-
Notifications
You must be signed in to change notification settings - Fork 1
/
parOES.m
307 lines (263 loc) · 13 KB
/
parOES.m
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
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%% %%%%%%%%%%%
%%%%%%%%%%% parallel Matlab simulation with Open Eye Simulator %%%%%%%%%%%
%%%%%%%%%%% Author: Lukas Klimmasch %%%%%%%%%%%
%%%%%%%%%%% %%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%% Usage: This scprit creates a pool of parallel workers and distributes %%%%%%%%%%%
%%%%%%%%%%% a number of jobs with different parameter combinations provided %%%%%%%%%%%
%%%%%%%%%%% in the first part of the file. %%%%%%%%%%%
%%%%%%%%%%% It is possible to start a series of experiments with all %%%%%%%%%%%
%%%%%%%%%%% combinations of two variables that can take different ranges. %%%%%%%%%%%
%%%%%%%%%%% To faciliate readability of resulting file names, one can %%%%%%%%%%%
%%%%%%%%%%% specify descriptors of the parameters that will be used to %%%%%%%%%%%
%%%%%%%%%%% compose an according file name. %%%%%%%%%%%
%%%%%%%%%%% A general parameter section provides the opportunity to choose %%%%%%%%%%%
%%%%%%%%%%% the number of iterations and the seed for each simulation run. %%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%&
function parOES(nWorkers)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% parameters that should be explored %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% varNames = {'gamma', 'interval'};
% var1 = {0.1, 0.3, 0.9};
% var2 = {10, 50, 100};
% varDescr = {'cDiscout', 'interval'};
% numberFormatVar1 = '%1.1f';
% numberFormatVar2 = '%d';
% experimentDirName = 'Discount Factor vs Interval' % no ';' intended.
% varNames = {'gamma', 'metCostRange'};
% var1 = {0.1, 0.3, 0.6, 0.9};
% var2 = {[0], [0.01], [0.025], [0.05]};
% varDescr = {'gamma', 'metCost'};
% numberFormatVar1 = '%1.1f';
% numberFormatVar2 = '[%1.3f]';
% experimentDirName = 'GammaVsMetCostsBias0.02'
% varNames = {'gamma', 'metCostRange'};
% var1 = {0.01, 0.07, 0.1, 0.2};
% var2 = {[0], [0.0017], [0.0033], [0.005]};
% varDescr = {'gamma', 'metCost'};
% numberFormatVar1 = '%1.2f';
% numberFormatVar2 = '[%1.4f]';
% experimentDirName = 'GammaVsMetCostsBiasFineGrain'
% varNames = {'actorLRRange', 'varianceRange'};
% var1 = {[1,1], [0.5,0.5], [1, 0], [0.5, 0]};
% var2 = {[5e-5,5e-5], [5e-5,0], [1e-5,1e-5], [1e-5,0]};
% varDescr = {'actorLRRange', 'varianceRange'};
% numberFormatVar1 = '[%1.1f-%1.1f]';
% numberFormatVar2 = '[%1.0e-%1.0e]';
% experimentDirName = 'actorLR_vs_varRange_norm_feat_mio' % no ';' intended.
% varNames = {'regularizer', 'actorLRRange'};
% var1 = {1e-4, 1e-5, 1e-6}; % {1e-2, 1e-3, 1e-4};
% var2 = {[1,1], [0.5,0.5], [1, 0], [0.5, 0]};
% varDescr = {'regul', 'actorLR'};
% numberFormatVar1 = '%1.0e';
% numberFormatVar2 = '[%1.2f-%1.2f]';
% experimentDirName = 'actorLR_vs_regul_norm_feat' % no ';' intended.
% varNames = {'criticLRRange', 'actorLRRange'};
% var1 = {[1, 1], [1, 0], [0.75, 0.75], [0.75, 0], [0.5, 0.5], [0.5, 0], [0.25, 0.25], [0.25, 0]};
% var2 = {[1, 1], [1, 0], [0.75, 0.75], [0.75, 0], [0.5, 0.5], [0.5, 0], [0.25, 0.25], [0.25, 0]};
% var1Descr = {'[1,1]', '[1,0]', '[0.75,0.75]', '[0.75,0]', '[0.5,0.5]', '[0.5,0]', '[0.25,0.25]', '[0.25,0]'};
% var2Descr = {'[1,1]', '[1,0]', '[0.75,0.75]', '[0.75,0]', '[0.5,0.5]', '[0.5,0]', '[0.25,0.25]', '[0.25,0]'};
% varDescr = {'CriticLR', 'ActorLR'};
% numberFormatVar1 = '[%1.2f-%1.2f]';
% numberFormatVar2 = '[%1.2f-%1.2f]';
% experimentDirName = 'CriticLR vs ActorLR';
% varNames = {'criticLRRange', 'varianceRange'};
% var1 = {[1, 0], [0.75, 0.75]};
% var2 = {[1e-4, 0], [5e-5, 0], [1e-5, 0], [5e-6, 0], [1e-4, 5e-6], [5e-5, 5e-6], [1e-5, 5e-6], [5e-6, 5e-6], [5e-5, 5e-5], [1e-5, 1e-5], [5e-6, 5e-6]};
% varDescr = {'critic', 'varRange'};
% numberFormatVar1 = '[%1.2f-%1.2f]';
% numberFormatVar2 = '[%1.0e-%1.0e]';
% experimentDirName = 'varDec_new'
% varNames = {'actorLRRange', 'varianceRange'};
% var1 = {[1, 1], [0.5, 0.5], [1, 0], [0.5, 0]};
% var2 = {[1e-4, 1e-4], [1e-5, 1e-5], [1e-4, 1e-6], [1e-5, 1e-6], [1e-4, 0], [1e-5, 0]};
% varDescr = {'actor', 'varRange'};
% numberFormatVar1 = '[%1.1f-%1.1f]';
% numberFormatVar2 = '[%1.0e-%1.0e]';
% experimentDirName = 'steplength_actorVsVariance_reg1e-5'
% varNames = {'actorLRRange', 'varianceRange'};
% var1 = {[1, 1], [0.5, 0.5], [1, 0], [0.5, 0]};
% % var2 = {[5e-5, 5e-5], [5e-5, 0],[1e-5, 1e-5], [1e-5, 0], [4e-5, 4e-5], [4e-5, 0],[3e-5, 3e-5], [3e-5, 0], [2e-5, 2e-5], [2e-5, 0]};
% var2 = {[4e-5, 4e-5], [4e-5, 0],[3e-5, 3e-5], [3e-5, 0], [2e-5, 2e-5], [2e-5, 0], [0.5e-5, 0.5e-5], [0.5e-5, 0]};
% varDescr = {'actor', 'varRange'};
% numberFormatVar1 = '[%1.1f-%1.1f]';
% numberFormatVar2 = '[%1.0e-%1.0e]';
%
% experimentDirName = 'steplength_actorVsVariance_reg1e-5_1mio'
% varNames = {'criticLRRange', 'gamma'};
% var1 = {[1, 1], [0.5, 0.5], [1, 0], [0.5, 0]};
% var2 = {0.1, 0.3, 0.6, 0.9};
% varDescr = {'actor', 'varRange'};
% numberFormatVar1 = '[%1.1f-%1.1f]';
% numberFormatVar2 = '[%1.1f]';
% experimentDirName = 'wobbling_gammaVsCriticLR_1mio'
% varNames = {'gamma', 'metCostRange'};
% var1 = {0.1, 0.3, 0.6, 0.9};
% % var2 = {[0], [0.01], [0.025], [0.05]};
% var2 = {[0], [0.01], [0.05], [0.1]};
% varDescr = {'gamma', 'metCost'};
% numberFormatVar1 = '%1.1f';
% numberFormatVar2 = '[%1.2f]';
% experimentDirName = 'GammaVsMetCosts_0,5mio'
% experimentDirName = 'KillTheBugs'
% varNames = {'criticLRRange', 'metCostRange'};
% var1 = {[1, 1], [1, 0], [0.5, 0.5], [0.5, 0]};
% % var2 = {[0], [0.01], [0.025], [0.05]};
% var2 = {[0.025, 0.025], [0.025, 0], [0.05, 0.05], [0.05, 0], [0, 0]};
% varDescr = {'critLR', 'metCost'};
% numberFormatVar1 = '[%1.1f-%1.1f]';
% numberFormatVar2 = '[%1.3f-%1.3f]';
% % 12 runs: 1992 mb per cpu on autumnchat
% experimentDirName = 'CritLRVsMetCosts_1mio'
% var1Descr = {'[1,1]', '[1,0]', '[0.75,0.75]', '[0.75,0]', '[0.5,0.5]', '[0.5,0]', '[0.25,0.25]', '[0.25,0]'};
% var2Descr = {'[1,1]', '[1,0]', '[0.75,0.75]', '[0.75,0]', '[0.5,0.5]', '[0.5,0]', '[0.25,0.25]', '[0.25,0]'};
% varDescr = {'CriticLR', 'ActorLR'};
% numberFormatVar1 = '[%1.2f-%1.2f]';
% numberFormatVar2 = '[%1.2f-%1.2f]';
% experimentDirName = 'CriticLR vs ActorLR';
% varNames = {'regularizer', 'desiredStdZT'};
% var1 = {5e-5, 1e-5, 5e-6, 1e-6};
% var2 = {0.005, 0.0075, 0.01, 0.015, 0.02, 0.025};
% var1Descr = {'5e-5', '1e-5', '5e-6', '1e-6'};
% var2Descr = {'0.005', '0.0075', '0.01', '0.015', '0.02', '0.025'};
% varDescr = {'regularizer', 'desiredStdZT'};
% experimentDirName = 'Regularizer_vs_desiredStdZT_short';
% varNames = {'lambdaMuscleFB', 'desiredStdZT'};
% var1 = {0, 0.5, 1, 2};
% var2 = {0.005, 0.0075, 0.01, 0.015, 0.02};
% var1Descr = {'0', '0.5', '1', '2'};
% var2Descr = {'0.005', '0.0075', '0.01', '0.015', '0.02'};
% varDescr = {'lambdaMuscleFB', 'desiredStdZT'};
% experimentDirName = 'lambdaMuscleFB_vs_desiredStdZT_seed2';
% varNames = {'gamma', 'metCostRange'};
% var1 = {0.01, 0.05, 0.1};
% % var2 = {[0], [0.01], [0.03], [0.05]};
% var2 = {[0], [0.05], [0.075], [0.1]};
% varDescr = {'gamma', 'metCost'};
% numberFormatVar1 = '%1.2f';
% numberFormatVar2 = '[%1.3f]';
% experimentDirName = 'GammaVsMetCosts_FineGrainNoBias'
% varNames = {'filterLeft', 'filterLeftProb'};
% % var1 = {8, 9, 10, 11, 12, 13};
% var1 = {13, 12, 11, 10, 9, 8};
% % var2 = {0.1, 0.25, 0.5, 0.75, 0.9, 1};
% var2 = {1};
% varDescr = {'filtL', 'filtLProb'};
% numberFormatVar1 = '%d';
% numberFormatVar2 = '%d';
% experimentDirName = 'uniformBlurrInput'
varNames = {'filterLeft', 'filterLeftProb'};
% var1 = {26, 27, 28};
var1 = {34};% {45, 46};%, 30, 31, 32, 33, 34, 35, 36, 37};
% var1 = fliplr(var1);
% var2 = {0.1, 0.25, 0.5, 0.75, 0.9, 1};
var2 = {1};
varDescr = {'filtBoth', 'prob'};
numberFormatVar1 = '%d';
numberFormatVar2 = '%d';
% experimentDirName = 'edgeDeprivNew'
experimentDirName = 'explFilterSizes'
globalParams = {};
globalName = 'fsize6std_longer_'
% varNames = {'filterRight', 'filterRightProb'};
% % var1 = {8, 9, 10, 11, 12, 13};
% var1 = {2};
% var2 = {0, 1};
% varDescr = {'MDfilterR', 'filtProb'};
% numberFormatVar1 = '%d';
% numberFormatVar2 = '%d';
% experimentDirName = 'disp+verg_influence'
%
% globalParams = {'objDistMin', 3, 'objDistMax', 3, 'fixDistMin', 3, 'fixDistMax', 3, ...
% 'initMethod', 1, 'actorLRRange', [0], 'varianceRange', [0], 'weight_range', [0,0,0]};
% globalName = 'OD3_initMet1_ALR0_Var0_weights0_'
% varNames = {'initMethod', 'lapSig'};
% % var1 = {8, 9, 10, 11, 12, 13};
% var1 = {4};
% % var2 = {0, 0.22, 0.5, 1, 2};
% var2 = {200};%, 200};
% varDescr = {'initMethod', 'lapSigma'};
% numberFormatVar1 = '%d';
% numberFormatVar2 = '%0.2f';
% experimentDirName = 'laplacianPolicy'
%
% globalParams = {'actorLRRange', [0], 'varianceRange', [0], 'weight_range', [0,0,0], ...
% 'nBasis', [400], 'nBasisUsed', [10], 'basisSize', [128], ...
% 'sc_eta', [0.2], 'temperature', [0.01], 'dsRatio', [1], ...
% 'pxFieldOfViewOrig', [40], 'pxFieldOfView', [40], 'stride', [4]};
% globalName = 'noLearn_fineScOnly_'
% varNames = {'initMethod', 'lapSig'};
% var1 = {4};%, 4};
% var2 = {0};
% varDescr = {'initMethod', 'lapSigma'};
% numberFormatVar1 = '%d';
% numberFormatVar2 = '%d';
% experimentDirName = 'vergenceInfluence'
% globalParams = {'actorLRRange', [0], 'varianceRange', [0], 'weight_range', [0,0,0], 'filterLeft', 29};
% globalName = 'noLearning_'
varNames = {'hiddenDim', 'actorLRRange'};
% var1 = {26, 27, 28};
var1 = {1, 2};% {45, 46};%, 30, 31, 32, 33, 34, 35, 36, 37};
% var1 = fliplr(var1);
% var2 = {0.1, 0.25, 0.5, 0.75, 0.9, 1};
var2 = {[0.5,0]};
varDescr = {'hdim', 'actorLR'};
numberFormatVar1 = '%d';
numberFormatVar2 = '[%1.2f-%1.2f]';
experimentDirName = 'stability'
globalParams = {};
globalName = 'actor_hidden_dim_'
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% general parameter section %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
nIters = 500000; % number of iterations
rSeeds = [1] % random seeds
% rSeeds = [2];
% rSeeds = [1, 5] % random seeds
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%% staring here, the rest is done automatically and should - in gerneral - not be altered %%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
nParams = length(var1) * length(var2);
paramValues = cell(nParams, 2); % column vector, each row contains a set of params for one run
iter = 1;
for i = 1 : length(var1)
for j = 1 : length(var2)
paramValues(iter, :) = {cell2mat(var1(i)), cell2mat(var2(j))};
% paramStrings(iter, :) = {cell2mat(var1Descr(i)), cell2mat(var2Descr(j))}; % temporary solution for not renaming all folders ...
paramStrings(iter, :) = {num2str(var1{i}, numberFormatVar1), num2str(var2{j}, numberFormatVar2)};
iter = iter + 1;
end
end
%% if no other descriptors are given, take original values
if isempty(varDescr)
varDescr = varNames;
end
myCluster = parcluster('local');
myCluster.NumWorkers = nWorkers;
% saveProfile(myCluster);
% create a parallel worker pool if there is none
if (isempty(gcp('nocreate')))
mPool = parpool(nWorkers);
end
% textureFiles = {'Textures_mcgillManMade40.mat', 'Textures_mcgillManMade100.mat'};
% textureFiles = {'mcGillTest2.mat', 'mcGillTest1.mat'}; % test files containing less images
% simulator = prepareSimulator(textureFiles); % at first, we try to use a shared simulator for all threads
%% main loop
if length(rSeeds) == 1
parfor ind = 1 : nParams
% for ind = 1 : nParams
% sprintf('%s=[%4.2f,%4.2f], %s=[%4.2f,%4.2f]', varNames{1}, paramValues{ind, 1}(1), paramValues{ind, 1}(2), varNames{2}, paramValues{ind, 2}(1), paramValues{ind, 2}(2))
sprintf('%s_%s_%s_%s', varDescr{1}, paramStrings{ind, 1}, varDescr{2}, paramStrings{ind, 2})
OES2Muscles(nIters, rSeeds(1), 1, ...
{varNames{1}, paramValues{ind, 1}, varNames{2}, paramValues{ind, 2}, globalParams{:}}, ...
experimentDirName, sprintf('%s%s_%s_%s_%s_seed%d', globalName, varDescr{1}, paramStrings{ind, 1}, varDescr{2}, paramStrings{ind, 2}, rSeeds(1)));
end
else
parfor ind1 = 1 : length(rSeeds)
for ind2 = 1 : nParams
% for ind2 = 1 : length(rSeeds)
sprintf('%s_%s_%s_%s', varDescr{1}, paramStrings{ind2, 1}, varDescr{2}, paramStrings{ind2, 2})
OES2Muscles(nIters, rSeeds(ind1), 1, ...
{varNames{1}, paramValues{ind2, 1}, varNames{2}, paramValues{ind2, 2}, globalParams{:}}, ...
experimentDirName, sprintf('%s%s_%s_%s_%s_seed%d', globalName, varDescr{1}, paramStrings{ind2, 1}, varDescr{2}, paramStrings{ind2, 2}, rSeeds(ind1)));
end
end
end