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
|
# Copyright (C) 2003,2004,2005 Michael Creel <michael.creel@uab.es>
#
# This program 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.
#
# This program 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, see <http://www.gnu.org/licenses/>.
## usage: [obj_value, score] = mle_obj(theta, data, model, modelargs, nslaves)
##
## Returns the average log-likelihood for a specified model
## This is for internal use by mle_estimate
function [obj_value, score] = mle_obj(theta, data, model, modelargs, nslaves)
n = rows(data);
if nargin < 5 nslaves = 0; endif
if nslaves > 0
global NSLAVES PARALLEL NEWORLD NSLAVES TAG;
nn = floor(n/(NSLAVES + 1)); # number of obsns per slave
# The command that the slave nodes will execute
cmd=['contrib = mle_obj_nodes(theta, data, model, modelargs, nn); ',...
'MPI_Send(contrib,0,TAG,NEWORLD);'];
# send items to slaves
NumCmds_Send({"theta", "nn", "cmd"}, {theta, nn, cmd});
# evaluate last block on master while slaves are busy
obj_value = mle_obj_nodes(theta, data, model, modelargs, nn);
# collect slaves' results
contrib = 0.0; # must be initialized to use MPI_Recv
for i = 1:NSLAVES
MPI_Recv(contrib,i,TAG,NEWORLD);
obj_value = obj_value + contrib;
endfor
# compute the average
obj_value = - obj_value / n;
score = "na"; # fix this later to allow analytic score in parallel
else # serial version
[contribs, score] = feval(model, theta, data, modelargs);
obj_value = - mean(contribs);
if isnumeric(score) score = - mean(score)'; endif # model passes "na" when score not available
endif
# let's bullet-proof this in case the model goes nuts
if (((abs(obj_value) == Inf)) || (isnan(obj_value)))
obj_value = realmax/10;
endif
endfunction
|
