-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathopt-alpha.cpp
executable file
·105 lines (89 loc) · 3.37 KB
/
opt-alpha.cpp
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
// (C) Copyright 2009, Jun Zhu (junzhu [at] cs [dot] cmu [dot] edu)
// This file is part of MedLDA.
// MedLDA is free software; you can redistribute it and/or modify it under
// the terms of the GNU General Public License as published by the Free
// Software Foundation; either version 2 of the License, or (at your
// option) any later version.
// MedLDA is distributed in the hope that it will be useful, but WITHOUT
// ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
// FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
// for more details.
// You should have received a copy of the GNU General Public License
// along with this program; if not, write to the Free Software
// Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
// USA
#include "opt-alpha.h"
/*
* case 1: different topics share the same prior
* Newton's method with the objective function and its derivatives
*/
double alhood(double a, double ss, int D, int K)
{ return(D * (lgamma(K * a) - K * lgamma(a)) + (a - 1) * ss); }
double d_alhood(double a, double ss, int D, int K)
{ return(D * (K * digamma(K * a) - K * digamma(a)) + ss); }
double d2_alhood(double a, int D, int K)
{ return(D * (K * K * trigamma(K * a) - K * trigamma(a))); }
double opt_alpha(double ss, int D, int K)
{
double a, log_a, init_a = 100;
double f, df, d2f;
int iter = 0;
log_a = log(init_a);
do
{
iter++;
a = exp(log_a);
if (/*isnan(a)*/a>1e300 || a<1e-300)
{
init_a = init_a * 10;
printf("warning : alpha is nan; new init = %5.5f\n", init_a);
a = init_a;
log_a = log(a);
}
f = alhood(a, ss, D, K);
df = d_alhood(a, ss, D, K);
d2f = d2_alhood(a, D, K);
log_a = log_a - df/(d2f * a + df);
printf("alpha maximization : %5.5f %5.5f\n", f, df);
}
while ((fabs(df) > NEWTON_THRESH) && (iter < MAX_ALPHA_ITER));
return(exp(log_a));
}
/*
* case 2: different topics have different priors
* Newton's method with the objective function and its derivatives
*/
double alhood(double a, double alpha_sum, double ss, int D, int K)
{ return(D * (lgamma(alpha_sum) - lgamma(a)) + (a - 1) * ss); }
double d_alhood(double a, double alpha_sum, double ss, int D, int K)
{ return(D * (digamma(alpha_sum) - digamma(a)) + ss); }
double d2_alhood(double a, double alpha_sum, int D, int K)
{ return(D * (trigamma(alpha_sum) - trigamma(a))); }
double opt_alpha(double ss, double &alpha_sum, int D, int K)
{
double a, log_a, init_a = 100, old_a = 0;
double f, df, d2f;
int iter = 0;
log_a = log(init_a);
do
{
iter++;
a = exp(log_a);
alpha_sum += a - old_a;
old_a = a;
if (/*isnan(a)*/a>1e300 || a<1e-300)
{
init_a = init_a * 10;
printf("warning : alpha is nan; new init = %5.5f\n", init_a);
a = init_a;
log_a = log(a);
}
f = alhood(a, alpha_sum, ss, D, K);
df = d_alhood(a, alpha_sum, ss, D, K);
d2f = d2_alhood(a, alpha_sum, D, K);
log_a = log_a - df/(d2f * a + df);
//printf("alpha maximization : %5.5f %5.5f\n", f, df);
}
while ((fabs(df) > NEWTON_THRESH) && (iter < MAX_ALPHA_ITER));
return(exp(log_a));
}