complex_float *w_save = NULL;
lapack_complex_float *z_save = NULL;
lapack_complex_float *h_r = NULL;
lapack_complex_float *z_r = NULL;
/* Iniitialize the scalar parameters */
init_scalars_chseqr( &job, &compz, &n, &ilo, &ihi, &ldh, &ldz, &lwork );
ldh_r = n+2;
ldz_r = n+2;
job_i = job;
compz_i = compz;
n_i = n;
ilo_i = ilo;
ihi_i = ihi;
ldh_i = ldh;
ldz_i = ldz;
lwork_i = lwork;
/* Allocate memory for the LAPACK routine arrays */
h = (lapack_complex_float *)
LAPACKE_malloc( ldh*n * sizeof(lapack_complex_float) );
w = (lapack_complex_float *)
LAPACKE_malloc( n * sizeof(lapack_complex_float) );
z = (lapack_complex_float *)
LAPACKE_malloc( ldz*n * sizeof(lapack_complex_float) );
work = (lapack_complex_float *)
LAPACKE_malloc( lwork * sizeof(lapack_complex_float) );
/* Allocate memory for the C interface function arrays */
h_i = (lapack_complex_float *)
LAPACKE_malloc( ldh*n * sizeof(lapack_complex_float) );
w_i = (lapack_complex_float *)
LAPACKE_malloc( n * sizeof(lapack_complex_float) );
z_i = (lapack_complex_float *)
LAPACKE_malloc( ldz*n * sizeof(lapack_complex_float) );
work_i = (lapack_complex_float *)
LAPACKE_malloc( lwork * sizeof(lapack_complex_float) );
/* Allocate memory for the backup arrays */
h_save = (lapack_complex_float *)
LAPACKE_malloc( ldh*n * sizeof(lapack_complex_float) );
w_save = (lapack_complex_float *)
LAPACKE_malloc( n * sizeof(lapack_complex_float) );
z_save = (lapack_complex_float *)
LAPACKE_malloc( ldz*n * sizeof(lapack_complex_float) );
/* Allocate memory for the row-major arrays */
h_r = (lapack_complex_float *)
LAPACKE_malloc( n*(n+2) * sizeof(lapack_complex_float) );
z_r = (lapack_complex_float *)
LAPACKE_malloc( n*(n+2) * sizeof(lapack_complex_float) );
/* Initialize input arrays */
init_h( ldh*n, h );
init_w( n, w );
init_z( ldz*n, z );
init_work( lwork, work );
/* Backup the ouptut arrays */
for( i = 0; i < ldh*n; i++ ) {
h_save[i] = h[i];
}
for( i = 0; i < n; i++ ) {
w_save[i] = w[i];
}
for( i = 0; i < ldz*n; i++ ) {
z_save[i] = z[i];
}
/* Call the LAPACK routine */
chseqr_( &job, &compz, &n, &ilo, &ihi, h, &ldh, w, z, &ldz, work, &lwork,
&info );
/* Initialize input data, call the column-major middle-level
* interface to LAPACK routine and check the results */
for( i = 0; i < ldh*n; i++ ) {
h_i[i] = h_save[i];
}
for( i = 0; i < n; i++ ) {
w_i[i] = w_save[i];
}
for( i = 0; i < ldz*n; i++ ) {
z_i[i] = z_save[i];
}
for( i = 0; i < lwork; i++ ) {
work_i[i] = work[i];
}
info_i = LAPACKE_chseqr_work( LAPACK_COL_MAJOR, job_i, compz_i, n_i, ilo_i,
ihi_i, h_i, ldh_i, w_i, z_i, ldz_i, work_i,
lwork_i );
failed = compare_chseqr( h, h_i, w, w_i, z, z_i, info, info_i, compz, ldh,
ldz, n );
if( failed == 0 ) {
printf( "PASSED: column-major middle-level interface to chseqr\n" );
} else {
printf( "FAILED: column-major middle-level interface to chseqr\n" );
}
/* Initialize input data, call the column-major high-level
* interface to LAPACK routine and check the results */
for( i = 0; i < ldh*n; i++ ) {
h_i[i] = h_save[i];
}
for( i = 0; i < n; i++ ) {
w_i[i] = w_save[i];
}
for( i = 0; i < ldz*n; i++ ) {
z_i[i] = z_save[i];
}
for( i = 0; i < lwork; i++ ) {
work_i[i] = work[i];
}
info_i = LAPACKE_chseqr( LAPACK_COL_MAJOR, job_i, compz_i, n_i, ilo_i,
ihi_i, h_i, ldh_i, w_i, z_i, ldz_i );
failed = compare_chseqr( h, h_i, w, w_i, z, z_i, info, info_i, compz, ldh,
ldz, n );
if( failed == 0 ) {
printf( "PASSED: column-major high-level interface to chseqr\n" );
} else {
printf( "FAILED: column-major high-level interface to chseqr\n" );
}
/* Initialize input data, call the row-major middle-level
* interface to LAPACK routine and check the results */
for( i = 0; i < ldh*n; i++ ) {
h_i[i] = h_save[i];
}
for( i = 0; i < n; i++ ) {
w_i[i] = w_save[i];
}
for( i = 0; i < ldz*n; i++ ) {
z_i[i] = z_save[i];
}
for( i = 0; i < lwork; i++ ) {
work_i[i] = work[i];
}
LAPACKE_cge_trans( LAPACK_COL_MAJOR, n, n, h_i, ldh, h_r, n+2 );
if( LAPACKE_lsame( compz, 'i' ) || LAPACKE_lsame( compz, 'v' ) ) {
LAPACKE_cge_trans( LAPACK_COL_MAJOR, n, n, z_i, ldz, z_r, n+2 );
}
info_i = LAPACKE_chseqr_work( LAPACK_ROW_MAJOR, job_i, compz_i, n_i, ilo_i,
ihi_i, h_r, ldh_r, w_i, z_r, ldz_r, work_i,
lwork_i );
LAPACKE_cge_trans( LAPACK_ROW_MAJOR, n, n, h_r, n+2, h_i, ldh );
if( LAPACKE_lsame( compz, 'i' ) || LAPACKE_lsame( compz, 'v' ) ) {
LAPACKE_cge_trans( LAPACK_ROW_MAJOR, n, n, z_r, n+2, z_i, ldz );
}
failed = compare_chseqr( h, h_i, w, w_i, z, z_i, info, info_i, compz, ldh,
ldz, n );
if( failed == 0 ) {
printf( "PASSED: row-major middle-level interface to chseqr\n" );
} else {
printf( "FAILED: row-major middle-level interface to chseqr\n" );
}
/* Initialize input data, call the row-major high-level
* interface to LAPACK routine and check the results */
for( i = 0; i < ldh*n; i++ ) {
h_i[i] = h_save[i];
}
for( i = 0; i < n; i++ ) {
w_i[i] = w_save[i];
}
for( i = 0; i < ldz*n; i++ ) {
z_i[i] = z_save[i];
}
for( i = 0; i < lwork; i++ ) {
work_i[i] = work[i];
}
/* Init row_major arrays */
LAPACKE_cge_trans( LAPACK_COL_MAJOR, n, n, h_i, ldh, h_r, n+2 );
if( LAPACKE_lsame( compz, 'i' ) || LAPACKE_lsame( compz, 'v' ) ) {
LAPACKE_cge_trans( LAPACK_COL_MAJOR, n, n, z_i, ldz, z_r, n+2 );
}
info_i = LAPACKE_chseqr( LAPACK_ROW_MAJOR, job_i, compz_i, n_i, ilo_i,
ihi_i, h_r, ldh_r, w_i, z_r, ldz_r );
LAPACKE_cge_trans( LAPACK_ROW_MAJOR, n, n, h_r, n+2, h_i, ldh );
if( LAPACKE_lsame( compz, 'i' ) || LAPACKE_lsame( compz, 'v' ) ) {
LAPACKE_cge_trans( LAPACK_ROW_MAJOR, n, n, z_r, n+2, z_i, ldz );
}
failed = compare_chseqr( h, h_i, w, w_i, z, z_i, info, info_i, compz, ldh,
ldz, n );
if( failed == 0 ) {
printf( "PASSED: row-major high-level interface to chseqr\n" );
} else {
printf( "FAILED: row-major high-level interface to chseqr\n" );
}
/* Release memory */
if( h != NULL ) {
LAPACKE_free( h );
}
if( h_i != NULL ) {
LAPACKE_free( h_i );
}
if( h_r != NULL ) {
LAPACKE_free( h_r );
}
if( h_save != NULL ) {
LAPACKE_free( h_save );
}
if( w != NULL ) {
LAPACKE_free( w );
}
if( w_i != NULL ) {
LAPACKE_free( w_i );
}
if( w_save != NULL ) {
LAPACKE_free( w_save );
}
if( z != NULL ) {
LAPACKE_free( z );
}
if( z_i != NULL ) {
LAPACKE_free( z_i );
}
if( z_r != NULL ) {
LAPACKE_free( z_r );
}
if( z_save != NULL ) {
LAPACKE_free( z_save );
}
if( work != NULL ) {
LAPACKE_free( work );
}
if( work_i != NULL ) {
LAPACKE_free( work_i );
}
return 0;
}
void TopicSearch::run_hybrid_random_walk_simulated_annealing(
vec iter_temperature,
double random_walk_prob,
double percent_random_walk){
size_t accepted_Z_instances;
bool valid_burn_in_period;
Timer timer = Timer();
timer.restart_time();
init_z();
mat multinomial_prob = init_topical_Multinomial_probabilities();
if (this->verbose_ >= 1)
cout << "Multinomial probabilities: " << endl << multinomial_prob;
if (this->burn_in_period_ > 0 && this->burn_in_period_ < this->max_iterations_){
valid_burn_in_period = true;
accepted_Z_instances = ceil((this->max_iterations_ - this->burn_in_period_) / this->spacing);
}
else {
valid_burn_in_period = false;
accepted_Z_instances = ceil(this->max_iterations_ / this->spacing);
}
for (size_t d = 0; d < this->num_documents_; d++){ // START For each document
size_t num_words = this->document_lengths_[d];
umat accepted_Z = zeros<umat>(num_words, accepted_Z_instances);
uvec proposed_Z = zeros<uvec>(num_words);
uvec current_Z = zeros<uvec>(num_words);
size_t acceptance_count = 0;
size_t count = 0;
size_t random_walk_count = ceil((percent_random_walk / 100) * num_words);
size_t num_random_walks = 0;
uvec sampled_z;
uvec sampled_z2;
vector <size_t> word_indices = this->document_word_indices_[d];
for (size_t n = 0; n < num_words; n++)
current_Z(n) = this->initial_z_(word_indices[n]);
long double ppZ = calc_ln_partition_probality(word_indices, current_Z);
for (size_t iter = 0; iter < this->max_iterations_; iter++){ // START TOPIC SEARCH
if (this->sample_uniform() <= random_walk_prob){ // do random walk from the previous state
num_random_walks++;
proposed_Z = current_Z;
for (size_t s = 0; s < random_walk_count; s++){
size_t idx = sample_uniform_int(num_words); // selects a word at random
while(1){
size_t topic = sample_uniform_int(this->num_topics_);
if (topic != current_Z(idx)){
proposed_Z(idx) = topic;
break;
}
}
}
}
else { // do sample from the topic specific Multinomial
for (size_t i = 0; i < num_words; i++)
proposed_Z(i) = sample_multinomial(multinomial_prob.col(current_Z(i)));
}
long double ppZ_prime = calc_ln_partition_probality(word_indices, proposed_Z);
long double tpZ_prime = calc_lnTP_hybrid_randomwalk(
num_words, proposed_Z, current_Z, multinomial_prob, random_walk_prob, random_walk_count);
long double tpZ = calc_lnTP_hybrid_randomwalk(
num_words, current_Z, proposed_Z, multinomial_prob, random_walk_prob, random_walk_count);
long double p_ratio = ppZ_prime - ppZ;
long double q_ratio = tpZ_prime - tpZ;
double acceptance_probability = min(1.0L, pow(exp(p_ratio + q_ratio), (1 / iter_temperature(iter)))); // MH acceptance probability
if (this->sample_uniform() <= acceptance_probability){
ppZ = ppZ_prime; // To avoid re-calculation
current_Z = proposed_Z;
acceptance_count++;
// if (this->verbose_ >= 1){
//
// cout << "doc " << d + 1 << " iter " << iter + 1;
// cout << " accepted";
// cout << " [a.p. = " << pow(exp(p_ratio + q_ratio), (1 / iter_temperature(iter)))
// // << " t.ratio = " << exp(q_ratio) << " p.ratio = " << exp(p_ratio)
// << " ln P(z') = " << ppZ_prime << " ln P(z) = " << ppZ
// << " ln T(z',z) = " << tpZ_prime << " ln T(z,z') = " << tpZ
// << " a.count = " << acceptance_count << " ]" << endl;
// }
}
if ((iter % this->spacing == 0)
&& (!valid_burn_in_period || (valid_burn_in_period
&& (this->burn_in_period_ < iter)))) {
accepted_Z.col(count) = current_Z;
count++;
}
} // END TOPIC SEARCH
// // Saves the results to the class variables
// sampled_z = find_mode(accepted_Z);
// sampled_z2 = accepted_Z.col(count - 1);
//// for (size_t n = 0; n < num_words; n++){
//// this->sampled_z_(word_indices[n]) = sampled_z(n);
//// this->beta_counts_(this->sampled_z_(word_indices[n]), this->word_ids_(word_indices[n])) += 1;
//// this->beta_counts_last_(sampled_z2(n), this->word_ids_(word_indices[n])) += 1;
//// }
//
// // Calculates theta counts
// this->theta_counts_.col(d) = calc_topic_counts(sampled_z, this->num_topics_);
// this->theta_counts_last_.col(d) = calc_topic_counts(sampled_z2, this->num_topics_);
//
// // Resets all used data structures
// current_Z.reset();
// proposed_Z.reset();
// accepted_Z.reset();
// sampled_z.reset();
// sampled_z2.reset();
// Saves the results to the class variable
sampled_z = find_mode(accepted_Z);
sampled_z2 = accepted_Z.col(count - 1);
for (size_t n = 0; n < num_words; n++){
this->sampled_z_(word_indices[n]) = sampled_z(n);
this->beta_counts_(this->sampled_z_(word_indices[n]), this->word_ids_(word_indices[n])) += 1;
this->beta_counts_last_(sampled_z2(n), this->word_ids_(word_indices[n])) += 1;
}
// Calculates theta counts
this->theta_counts_.col(d) = calc_topic_counts(sampled_z, this->num_topics_);
this->theta_counts_last_.col(d) = calc_topic_counts(sampled_z2, this->num_topics_);
// Resets all used data structures
current_Z.reset();
proposed_Z.reset();
accepted_Z.reset();
sampled_z.reset();
sampled_z2.reset();
if (this->verbose_ >= 1)
cout << "doc " << d + 1 << " accepted # " << acceptance_count << " random walks # " << num_random_walks << endl;
} // END For each document
if (this->verbose_ >= 1){
cout << endl << "Total execution time: " << timer.get_time() << "s" << endl;
cout << "Model perplexity: " << calc_corpus_perplexity() << endl; // using beta_counts_ and theta_counts_
cout << "Log partition probability: " << calc_ln_corpus_partition_probality() << endl;
}
}
void TopicSearch::run_hybrid_random_walk_simulated_annealing_uniform2(
vec iter_temperature,
double random_walk_prob,
double percent_random_walk){
size_t accepted_Z_instances;
bool valid_burn_in_period;
Timer timer = Timer();
timer.restart_time();
init_z();
if (this->burn_in_period_ > 0 && this->burn_in_period_ < this->max_iterations_){
valid_burn_in_period = true;
accepted_Z_instances = ceil((this->max_iterations_ - this->burn_in_period_) / this->spacing);
}
else {
valid_burn_in_period = false;
accepted_Z_instances = ceil(this->max_iterations_ / this->spacing);
}
for (size_t d = 0; d < this->num_documents_; d++){ // START For each document
size_t num_words = this->document_lengths_[d];
umat accepted_Z = zeros<umat>(num_words, accepted_Z_instances);
vec accepted_Z_pp = zeros<vec>(accepted_Z_instances);
uvec proposed_Z = zeros<uvec>(num_words);
uvec current_Z = zeros<uvec>(num_words);
size_t acceptance_count = 0;
size_t count = 0;
size_t num_random_walks = 0;
size_t num_random_walks2 = 0;
uvec sampled_z;
uvec sampled_z2;
vector <size_t> word_indices = this->document_word_indices_[d];
long double ppZ;
long double ppZ_prime;
long double tpZ_prime;
long double tpZ;
long double p_ratio;
long double q_ratio;
double acceptance_probability;
double multi_jump_prob = percent_random_walk / 100.0;
for (size_t n = 0; n < num_words; n++)
current_Z(n) = this->initial_z_(word_indices[n]);
ppZ = calc_ln_partition_probality(word_indices, current_Z);
for (size_t iter = 0; iter < this->max_iterations_; iter++){ // START TOPIC SEARCH
if (this->sample_uniform() <= random_walk_prob){ // do random walk from the previous state
num_random_walks++;
proposed_Z = current_Z;
for (size_t i = 0; i < num_words; i++){
if (this->sample_uniform() <= multi_jump_prob){ /// for each word with some probability multinomial jump
proposed_Z(i) = sample_multinomial(this->init_beta_sample_.col(this->word_ids_(word_indices[i])));
num_random_walks2++;
}
}
}
else { // do sample from a uniform
for (size_t i = 0; i < num_words; i++)
proposed_Z(i) = sample_uniform_int(this->num_topics_);
}
ppZ_prime = calc_ln_partition_probality(word_indices, proposed_Z);
p_ratio = ppZ_prime - ppZ;
tpZ_prime = calc_lnTP_hybrid_multi_randomwalk(num_words, word_indices, current_Z, random_walk_prob, multi_jump_prob);
tpZ = calc_lnTP_hybrid_multi_randomwalk(num_words, word_indices, proposed_Z, random_walk_prob, multi_jump_prob);
q_ratio = tpZ_prime - tpZ;
assert(iter_temperature(iter) > 0);
// acceptance_probability = min(1.0L, pow(exp(p_ratio), (1 / iter_temperature(iter)))); // MH acceptance probability
acceptance_probability = min(1.0L, pow(exp(p_ratio + q_ratio), (1.0 / iter_temperature(iter)))); // MH acceptance probability
if (this->sample_uniform() <= acceptance_probability){
current_Z = proposed_Z;
ppZ = ppZ_prime;
acceptance_count++;
// if (this->verbose_ >= 1){
// cout << "doc " << d + 1 << " iter " << iter + 1;
// cout << " accepted";
// cout << " [a.p. = " << pow(exp(p_ratio), (1 / iter_temperature(iter)))
// << " ln P(z') = " << ppZ_prime << " ln P(z) = " << ppZ
// << " a.count = " << acceptance_count << " ]" << endl;
// }
}
if (((iter + 1) % this->spacing == 0) && (!valid_burn_in_period || (valid_burn_in_period && (this->burn_in_period_ < iter)))) {
accepted_Z.col(count) = current_Z;
accepted_Z_pp(count) = ppZ_prime;
count++;
}
} // END TOPIC SEARCH
// Saves the results to the class variable
sampled_z = find_mode(accepted_Z);
sampled_z2 = accepted_Z.col(count - 1);
for (size_t n = 0; n < num_words; n++){
this->sampled_z_(word_indices[n]) = sampled_z(n);
this->beta_counts_(this->sampled_z_(word_indices[n]), this->word_ids_(word_indices[n])) += 1;
this->beta_counts_last_(sampled_z2(n), this->word_ids_(word_indices[n])) += 1;
}
// Calculates theta counts
this->theta_counts_.col(d) = calc_topic_counts(sampled_z, this->num_topics_);
this->theta_counts_last_.col(d) = calc_topic_counts(sampled_z2, this->num_topics_);
// Resets all used data structures
current_Z.reset();
proposed_Z.reset();
accepted_Z.reset();
sampled_z.reset();
sampled_z2.reset();
// if (this->verbose_ >= 1)
cout << "doc " << d + 1 << " accepted # " << acceptance_count << " random walks # " << num_random_walks << " actual random walks # " << num_random_walks2 << endl;
} // END For each document
// if (this->verbose_ >= 1){
cout << endl << "Total execution time: " << timer.get_time() << "s" << endl;
cout << "Model perplexity: " << calc_corpus_perplexity() << endl; // using beta_counts_ and theta_counts_
cout << "Log partition probability: " << calc_ln_corpus_partition_probality() << endl;
// }
}
virtual void init(const Opm::parameter::ParameterGroup& param,
const Grid& grid,
const Fluid& fluid,
typename Grid::Vector gravity,
State& simstate)
{
typedef typename Fluid::CompVec CompVec;
typedef typename Fluid::PhaseVec PhaseVec;
if (param.getDefault("heterogenous_initial_mix", false)) {
CompVec init_oil(0.0);
init_oil[Fluid::Oil] = 1.0;
CompVec init_water(0.0);
init_water[Fluid::Water] = 1.0;
simstate.cell_z_.resize(grid.numCells());
std::fill(simstate.cell_z_.begin(), simstate.cell_z_.begin() + simstate.cell_z_.size()/2, init_oil);
std::fill(simstate.cell_z_.begin() + simstate.cell_z_.size()/2, simstate.cell_z_.end(), init_water);
OPM_MESSAGE("******* Assuming zero capillary pressures *******");
PhaseVec init_p(100.0*Opm::unit::barsa);
simstate.cell_pressure_.resize(grid.numCells(), init_p);
// if (gravity.two_norm() != 0.0) {
// double ref_gravpot = grid.cellCentroid(0)*gravity;
// double rho = init_z*fluid_.surfaceDensities(); // Assuming incompressible, and constant initial z.
// for (int cell = 1; cell < grid.numCells(); ++cell) {
// double press = rho*(grid.cellCentroid(cell)*gravity - ref_gravpot) + simstate.cell_pressure_[0][0];
// simstate.cell_pressure_[cell] = PhaseVec(press);
// }
// }
} else if (param.getDefault("unstable_initial_mix", false)) {
CompVec init_oil(0.0);
init_oil[Fluid::Oil] = 1.0;
init_oil[Fluid::Gas] = 0.0;
CompVec init_water(0.0);
init_water[Fluid::Water] = 1.0;
CompVec init_gas(0.0);
init_gas[Fluid::Gas] = 150.0;
simstate.cell_z_.resize(grid.numCells());
std::fill(simstate.cell_z_.begin(),
simstate.cell_z_.begin() + simstate.cell_z_.size()/3,
init_water);
std::fill(simstate.cell_z_.begin() + simstate.cell_z_.size()/3,
simstate.cell_z_.begin() + 2*(simstate.cell_z_.size()/3),
init_oil);
std::fill(simstate.cell_z_.begin() + 2*(simstate.cell_z_.size()/3),
simstate.cell_z_.end(),
init_gas);
OPM_MESSAGE("******* Assuming zero capillary pressures *******");
PhaseVec init_p(100.0*Opm::unit::barsa);
simstate.cell_pressure_.resize(grid.numCells(), init_p);
if (gravity.two_norm() != 0.0) {
typename Fluid::FluidState state = fluid.computeState(simstate.cell_pressure_[0], simstate.cell_z_[0]);
simstate.cell_z_[0] *= 1.0/state.total_phase_volume_density_;
for (int cell = 1; cell < grid.numCells(); ++cell) {
double fluid_vol_dens;
int cnt =0;
do {
double rho = 0.5*((simstate.cell_z_[cell]+simstate.cell_z_[cell-1])*fluid.surfaceDensities());
double press = rho*((grid.cellCentroid(cell) - grid.cellCentroid(cell-1))*gravity) + simstate.cell_pressure_[cell-1][0];
simstate.cell_pressure_[cell] = PhaseVec(press);
state = fluid.computeState(simstate.cell_pressure_[cell], simstate.cell_z_[cell]);
fluid_vol_dens = state.total_phase_volume_density_;
simstate.cell_z_[cell] *= 1.0/fluid_vol_dens;
++cnt;
} while (std::fabs((fluid_vol_dens-1.0)) > 1.0e-8 && cnt < 10);
}
} else {
std::cout << "---- Exit - BlackoilSimulator.hpp: No gravity, no fun ... ----" << std::endl;
exit(-1);
}
} else if (param.getDefault("CO2-injection", false)) {
CompVec init_water(0.0);
// Initially water filled (use Oil-component for water in order
// to utilise blackoil mechanisms for brine-co2 interaction)
init_water[Fluid::Oil] = 1.0;
simstate.cell_z_.resize(grid.numCells());
std::fill(simstate.cell_z_.begin(),simstate.cell_z_.end(),init_water);
double datum_pressure_barsa = param.getDefault<double>("datum_pressure", 200.0);
double datum_pressure = Opm::unit::convert::from(datum_pressure_barsa, Opm::unit::barsa);
PhaseVec init_p(datum_pressure);
simstate.cell_pressure_.resize(grid.numCells(), init_p);
// Simple initial condition based on "incompressibility"-assumption
double zMin = grid.cellCentroid(0)[2];
for (int cell = 1; cell < grid.numCells(); ++cell) {
if (grid.cellCentroid(cell)[2] < zMin)
zMin = grid.cellCentroid(cell)[2];
}
typename Fluid::FluidState state = fluid.computeState(init_p, init_water);
simstate.cell_z_[0] *= 1.0/state.total_phase_volume_density_;
double density = (init_water*fluid.surfaceDensities())/state.total_phase_volume_density_;
for (int cell = 0; cell < grid.numCells(); ++cell) {
double pressure(datum_pressure + (grid.cellCentroid(cell)[2] - zMin)*gravity[2]*density);
simstate.cell_pressure_[cell] = PhaseVec(pressure);
state = fluid.computeState(simstate.cell_pressure_[cell], simstate.cell_z_[cell]);
simstate.cell_z_[cell] *= 1.0/state.total_phase_volume_density_;
}
} else {
CompVec init_z(0.0);
double initial_mixture_gas = param.getDefault("initial_mixture_gas", 0.0);
double initial_mixture_oil = param.getDefault("initial_mixture_oil", 1.0);
double initial_mixture_water = param.getDefault("initial_mixture_water", 0.0);
init_z[Fluid::Water] = initial_mixture_water;
init_z[Fluid::Gas] = initial_mixture_gas;
init_z[Fluid::Oil] = initial_mixture_oil;
simstate.cell_z_.resize(grid.numCells(), init_z);
OPM_MESSAGE("******* Assuming zero capillary pressures *******");
PhaseVec init_p(param.getDefault("initial_pressure", 100.0*Opm::unit::barsa));
simstate.cell_pressure_.resize(grid.numCells(), init_p);
if (gravity.two_norm() != 0.0) {
double ref_gravpot = grid.cellCentroid(0)*gravity;
double rho = init_z*fluid.surfaceDensities(); // Assuming incompressible, and constant initial z.
for (int cell = 1; cell < grid.numCells(); ++cell) {
double press = rho*(grid.cellCentroid(cell)*gravity - ref_gravpot) + simstate.cell_pressure_[0][0];
simstate.cell_pressure_[cell] = PhaseVec(press);
}
}
}
}
