|
|
|
@207
|
[207]
|
06/02/08 19:04:49 |
edsuom |
Got Rao-Blackwellized version working locally & remotely, with better but … |
|
|
|
@206
|
[206]
|
06/01/08 02:06:56 |
edsuom |
Got Rao-Blackwellized (on z only) version running locally, not yet … |
|
|
|
@205
|
[205]
|
06/01/08 00:22:10 |
edsuom |
Rao-Blackwellization on z looks very promising, but let's ditch it on … |
|
|
|
@203
|
[203]
|
05/30/08 08:27:51 |
edsuom |
Switching from attempting estimate of L with integration to … |
|
|
|
@202
|
[202]
|
05/28/08 23:43:21 |
edsuom |
Using normal proposal on z instead of uniform; same result at convergence, … |
|
|
|
@201
|
[201]
|
05/28/08 17:18:11 |
edsuom |
Working on reducing variance in likelihood estimates |
|
|
|
@200
|
[200]
|
05/27/08 16:46:38 |
edsuom |
Integration over log-volatility simulations using Metropolis-within-Gibbs … |
|
|
|
@199
|
[199]
|
05/27/08 15:37:01 |
edsuom |
Integration over log-volatility simulations using Metropolis-within-Gibbs … |
|
|
|
@198
|
[198]
|
05/27/08 00:19:36 |
edsuom |
Tried Metropolis-within-Gibbs, but ran into the difficulty of getting the … |
|
|
|
@196
|
[196]
|
05/24/08 00:05:18 |
edsuom |
Huge progress, got PMC running with noticably less variance in population … |
|
|
|
@195
|
[195]
|
05/23/08 16:16:25 |
edsuom |
Ditching Rao-Blackwellization for now |
|
|
|
@194
|
[194]
|
05/23/08 15:04:33 |
edsuom |
PMC now running with more efficient proposal code and sequencing of said … |
|
|
|
@193
|
[193]
|
05/23/08 13:59:26 |
edsuom |
Rewrote PriorContainer? to be (hopefully) much more efficient with … |
|
|
|
@192
|
[192]
|
05/23/08 11:43:54 |
edsuom |
Got C code working nicely and integrated with pmc; working on making … |
|
|
|
@191
|
[191]
|
05/22/08 17:21:43 |
edsuom |
Figured out N1=10, N2=2, Ne=3 sim params |
|
|
|
@190
|
[190]
|
05/21/08 23:13:52 |
edsuom |
Got C-based hybrid Gibbs working; Three-sample Lh evaluation appears … |
|
|
|
@189
|
[189]
|
05/21/08 18:16:16 |
edsuom |
C code for model.likelihood appears to be working, but unit test for the … |
|
|
|
@188
|
[188]
|
05/21/08 13:20:34 |
edsuom |
About to switch to using a single set of proposal samples instead of two |
|
|
|
@187
|
[187]
|
05/20/08 18:09:32 |
edsuom |
Working on putting the likelihood method together, finally |
|
|
|
@186
|
[186]
|
05/20/08 15:42:21 |
edsuom |
Working on putting the likelihood method together, finally |
|
|
|
@185
|
[185]
|
05/19/08 15:06:13 |
edsuom |
Unit testing support C code for model |
|
|
|
@184
|
[184]
|
05/17/08 01:31:43 |
edsuom |
Grinding along with C code for new model |
|
|
|
@183
|
[183]
|
05/16/08 16:40:27 |
edsuom |
Grinding along with C code for new model |
|
|
|
@182
|
[182]
|
05/15/08 13:35:24 |
edsuom |
Grinding along with C code for new model |
|
|
|
@181
|
[181]
|
05/15/08 10:05:12 |
edsuom |
Unit testing model support C code |
|
|
|
@180
|
[180]
|
05/15/08 01:27:07 |
edsuom |
Grinding along with C code for new model |
|
|
|
@179
|
[179]
|
05/14/08 17:30:57 |
edsuom |
Getting support C code for new model working and tested |
|
|
|
@178
|
[178]
|
05/12/08 21:29:10 |
edsuom |
Rewrote zProposal to use truncated normal |
|
|
|
@177
|
[177]
|
05/10/08 22:37:06 |
edsuom |
Working on new truncated-normal zProposal |
|
|
|
@176
|
[176]
|
05/09/08 22:13:35 |
edsuom |
Redoing Rao-Blackwellization as the sum of the joint proposal-jump … |
|
|
|
@175
|
[175]
|
05/08/08 23:30:51 |
edsuom |
Got D-Kernel PMC running with what I hope is Rao-Blackwellization |
|
|
|
@174
|
[174]
|
05/08/08 12:49:59 |
edsuom |
Working on D-kernel PMC |
|
|
|
@173
|
[173]
|
05/05/08 16:21:36 |
edsuom |
Starting work on Rao-Blackwellization |
|
|
|
@172
|
[172]
|
05/01/08 21:33:08 |
edsuom |
Trying to get an actual PMC run going |
|
|
|
@171
|
[171]
|
05/01/08 15:10:42 |
edsuom |
Finally doing some inital runs & debugging |
|
|
|
@170
|
[170]
|
04/30/08 23:02:46 |
edsuom |
Putting together svpmc example |
|
|
|
@169
|
[169]
|
04/30/08 23:00:30 |
edsuom |
Putting together svpmc example |
|
|
|
@168
|
[168]
|
04/30/08 20:13:51 |
edsuom |
Got model unit testing working again; more complete unit testing of … |
|
|
|
@167
|
[167]
|
04/30/08 13:37:56 |
edsuom |
Getting model unit tests working again after refactoring |
|
|
|
@166
|
[166]
|
04/29/08 22:46:52 |
edsuom |
Got pmc module cleanly prototyped, now need to unit test; then we should … |
|
|
|
@165
|
[165]
|
04/29/08 17:21:15 |
edsuom |
ProjectManager? tested OK! |
|
|
|
@164
|
[164]
|
04/29/08 15:35:34 |
edsuom |
Working on ProjectManager?, pulling things together |
|
|
|
@163
|
[163]
|
04/29/08 01:32:02 |
edsuom |
Working on prior and parameter containers, starting to tie everything … |
|
|
|
@161
|
[161]
|
04/28/08 21:37:01 |
edsuom |
Working on prior and parameter containers, starting to tie everything … |
|
|
|
@160
|
[160]
|
04/26/08 23:27:47 |
edsuom |
Working on project specification processing |
|
|
|
@159
|
[159]
|
04/25/08 23:28:27 |
edsuom |
Innovation & volatility shocks can have non-unity variances; working on … |
|
|
|
@158
|
[158]
|
04/25/08 00:24:02 |
edsuom |
Got log-volatility Metropolis-within-Gibbs simulation working!!! |
|
|
|
@157
|
[157]
|
04/24/08 20:15:13 |
edsuom |
Got log-volatility Metropolis-within-Gibbs simulation working!!! |
|
|
|
@156
|
[156]
|
04/23/08 15:34:26 |
edsuom |
Working on Model.logLikelihood: VAR process tested |
|
|
|
@155
|
[155]
|
04/23/08 02:17:19 |
edsuom |
Got hybrid Gibbs Model.likelihood prototyped |
|
|
|
@154
|
[154]
|
04/22/08 19:38:33 |
edsuom |
weave.Weaver.inline has the same value-returning behavior as weave.inline; … |
|
|
|
@153
|
[153]
|
04/21/08 21:26:43 |
edsuom |
Working on Model.likelihood, some really hairy C code |
|
|
|
@152
|
[152]
|
04/19/08 23:20:46 |
edsuom |
Working on hybrid Gibbs log-volatility simulation and likelihood … |
|
|
|
@151
|
[151]
|
04/19/08 20:26:56 |
edsuom |
Tested FlexArray?; last work on Model.likelihood before switching it to … |
|
|
|
@150
|
[150]
|
04/17/08 21:17:12 |
edsuom |
Working on parameter containers and associated glue-type stuff |
|
|
|
@149
|
[149]
|
04/17/08 17:11:23 |
edsuom |
Working on model.likelihood |
|
|
|
@148
|
[148]
|
04/16/08 20:05:10 |
edsuom |
Refined testing of Model's multivariate stuff; working on pInnovation |
|
|
|
@147
|
[147]
|
04/16/08 16:27:16 |
edsuom |
Got decorrelation working; weave.Weaver now uses inline as its method … |
|
|
|
@146
|
[146]
|
04/11/08 00:47:57 |
edsuom |
Unit tested Model.draw_h |
|
|
|
@145
|
[145]
|
04/10/08 22:47:22 |
edsuom |
Improvements to weave.Weaver; much better unit testing for VAR |
|
|
|
@144
|
[144]
|
04/09/08 16:46:00 |
edsuom |
Working on model |
|
|
|
@143
|
[143]
|
04/08/08 21:42:51 |
edsuom |
Working on model |
|
|
|
@141
|
[141]
|
04/08/08 03:38:21 |
edsuom |
Got VAR(1)+drift process working forward & back |
|
|
|
@140
|
[140]
|
04/07/08 22:25:54 |
edsuom |
Got cool & convenient Weaver mixin written and tested; refactored … |
|
|
|
@139
|
[139]
|
04/05/08 17:11:44 |
edsuom |
Got NormalWalk? working efficiently and tested |
|
|
|
@138
|
[138]
|
04/05/08 15:45:20 |
edsuom |
Working on MCMC for normal random walk |
|
|
|
@137
|
[137]
|
04/05/08 13:51:58 |
edsuom |
New base class 'Weave' for convenience with Weave inline operations; … |
|
|
|
@136
|
[136]
|
04/05/08 12:15:17 |
edsuom |
Working on MCMC simulation of the normals underlying the volatilities |
|
|
|
@135
|
[135]
|
04/03/08 22:20:51 |
edsuom |
Renaming pypmc to svpmc to reflect specific direction of the subpackage |
|
|
|
@134
|
[134]
|
04/03/08 22:14:41 |
edsuom |
Renaming pypmc to svpmc to reflect specific direction of the subpackage |
|
copied from projects/AsynCluster/trunk/pypmc:
|
|
|
|
@131
|
[131]
|
03/27/08 20:15:08 |
edsuom |
Replacing long-inactive gevolver subpackage with pypmc |
