@@ -181,8 +181,8 @@ output1 = gng_m1("example", ncore=4)
181
181
## Chain 1:
182
182
## Chain 1:
183
183
## Chain 1:
184
- ## Chain 1: Gradient evaluation took 0.001939 seconds
185
- ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 19.39 seconds.
184
+ ## Chain 1: Gradient evaluation took 0.001829 seconds
185
+ ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 18.29 seconds.
186
186
## Chain 1: Adjust your expectations accordingly!
187
187
## Chain 1:
188
188
## Chain 1:
@@ -196,18 +196,18 @@ output1 = gng_m1("example", ncore=4)
196
196
## Chain 1:
197
197
## Chain 1: Begin stochastic gradient ascent.
198
198
## Chain 1: iter ELBO delta_ELBO_mean delta_ELBO_med notes
199
- ## Chain 1: 100 -830.450 1.000 1.000
200
- ## Chain 1: 200 -815.664 0.509 1.000
201
- ## Chain 1: 300 -812.693 0.341 0.018
202
- ## Chain 1: 400 -809.323 0.256 0.018
203
- ## Chain 1: 500 -809.234 0.205 0.004 MEDIAN ELBO CONVERGED
199
+ ## Chain 1: 100 -820.269 1.000 1.000
200
+ ## Chain 1: 200 -810.308 0.506 1.000
201
+ ## Chain 1: 300 -815.111 0.339 0.012
202
+ ## Chain 1: 400 -809.368 0.256 0.012
203
+ ## Chain 1: 500 -809.646 0.205 0.007 MEDIAN ELBO CONVERGED
204
204
## Chain 1:
205
205
## Chain 1: Drawing a sample of size 1000 from the approximate posterior...
206
206
## Chain 1: COMPLETED.
207
207
```
208
208
209
209
```
210
- ## Warning: Pareto k diagnostic value is 1.09 . Resampling is disabled.
210
+ ## Warning: Pareto k diagnostic value is 1.25 . Resampling is disabled.
211
211
## Decreasing tol_rel_obj may help if variational algorithm has terminated
212
212
## prematurely. Otherwise consider using sampling instead.
213
213
```
@@ -352,16 +352,16 @@ output1$allIndPars
352
352
353
353
```
354
354
## subjID xi ep rho
355
- ## 1 1 0.03912684 0.1390364 5.971566
356
- ## 2 2 0.03559554 0.1622292 6.154059
357
- ## 3 3 0.04195460 0.1277940 5.922376
358
- ## 4 4 0.03149474 0.1494447 6.223886
359
- ## 5 5 0.03442572 0.1491020 6.168325
360
- ## 6 6 0.04100730 0.1539260 6.288472
361
- ## 7 7 0.04275452 0.1481033 5.792658
362
- ## 8 8 0.03397865 0.1612648 6.510263
363
- ## 9 9 0.03957498 0.1452006 6.064876
364
- ## 10 10 0.04719602 0.1302818 5.554479
355
+ ## 1 1 0.03937858 0.1388763 5.991021
356
+ ## 2 2 0.03602277 0.1614945 6.180092
357
+ ## 3 3 0.04288713 0.1274827 5.941119
358
+ ## 4 4 0.03170505 0.1484355 6.262789
359
+ ## 5 5 0.03462090 0.1485741 6.184602
360
+ ## 6 6 0.04236850 0.1536645 6.334553
361
+ ## 7 7 0.04314376 0.1491778 5.797528
362
+ ## 8 8 0.03471143 0.1611320 6.538876
363
+ ## 9 9 0.03987275 0.1451317 6.083010
364
+ ## 10 10 0.04784353 0.1302289 5.546315
365
365
```
366
366
-->
367
367
@@ -454,8 +454,8 @@ output3 = gng_m3(data="example", niter=2000, nwarmup=1000, modelRegressor=TRUE)
454
454
## Chain 1:
455
455
## Chain 1:
456
456
## Chain 1:
457
- ## Chain 1: Gradient evaluation took 0.004173 seconds
458
- ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 41.73 seconds.
457
+ ## Chain 1: Gradient evaluation took 0.00253 seconds
458
+ ## Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 25.3 seconds.
459
459
## Chain 1: Adjust your expectations accordingly!
460
460
## Chain 1:
461
461
## Chain 1:
@@ -469,18 +469,18 @@ output3 = gng_m3(data="example", niter=2000, nwarmup=1000, modelRegressor=TRUE)
469
469
## Chain 1:
470
470
## Chain 1: Begin stochastic gradient ascent.
471
471
## Chain 1: iter ELBO delta_ELBO_mean delta_ELBO_med notes
472
- ## Chain 1: 100 -833.499 1.000 1.000
473
- ## Chain 1: 200 -819.287 0.509 1.000
474
- ## Chain 1: 300 -819.175 0.339 0.017
475
- ## Chain 1: 400 -823.919 0.256 0.017
476
- ## Chain 1: 500 -818.524 0.206 0.007 MEDIAN ELBO CONVERGED
472
+ ## Chain 1: 100 -823.918 1.000 1.000
473
+ ## Chain 1: 200 -826.958 0.502 1.000
474
+ ## Chain 1: 300 -814.838 0.340 0.015
475
+ ## Chain 1: 400 -818.443 0.256 0.015
476
+ ## Chain 1: 500 -817.985 0.205 0.004 MEDIAN ELBO CONVERGED
477
477
## Chain 1:
478
478
## Chain 1: Drawing a sample of size 1000 from the approximate posterior...
479
479
## Chain 1: COMPLETED.
480
480
```
481
481
482
482
```
483
- ## Warning: Pareto k diagnostic value is 1.34 . Resampling is disabled.
483
+ ## Warning: Pareto k diagnostic value is 1.14 . Resampling is disabled.
484
484
## Decreasing tol_rel_obj may help if variational algorithm has terminated
485
485
## prematurely. Otherwise consider using sampling instead.
486
486
```
0 commit comments