-
Notifications
You must be signed in to change notification settings - Fork 1
Expand file tree
/
Copy pathfeatures.cpp
More file actions
141 lines (123 loc) · 4.87 KB
/
features.cpp
File metadata and controls
141 lines (123 loc) · 4.87 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
#include "PartialModel.hpp"
#include "aux.hpp"
#include "odds.hpp"
using namespace std;
namespace PoissonFactorization {
namespace Partial {
template <>
void Model<Variable::Feature, Kind::Gamma>::initialize_factor(size_t t) {
// initialize p of Φ
LOG(debug) << "Initializing P of Φ";
#pragma omp parallel for if (DO_PARALLEL)
for (size_t g = 0; g < dim1; ++g)
prior.p(g, t) = prob_to_neg_odds(sample_beta<Float>(
parameters.hyperparameters.phi_p_1, parameters.hyperparameters.phi_p_2,
EntropySource::rngs[omp_get_thread_num()]));
// initialize r of Φ
LOG(debug) << "Initializing R of Φ";
#pragma omp parallel for if (DO_PARALLEL)
for (size_t g = 0; g < dim1; ++g)
// NOTE: std::gamma_distribution takes a shape and scale parameter
prior.r(g, t) = std::gamma_distribution<Float>(
parameters.hyperparameters.phi_r_1,
1 / parameters.hyperparameters.phi_r_2)(
EntropySource::rngs[omp_get_thread_num()]);
// initialize Φ
LOG(debug) << "Initializing Φ";
#pragma omp parallel for if (DO_PARALLEL)
for (size_t g = 0; g < dim1; ++g)
// NOTE: std::gamma_distribution takes a shape and scale parameter
matrix(g, t)
= std::gamma_distribution<Float>(prior.r(g, t), 1 / prior.p(g, t))(
EntropySource::rngs[omp_get_thread_num()]);
}
template <>
void Model<Variable::Feature, Kind::Dirichlet>::initialize_factor(size_t t) {
auto x = sample_dirichlet<Float>(prior.alpha.begin_col(t),
prior.alpha.end_col(t),
EntropySource::rngs[omp_get_thread_num()]);
for (size_t g = 0; g < dim1; ++g)
matrix(g, t) = x[g];
}
template <>
void Model<Variable::Feature, Kind::Gamma>::initialize() {
LOG(debug) << "Initializing Φ from Gamma distribution";
#pragma omp parallel for if (DO_PARALLEL)
for (size_t g = 0; g < dim1; ++g) {
const size_t thread_num = omp_get_thread_num();
for (size_t t = 0; t < dim2; ++t)
// NOTE: gamma_distribution takes a shape and scale parameter
matrix(g, t) = std::gamma_distribution<Float>(
prior.r(g, t), 1 / prior.p(g, t))(EntropySource::rngs[thread_num]);
}
}
template <>
void Model<Variable::Feature, Kind::Dirichlet>::initialize() {
LOG(debug) << "Initializing Φ from Dirichlet distribution";
#pragma omp parallel for if (DO_PARALLEL)
for (size_t t = 0; t < dim2; ++t)
initialize_factor(t);
}
template <>
// TODO ensure no NaNs or infinities are generated
double Model<Variable::Feature, Kind::Gamma>::log_likelihood() const {
double l = 0;
for (size_t t = 0; t < dim2; ++t)
l += log_likelihood_factor(t);
return l;
}
template <>
// TODO ensure no NaNs or infinities are generated
double Model<Variable::Feature, Kind::Gamma>::log_likelihood_factor(
size_t t) const {
double l = 0;
#pragma omp parallel for reduction(+ : l) if (DO_PARALLEL)
for (size_t g = 0; g < dim1; ++g)
// NOTE: log_gamma takes a shape and scale parameter
l += log_gamma(matrix(g, t), prior.r(g, t), 1.0 / prior.p(g, t));
if (parameters.respect_phi_prior_likelihood)
#pragma omp parallel for reduction(+ : l) if (DO_PARALLEL)
for (size_t g = 0; g < dim1; ++g)
// NOTE: log_gamma takes a shape and scale parameter
l += log_gamma(prior.r(g, t), parameters.hyperparameters.phi_r_1,
1.0 / parameters.hyperparameters.phi_r_2);
if (parameters.respect_phi_prior_likelihood)
#pragma omp parallel for reduction(+ : l) if (DO_PARALLEL)
for (size_t g = 0; g < dim1; ++g)
// TODO FIXME this needs to use the generalized beta prime distribution
l += log_beta_neg_odds(prior.p(g, t), parameters.hyperparameters.phi_p_1,
parameters.hyperparameters.phi_p_2);
// l += log_generalized_beta_prime(prior.p(g, t),
// parameters.hyperparameters.phi_p_1,
// parameters.hyperparameters.phi_p_2,
// experiment.contributions_gene_type(g, t);
// expected); // experiment.expected_gene_type(...);
LOG(verbose) << "Feature log likelihood factor " << t << ": " << l;
return l;
}
template <>
// TODO ensure no NaNs or infinities are generated
// TODO check whether using OMP is actually faster here!
double Model<Variable::Feature, Kind::Dirichlet>::log_likelihood() const {
double l = 0;
for (size_t t = 0; t < dim2; ++t)
l += log_likelihood_factor(t);
return l;
}
template <>
// TODO ensure no NaNs or infinities are generated
// TODO check whether using OMP is actually faster here!
double Model<Variable::Feature, Kind::Dirichlet>::log_likelihood_factor(
size_t t) const {
std::vector<Float> p(dim1);
#pragma omp parallel for if (DO_PARALLEL)
for (size_t g = 0; g < dim1; ++g)
p[g] = matrix(g, t);
std::vector<Float> alpha(dim1);
#pragma omp parallel for if (DO_PARALLEL)
for (size_t g = 0; g < dim1; ++g)
alpha[g] = prior.alpha(g, t) + prior.alpha_prior;
return log_dirichlet(p, alpha);
}
}
}