-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathcli.html
More file actions
979 lines (796 loc) · 73.5 KB
/
Copy pathcli.html
File metadata and controls
979 lines (796 loc) · 73.5 KB
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
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
<!DOCTYPE html>
<html lang="en" data-content_root="./" >
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" /><meta name="viewport" content="width=device-width, initial-scale=1" />
<title>Command-line interface — DeepQMC 1.3.0 documentation</title>
<script data-cfasync="false">
document.documentElement.dataset.mode = localStorage.getItem("mode") || "";
document.documentElement.dataset.theme = localStorage.getItem("theme") || "";
</script>
<!--
this give us a css class that will be invisible only if js is disabled
-->
<noscript>
<style>
.pst-js-only { display: none !important; }
</style>
</noscript>
<!-- Loaded before other Sphinx assets -->
<link href="_static/styles/theme.css?digest=a95f357e85573c9b56d5" rel="stylesheet" />
<link href="_static/styles/pydata-sphinx-theme.css?digest=a95f357e85573c9b56d5" rel="stylesheet" />
<link rel="stylesheet" type="text/css" href="_static/pygments.css?v=8f2a1f02" />
<link rel="stylesheet" type="text/css" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/katex.min.css" />
<link rel="stylesheet" type="text/css" href="_static/katex-math.css?v=05624691" />
<link rel="stylesheet" type="text/css" href="_static/custom.css?v=db83303b" />
<!-- So that users can add custom icons -->
<script defer src="_static/scripts/fontawesome.js?digest=a95f357e85573c9b56d5"></script>
<!-- Pre-loaded scripts that we'll load fully later -->
<link rel="preload" as="script" href="_static/scripts/bootstrap.js?digest=a95f357e85573c9b56d5" />
<link rel="preload" as="script" href="_static/scripts/pydata-sphinx-theme.js?digest=a95f357e85573c9b56d5" />
<script src="_static/documentation_options.js?v=1f29e9d3"></script>
<script src="_static/doctools.js?v=fd6eb6e6"></script>
<script src="_static/sphinx_highlight.js?v=6ffebe34"></script>
<script src="_static/katex.min.js?v=5cc8ed51"></script>
<script src="_static/auto-render.min.js?v=af98beb9"></script>
<script src="_static/katex_autorenderer.js?v=bebc588a"></script>
<script crossorigin="anonymous" integrity="sha256-Ae2Vz/4ePdIu6ZyI/5ZGsYnb+m0JlOmKPjt6XZ9JJkA=" src="https://cdnjs.cloudflare.com/ajax/libs/require.js/2.3.4/require.min.js"></script>
<script>DOCUMENTATION_OPTIONS.pagename = 'cli';</script>
<script>DOCUMENTATION_OPTIONS.search_as_you_type = false;</script>
<link rel="index" title="Index" href="genindex.html" />
<link rel="search" title="Search" href="search.html" />
<link rel="next" title="Tutorial" href="tutorial.html" />
<link rel="prev" title="Installation" href="installation.html" />
<meta name="viewport" content="width=device-width, initial-scale=1"/>
<meta name="docsearch:language" content="en"/>
<meta name="docsearch:version" content="1.3.0" />
<script src="_static/searchtools.js"></script>
<script src="_static/language_data.js"></script>
<script src="searchindex.js"></script>
</head>
<body data-default-mode="">
<div id="pst-skip-link" class="skip-link d-print-none"><a href="#main-content">Skip to main content</a></div>
<div id="pst-scroll-pixel-helper"></div>
<button type="button" class="btn rounded-pill" id="pst-back-to-top">
<i class="fa-solid fa-arrow-up"></i>Back to top</button>
<dialog id="pst-search-dialog">
<form class="bd-search d-flex align-items-center"
action="search.html"
method="get">
<i class="fa-solid fa-magnifying-glass"></i>
<input type="search"
class="form-control"
name="q"
placeholder="Search the docs ..."
aria-label="Search the docs ..."
autocomplete="off"
autocorrect="off"
autocapitalize="off"
spellcheck="false"/>
<span class="search-button__kbd-shortcut"><kbd class="kbd-shortcut__modifier">Ctrl</kbd>+<kbd>K</kbd></span>
</form>
</dialog>
<div class="pst-async-banner-revealer d-none">
<aside id="bd-header-version-warning" class="d-none d-print-none" aria-label="Version warning"></aside>
</div>
<header id="pst-header" class="bd-header navbar navbar-expand-lg bd-navbar d-print-none">
<div class="bd-header__inner bd-page-width">
<button class="pst-navbar-icon sidebar-toggle primary-toggle" aria-label="Site navigation">
<span class="fa-solid fa-bars"></span>
</button>
<div class="col-lg-3 navbar-header-items__start">
<div class="navbar-item">
<a class="navbar-brand logo" href="index.html">
<p class="title logo__title">DeepQMC 1.3.0 documentation</p>
</a></div>
</div>
<div class="col-lg-9 navbar-header-items">
<div class="me-auto navbar-header-items__center">
<div class="navbar-item">
<nav>
<ul class="bd-navbar-elements navbar-nav">
<li class="nav-item ">
<a class="nav-link nav-internal" href="installation.html">
Installation
</a>
</li>
<li class="nav-item current active">
<a class="nav-link nav-internal" href="#">
Command-line interface
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="tutorial.html">
Tutorial
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="examples.html">
Examples
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="api.html">
API
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="refs.html">
References
</a>
</li>
</ul>
</nav></div>
</div>
<div class="navbar-header-items__end">
<div class="navbar-item navbar-persistent--container">
<button class="btn btn-sm pst-navbar-icon search-button search-button__button pst-js-only" title="Search" aria-label="Search" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="fa-solid fa-magnifying-glass fa-lg"></i>
</button>
</div>
<div class="navbar-item">
<div class="theme-switch-container dropdown pst-js-only" data-bs-toggle="tooltip" data-bs-placement="bottom" title="Color mode">
<button class="btn btn-sm nav-link pst-navbar-icon theme-switch-button dropdown-toggle" aria-label="Color mode" data-bs-toggle="dropdown">
<i class="theme-switch fa-solid fa-sun fa-lg fa-fw" data-mode="light" title="Light"></i>
<i class="theme-switch fa-solid fa-moon fa-lg fa-fw" data-mode="dark" title="Dark"></i>
<i class="theme-switch fa-solid fa-circle-half-stroke fa-lg fa-fw" data-mode="auto" title="System Settings"></i>
</button>
<ul class="dropdown-menu dropdown-menu-end">
<li><button class="dropdown-item d-flex align-items-center theme-change-button" data-mode="auto"><i class="fa-solid fa-circle-half-stroke fa-lg fa-fw me-1"></i>System Settings</button></li>
<li><button class="dropdown-item d-flex align-items-center theme-change-button" data-mode="light"><i class="fa-solid fa-sun fa-lg fa-fw me-1"></i>Light</button></li>
<li><button class="dropdown-item d-flex align-items-center theme-change-button" data-mode="dark"><i class="fa-solid fa-moon fa-lg fa-fw me-1"></i>Dark</button></li>
</ul>
</div></div>
<div class="navbar-item"><ul class="navbar-icon-links"
aria-label="Icon Links">
<li class="nav-item">
<a href="https://github.com/deepqmc/deepqmc" title="GitHub" class="nav-link pst-navbar-icon" rel="noopener" target="_blank" data-bs-toggle="tooltip" data-bs-placement="bottom"><i class="fa-brands fa-github fa-lg" aria-hidden="true"></i><span class="visually-hidden">GitHub</span></a>
</li>
</ul></div>
</div>
</div>
<div class="navbar-persistent--mobile">
<button class="btn btn-sm pst-navbar-icon search-button search-button__button pst-js-only" title="Search" aria-label="Search" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="fa-solid fa-magnifying-glass fa-lg"></i>
</button>
</div>
</div>
</header>
<div class="bd-container">
<div class="bd-container__inner bd-page-width">
<dialog id="pst-primary-sidebar-modal"></dialog>
<div id="pst-primary-sidebar" class="bd-sidebar-primary bd-sidebar">
<div class="sidebar-header-items sidebar-primary__section">
<div class="sidebar-header-items__center">
<div class="navbar-item">
<nav>
<ul class="bd-navbar-elements navbar-nav">
<li class="nav-item ">
<a class="nav-link nav-internal" href="installation.html">
Installation
</a>
</li>
<li class="nav-item current active">
<a class="nav-link nav-internal" href="#">
Command-line interface
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="tutorial.html">
Tutorial
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="examples.html">
Examples
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="api.html">
API
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="refs.html">
References
</a>
</li>
</ul>
</nav></div>
</div>
<div class="sidebar-header-items__end">
<div class="navbar-item">
<div class="theme-switch-container dropdown pst-js-only" data-bs-toggle="tooltip" data-bs-placement="bottom" title="Color mode">
<button class="btn btn-sm nav-link pst-navbar-icon theme-switch-button dropdown-toggle" aria-label="Color mode" data-bs-toggle="dropdown">
<i class="theme-switch fa-solid fa-sun fa-lg fa-fw" data-mode="light" title="Light"></i>
<i class="theme-switch fa-solid fa-moon fa-lg fa-fw" data-mode="dark" title="Dark"></i>
<i class="theme-switch fa-solid fa-circle-half-stroke fa-lg fa-fw" data-mode="auto" title="System Settings"></i>
</button>
<ul class="dropdown-menu dropdown-menu-end">
<li><button class="dropdown-item d-flex align-items-center theme-change-button" data-mode="auto"><i class="fa-solid fa-circle-half-stroke fa-lg fa-fw me-1"></i>System Settings</button></li>
<li><button class="dropdown-item d-flex align-items-center theme-change-button" data-mode="light"><i class="fa-solid fa-sun fa-lg fa-fw me-1"></i>Light</button></li>
<li><button class="dropdown-item d-flex align-items-center theme-change-button" data-mode="dark"><i class="fa-solid fa-moon fa-lg fa-fw me-1"></i>Dark</button></li>
</ul>
</div></div>
<div class="navbar-item"><ul class="navbar-icon-links"
aria-label="Icon Links">
<li class="nav-item">
<a href="https://github.com/deepqmc/deepqmc" title="GitHub" class="nav-link pst-navbar-icon" rel="noopener" target="_blank" data-bs-toggle="tooltip" data-bs-placement="bottom"><i class="fa-brands fa-github fa-lg" aria-hidden="true"></i><span class="visually-hidden">GitHub</span></a>
</li>
</ul></div>
</div>
</div>
<div class="sidebar-primary-items__start sidebar-primary__section">
<div class="sidebar-primary-item">
<div
id="pst-page-navigation-heading-2"
class="page-toc tocsection onthispage">
<i class="fa-solid fa-list"></i> On this page
</div>
<nav id="pst-page-toc-nav" class="page-toc" aria-labelledby="pst-page-navigation-heading-2">
<ul class="pst-show_toc_level nav section-nav flex-column">
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#basics">Basics</a></li>
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#execution-on-multiple-gpus">Execution on multiple GPUs</a></li>
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#hyperparameters">Hyperparameters</a><ul class="nav section-nav flex-column">
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#task">Task</a></li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#hamiltonian">Hamiltonian</a><ul class="nav section-nav flex-column">
<li class="toc-h4 nav-item toc-entry"><a class="reference internal nav-link" href="#pseudopotentials-and-pseudohamiltonains">Pseudopotentials and pseudohamiltonains</a></li>
</ul>
</li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#sampling">Sampling</a></li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#optimization">Optimization</a></li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#excited-states">Excited States</a><ul class="nav section-nav flex-column">
<li class="toc-h4 nav-item toc-entry"><a class="reference internal nav-link" href="#excited-states-with-via-orthogonalization">Excited states with via orthogonalization</a></li>
<li class="toc-h4 nav-item toc-entry"><a class="reference internal nav-link" href="#targeting-spin-sectors-with-a-spin-penalty">Targeting spin sectors with a spin penalty</a></li>
<li class="toc-h4 nav-item toc-entry"><a class="reference internal nav-link" href="#sharing-parameters-across-excited-states">Sharing parameters across excited states</a></li>
</ul>
</li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#transferable-training">Transferable Training</a><ul class="nav section-nav flex-column">
<li class="toc-h4 nav-item toc-entry"><a class="reference internal nav-link" href="#sharing-parameters-for-a-fixed-dataset">Sharing parameters for a fixed dataset</a></li>
<li class="toc-h4 nav-item toc-entry"><a class="reference internal nav-link" href="#dynamic-geometry-sampling">Dynamic geometry sampling</a></li>
</ul>
</li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#ansatz">Ansatz</a></li>
</ul>
</li>
</ul>
</nav></div>
</div>
<div class="sidebar-primary-items__end sidebar-primary__section">
<div class="sidebar-primary-item">
<div id="ethical-ad-placement"
class="flat"
data-ea-publisher="readthedocs"
data-ea-type="readthedocs-sidebar"
data-ea-manual="true">
</div></div>
</div>
</div>
<main id="main-content" class="bd-main" role="main">
<div class="bd-content">
<div class="bd-article-container">
<div class="bd-header-article d-print-none">
<div class="header-article-items header-article__inner">
<div class="header-article-items__start">
<div class="header-article-item">
<nav aria-label="Breadcrumb" class="d-print-none">
<ul class="bd-breadcrumbs">
<li class="breadcrumb-item breadcrumb-home">
<a href="index.html" class="nav-link" aria-label="Home">
<i class="fa-solid fa-home"></i>
</a>
</li>
<li class="breadcrumb-item active" aria-current="page"><span class="ellipsis">Command-line interface</span></li>
</ul>
</nav>
</div>
</div>
</div>
</div>
<div id="searchbox"></div>
<article class="bd-article">
<section id="command-line-interface">
<span id="cli"></span><h1>Command-line interface<a class="headerlink" href="#command-line-interface" title="Link to this heading">#</a></h1>
<p>The most convenient and reproducible way of running an experiment with DeepQMC is through the <a class="reference external" href="https://hydra.cc/">hydra</a> based command-line interface (CLI).
The tutorial exemplifies a basic training and evaluation through the command line.
For more advanced functionality such as multiruns or interaction with slurm see the <a class="reference external" href="https://hydra.cc/docs/intro/">hydra docs</a>.</p>
<p>The CLI provides simple access to the functionalities of the DeepQMC package. The main tasks comprise <code class="docutils literal notranslate"><span class="pre">train</span></code>, <code class="docutils literal notranslate"><span class="pre">restart</span></code> and <code class="docutils literal notranslate"><span class="pre">evaluate</span></code>, which are thin wrappers around the <a class="reference internal" href="api.html#training-and-evaluation"><span class="std std-ref">train</span></a> function.</p>
<blockquote>
<div><p>Available tasks:</p>
<ul class="simple">
<li><p><code class="docutils literal notranslate"><span class="pre">train</span></code>: Trains the ansatz with variational Monte Carlo.</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">evaluate</span></code>: Evaluates observables (i.e. the energy) of an ansatz via Monte Carlo sampling.</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">restart</span></code>: Restarts/continues the training from a stored training checkpoint.</p></li>
</ul>
</div></blockquote>
<p>The train function creates a directory which contains the logs as well as the hyperparameters for the training (<code class="docutils literal notranslate"><span class="pre">.hydra</span></code>). For <code class="docutils literal notranslate"><span class="pre">restart</span></code> and <code class="docutils literal notranslate"><span class="pre">evaluate</span></code> the restdir of the former training run has to be provided. Specifying arguments when executing the command will overwrite the configuration stored in the restdir. This enables changing certain parameters, such as the number of training / evaluation steps, but can result in errors if the requested hyperparameters conflict with the recovered train state.</p>
<section id="basics">
<h2>Basics<a class="headerlink" href="#basics" title="Link to this heading">#</a></h2>
<p>A training can be run via:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc hydra.run.dir=workdir
</pre></div>
</div>
<p>Running the plain deepqmc command launches a default job on the LiH molecule, which is good for testing the functionality of the code.
The <cite>hydra.run.dir=workdir</cite> arguments sets the working directory of the job to “workdir”.
In the working directory several files are created, including:</p>
<ul class="simple">
<li><p><code class="docutils literal notranslate"><span class="pre">deepqmc.log</span></code> - Stores the console log of the run</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">training/events.out.tfevents.*</span></code> - Tensorboard event file</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">training/result.h5</span></code> - HDF5 file with the training trajectory</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">training/state-*.pt</span></code> - Checkpoint files with the saved state of the ansatz, optimizer and sampler at particular steps</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">training/.hydra</span></code> - Folder containing the <a class="reference external" href="https://hydra.cc/">hydra</a> config of the run</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">training/pyscf_chkpts</span></code> - Folder containing the <a class="reference external" href="https://pyscf.org/">PySCF</a> checkpoints for pretraining</p></li>
</ul>
<p>The evaluation of the energy of a trained wavefunction ansatz is obtained via:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc task=evaluate task.restdir=workdir/training
</pre></div>
</div>
<p>The training can be continued or recoverd from a training checkpoint:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc task=restart task.restdir=workdir/training
</pre></div>
</div>
<p>This again generates a Tensorboard event file <code class="docutils literal notranslate"><span class="pre">evaluation/events.out.tfevents.*</span></code> and an HDF5 file <code class="docutils literal notranslate"><span class="pre">evaluation/result.h5</span></code> file holding the sampled local energies and other observables (see <a class="reference internal" href="tutorial.html#logging"><span class="std std-ref">Tutorial/Logging</span></a>).
If a training is evaluated or continued the hydra configurations are chained together, with the latter runs referring to the configs of the former.
Note that moving the working directories during subsequent training, restarts and evaluations may lead to broken references in the config paths in <code class="docutils literal notranslate"><span class="pre">training/.hydra</span></code> and should be avoided (if necessary manual correction of the paths is required).</p>
</section>
<section id="execution-on-multiple-gpus">
<h2>Execution on multiple GPUs<a class="headerlink" href="#execution-on-multiple-gpus" title="Link to this heading">#</a></h2>
<p>DeepQMC can utilize multiple GPUs increased performance.
The algorithm is parallelised over the electron position samples, therefore the number of such samples in a batch (<code class="docutils literal notranslate"><span class="pre">electron_batch_size</span></code>) must be divisible with the number of utilized GPUs.
DeepQMC relies on JAX to automatically detect and use all available GPUs, without any configuration from the user.
It respects the <code class="docutils literal notranslate"><span class="pre">CUDA_VISIBLE_DEVICES</span></code> environment variable if it’s defined, and only uses the GPUs specified there.
A short log message at the beginning of the run informs the user of the number of utilized GPUs.</p>
<div class="highlight-text notranslate"><div class="highlight"><pre><span></span>INFO:deepqmc.app: Running on X GPU_NAME with Y processes
</pre></div>
</div>
</section>
<section id="hyperparameters">
<span id="id2"></span><h2>Hyperparameters<a class="headerlink" href="#hyperparameters" title="Link to this heading">#</a></h2>
<p>In the following the most relevant settings for running experiments with DeepQMC are discussed.
Various application examples with a more in depths explanation of the arguments are provided under the <a class="reference internal" href="examples.html#examples"><span class="std std-ref">examples</span></a> page.</p>
<section id="task">
<h3>Task<a class="headerlink" href="#task" title="Link to this heading">#</a></h3>
<p>DeepQMC provides the above mentioned configurations for the <code class="docutils literal notranslate"><span class="pre">train</span></code>, <code class="docutils literal notranslate"><span class="pre">evaluate</span></code> and <code class="docutils literal notranslate"><span class="pre">restart</span></code> task.
In order to override default hyperparameters of the experimental setup, such as the <code class="docutils literal notranslate"><span class="pre">sample_size</span></code> or the number of training <code class="docutils literal notranslate"><span class="pre">steps</span></code> or <code class="docutils literal notranslate"><span class="pre">pretrain_steps</span></code>, hydra provides a simple syntax:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc task=train task.electron_batch_size=2048 task.steps=50000 task.pretrain_steps=5000
</pre></div>
</div>
<p>The working directory for logging and checkpointing is is defined through:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc hydra.run.dir=workdir
</pre></div>
</div>
<p>If no Logging directory is provided a directory <code class="docutils literal notranslate"><span class="pre">outputs/YEAR-MONTH-DAY/HOUR-MINUTE-SECOND</span></code> is being used.
Note that the working directory of an <code class="docutils literal notranslate"><span class="pre">evaluate</span></code> and <code class="docutils literal notranslate"><span class="pre">restart</span></code> task cannot match the value of their <code class="docutils literal notranslate"><span class="pre">restdir</span></code> option.</p>
</section>
<section id="hamiltonian">
<h3>Hamiltonian<a class="headerlink" href="#hamiltonian" title="Link to this heading">#</a></h3>
<p>DeepQMC aims at solving the molecular Hamiltonian. Molecules can be selected from a range of predefined configurations located in <code class="docutils literal notranslate"><span class="pre">.../deepqmc/src/deepqmc/conf/hamil/mol</span></code>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc hamil/mol=LiH
</pre></div>
</div>
<p>The hydra syntax allows specifying molecules on the command line:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc hamil.mol.coords=[[0,0,0],[0.742,0,0]] hamil.mol.charges=[1,1] hamil.mol.charge=0 hamil.mol.spin=0 hamil.mol.unit=angstrom
</pre></div>
</div>
<p>In practice it is often more convenient to create custom YAML files (for examples check the <code class="docutils literal notranslate"><span class="pre">.../deepqmc/src/deepqmc/conf/hamil/mol</span></code> folder) and load them with:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc hamil/mol=from_file hamil.mol.file=relative/path/to/molecule/file.yaml
</pre></div>
</div>
<section id="pseudopotentials-and-pseudohamiltonains">
<h4>Pseudopotentials and pseudohamiltonains<a class="headerlink" href="#pseudopotentials-and-pseudohamiltonains" title="Link to this heading">#</a></h4>
<p>DeepQMC implements the option to use pseudopotentials <a class="reference internal" href="refs.html#burkatzki07" id="id3"><span>[Burkatzki07]</span></a> <a class="reference internal" href="refs.html#bennett17" id="id4"><span>[Bennett17]</span></a> and pseudohamiltonains <a class="reference internal" href="refs.html#ichibha23" id="id5"><span>[Ichibha23]</span></a> <a class="reference internal" href="refs.html#fu26" id="id6"><span>[Fu26]</span></a>.
Pseudopotentials and pseudohamiltonians are enabled via the <cite>ecp_type</cite> and <cite>ecp_mask</cite> arguments of the hamiltonian.
For the ccECP type pseudopotential use <cite>ecp_type=’ccECP’</cite>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc hamil/mol='Sc' +hamil.ecp_type='ccECP'
</pre></div>
</div>
<p>For the pseudohamiltonians use <cite>ecp_type=PH</cite>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc hamil/mol='Fe' +hamil.ecp_type='PH'
</pre></div>
</div>
<p>DeepQMC supports the Gaussian-type pseudopotentials from PySCF as well and provides pseudohamiltonians for S, Cr, Mn, Fe, Co, Ni, Cu and Zn.</p>
<p>Furthermore, DeepQMC uses the <a class="reference external" href="https://github.com/microsoft/folx">folx</a> package to utilize the forward laplacian framework <a class="reference internal" href="refs.html#li24" id="id7"><span>[Li24]</span></a>.
The forward Laplacian significantly accelerates the computation of the Laplacian at the cost of a an increased memory footprint, resulting in about a 2x overall speed-up of the simulation of intermediately sized systems.
The use of the forward laplacian is always recommended if compatible with the computational setup:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc hamil=qc_forward_laplacian
</pre></div>
</div>
</section>
</section>
<section id="sampling">
<span id="id8"></span><h3>Sampling<a class="headerlink" href="#sampling" title="Link to this heading">#</a></h3>
<p>Different sampler configurations can be found in <code class="docutils literal notranslate"><span class="pre">.../deepqmc/src/deepqmc/conf/task/sampler_factory</span></code>.
A typical usecase would be to pick as sampler form these configurations and, if required, change some argument from the command line:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc task/sampler_factory=decorr_langevin task.sampler_factory.elec_sampler.samplers.0.length=30
</pre></div>
</div>
</section>
<section id="optimization">
<h3>Optimization<a class="headerlink" href="#optimization" title="Link to this heading">#</a></h3>
<p>For the optimization either <a class="reference external" href="https://kfac-jax.readthedocs.io/en/latest/">KFAC</a> or optimizers from <a class="reference external" href="https://optax.readthedocs.io/en/latest/">optax</a> may be used.
While the use of <a class="reference external" href="https://kfac-jax.readthedocs.io/en/latest/">KFAC</a> is highly recommended due to the significantly improved convergence, at times it can be useful to run with other optimizers such as <a class="reference external" href="https://optax.readthedocs.io/en/latest/api.html#adamw">AdamW</a>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc task/opt=adamw
</pre></div>
</div>
</section>
<section id="excited-states">
<h3>Excited States<a class="headerlink" href="#excited-states" title="Link to this heading">#</a></h3>
<p>DeepQMC implements penalty-based optimisation of electronic excited states.</p>
<section id="excited-states-with-via-orthogonalization">
<h4>Excited states with via orthogonalization<a class="headerlink" href="#excited-states-with-via-orthogonalization" title="Link to this heading">#</a></h4>
<p>To simulate the two lowest lying states of a molecule use:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc task.electronic_states=2
</pre></div>
</div>
<p>When simulating excited states it can be useful to pretrain with respect to orthogonal (excited) states. This is achieved by specifying a suitable cas space:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc task.electronic_states=2 +task.pretrain_kwargs.scf_kwargs.cas=[2,2]
</pre></div>
</div>
</section>
<section id="targeting-spin-sectors-with-a-spin-penalty">
<h4>Targeting spin sectors with a spin penalty<a class="headerlink" href="#targeting-spin-sectors-with-a-spin-penalty" title="Link to this heading">#</a></h4>
<p>To target states of a particular spin sector, a spin penalty can be applied:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc task.electronic_states=2 +task.loss_function_factory.spin_penalty=10
</pre></div>
</div>
<p>Setting the spin penalty penalises high spin states, i.e. favours singlet (doublet) states over triplet (quartet) states, etc.
When simulating states with higher total spin, the spin penalty is combined with setting the magnetic quantum number, i.e. setting <code class="docutils literal notranslate"><span class="pre">mol.spin=2</span></code> in the molecule configuration for a triplet.
Note that when combined with cas pretraining it is required to fix the spin in the calculation of the baseline to provide sensible pretraining targets.</p>
<p>For more details on the configuration of excited state calculations see <a class="reference internal" href="refs.html#szabo24" id="id10"><span>[Szabo24]</span></a>.</p>
</section>
<section id="sharing-parameters-across-excited-states">
<h4>Sharing parameters across excited states<a class="headerlink" href="#sharing-parameters-across-excited-states" title="Link to this heading">#</a></h4>
<p>Wave function parameters can be shared across excited states by using the <code class="docutils literal notranslate"><span class="pre">merge_keys</span></code> argument.
This accepts a list of strings that are matched against the parameter names, i.e. all the parameters of the graph neural network / transformer layers:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc ansatz=psiformer task.electronic_states=2 +task.merge_keys="['embedding', 'gnn']"
</pre></div>
</div>
<p>A list of the shared parameters will be logged at the beginning of the training.</p>
</section>
</section>
<section id="transferable-training">
<h3>Transferable Training<a class="headerlink" href="#transferable-training" title="Link to this heading">#</a></h3>
<p>DeepQMC implements geometrically transferable training, that is a single ansatz can be trained across multiple molecular configurations.
This significantly reduces the computational cost over equivalent independent single-point simulations, improves relative energies and can be used for interpolation of entire potential energy surfaces, including for ab initio geometry optimization <a class="reference internal" href="refs.html#szabo26" id="id11"><span>[Szabo26]</span></a>.
Transferable simulation can be done in various modes and can be combined with excited state simulation and spin penalties.
Here we provide a basic introduction. For more examples see the <a class="reference internal" href="examples.html#examples"><span class="std std-ref">examples</span></a> page.</p>
<section id="sharing-parameters-for-a-fixed-dataset">
<h4>Sharing parameters for a fixed dataset<a class="headerlink" href="#sharing-parameters-for-a-fixed-dataset" title="Link to this heading">#</a></h4>
<p>The simplest version of geometric transferability can be thought of as sharing wave function parameters across multiple parallel DeepQMC simulations on different geometric configurations of the same molecule.
Therefore, we instantiate a joint training run on multiple fixed geometries by providing the optional <code class="docutils literal notranslate"><span class="pre">mols</span></code> argument to the train function.
The most convenient way to instantiate a transferable training run is by using the <code class="docutils literal notranslate"><span class="pre">task=train_transferable</span></code> config and providing a directory in which molecule configurations are stored as yaml files.
Here we generate a dataset composed of two configurations of the LiH molecule:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ mkdir -p mols_dir
$ cat > mols_dir/LiH_eq.yaml <<EOF
coords: [[0.0, 0.0, 0.0], [3.014, 0.0, 0.0]]
charges: [3, 1]
charge: 0
spin: 0
EOF
$ cat > mols_dir/LiH_stretch.yaml <<EOF
coords: [[0.0, 0.0, 0.0], [6, 0.0, 0.0]]
charges: [3, 1]
charge: 0
spin: 0
EOF
</pre></div>
</div>
<p>We then run a transferable DeepQMC simulation:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc hydra.run.dir=workdir_transferable task=train_transferable task.seed=42 ansatz=transpsiformer task.molecule_batch_size=2 hamil/mol=LiH task.mols.directory=mols_dir
</pre></div>
</div>
<p>Additional to providing the mols directory the molecule in the Hamiltonian needs to be set accordingly (this molecule is used for determining shapes of the wave function etc.).
Note that all the molecule configurations in <code class="docutils literal notranslate"><span class="pre">mols_dir</span></code> as well as the molecule specified in the Hamiltonian need to have the same charges, total charge, spin, etc. that is have to be equivalent up to their geometry.</p>
<p>We use the transpsiformer ansatz described in <a class="reference internal" href="refs.html#schaetzle25" id="id12"><span>[Schaetzle25]</span></a>, which is an extension of the Psiformer <a class="reference internal" href="refs.html#glehn23" id="id13"><span>[Glehn23]</span></a> that explicitly accounts for changes in the nuclear geometry.</p>
<p>Setting <code class="docutils literal notranslate"><span class="pre">molecule_batch_size=2</span></code> means that two molecules per iteration are run in parallel, each with <code class="docutils literal notranslate"><span class="pre">electron_batch_size</span></code> many walkers.
For large datasets it is possible to work only on a subset of the molecular geometries on each iteration (i.e. <code class="docutils literal notranslate"><span class="pre">molecule_batch_size<len(mols)</span></code>) and cycle through the data iteratively.
The exact choice of <code class="docutils literal notranslate"><span class="pre">electron_batch_size</span></code> and <code class="docutils literal notranslate"><span class="pre">molecule_batch_size</span></code> may depend on the experimental setup and differs among practitioners.
We typically use a <code class="docutils literal notranslate"><span class="pre">molecule_batch_size</span></code> of 4-32, with an inversely proportional <code class="docutils literal notranslate"><span class="pre">electron_batch_size</span></code> of 2048-256 scaled to optimize GPU bandwidth.</p>
<p>In this training mode individual markov chains are retained for each geometry.
Therefore, if the <code class="docutils literal notranslate"><span class="pre">molecule_batch_size</span></code> is significantly smaller than the number of molecules in the training set, walkers can go stale and it can be useful to increase the number of <a class="reference internal" href="#sampling"><span class="std std-ref">decorrelation steps</span></a>.</p>
<p>After a shared optimization the transferable ansatz can be evaluated on the geometries of interest independently:</p>
<blockquote>
<div><p>$ deepqmc hydra.run.dir=workdir_eval task=evaluate task.restdir=workdir_transferable/training +task.molecule_batch_size=1 hamil/mol=from_file hamil.mol.file=mols_dir/LiH_eq</p>
</div></blockquote>
</section>
<section id="dynamic-geometry-sampling">
<h4>Dynamic geometry sampling<a class="headerlink" href="#dynamic-geometry-sampling" title="Link to this heading">#</a></h4>
<p>Instead of training on a fixed set of molecular geometries, DeepQMC can dynamically resample the nuclear geometry throughout the optimization, directly in internal (bond length, angle and dihedral) coordinates.
This is the approach used for learning continuous potential energy surfaces and for ab initio geometry optimization <a class="reference internal" href="refs.html#szabo26" id="id14"><span>[Szabo26]</span></a>, see also <a class="reference internal" href="api.html#nuclear-geometry"><span class="std std-ref">Nuclear geometry</span></a>.</p>
<p>A dynamically sampled geometry is configured via the <code class="docutils literal notranslate"><span class="pre">nuc_sampler</span></code> argument of the sampler factory.
A convenient way to do this is to use a <a class="reference internal" href="api.html#deepqmc.sampling.nuclei_samplers.ZMatrixSampler" title="deepqmc.sampling.nuclei_samplers.ZMatrixSampler"><code class="xref py py-class docutils literal notranslate"><span class="pre">ZMatrixSampler</span></code></a> built from a <a class="reference internal" href="api.html#deepqmc.geom.zmatrix.StochasticZMatrixTemplate" title="deepqmc.geom.zmatrix.StochasticZMatrixTemplate"><code class="xref py py-class docutils literal notranslate"><span class="pre">StochasticZMatrixTemplate</span></code></a>.
The template specifies, for every atom (in the same order as the nuclear charges of the Hamiltonian’s molecule), which of its bond length, bond angle and dihedral angle relative to previously listed atoms are resampled every iteration, and from which noise distribution.
Custom sampler configurations are added as YAML files under <code class="docutils literal notranslate"><span class="pre">.../deepqmc/src/deepqmc/conf/task/sampler_factory/nuc_sampler</span></code>.
For example, the following configuration continuously stretches and compresses the Li-H bond of the LiH molecule around its reference length, clipped so that it never becomes unphysically short (bond lengths are in bohr, the atomic unit of length):</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ cat > src/deepqmc/conf/task/sampler_factory/nuc_sampler/LiH_zmat.yaml <<EOF
_target_: deepqmc.sampling.nuclei_samplers.ZMatrixSampler
_partial_: true
z_matrix_template:
_target_: deepqmc.geom.zmatrix.StochasticZMatrixTemplate.from_simplified_config
lines:
- charge: 3
atom_idxs: [null, null, null]
distribution_factories: [null, null, null]
- charge: 1
atom_idxs: [0, null, null]
distribution_factories:
- _target_: deepqmc.geom.zmatrix.ClippedNormalDistributionFactory
scale: 2.0
low: 1.5
- null
- null
EOF
</pre></div>
</div>
<p>Each entry of <code class="docutils literal notranslate"><span class="pre">lines</span></code> corresponds to one atom, with <code class="docutils literal notranslate"><span class="pre">atom_idxs</span></code> giving the (zero-based) indices of the previously listed atoms its bond, angle and dihedral are measured from (<code class="docutils literal notranslate"><span class="pre">null</span></code> where not applicable, e.g. the first atom of a Z-matrix never has any of these, and the second, as here, has at most a bond).
<a class="reference internal" href="api.html#deepqmc.geom.zmatrix.ClippedNormalDistributionFactory" title="deepqmc.geom.zmatrix.ClippedNormalDistributionFactory"><code class="xref py py-class docutils literal notranslate"><span class="pre">ClippedNormalDistributionFactory</span></code></a> samples from a normal distribution centered on the reference value found in the Hamiltonian’s molecule, clipped to an absolute range (here, at least 1.5 bohr) to avoid unphysical geometries.
Other distributions, such as <a class="reference internal" href="api.html#deepqmc.geom.zmatrix.CenteredUniformDistributionFactory" title="deepqmc.geom.zmatrix.CenteredUniformDistributionFactory"><code class="xref py py-class docutils literal notranslate"><span class="pre">CenteredUniformDistributionFactory</span></code></a>, sample uniformly within an offset around the reference value instead. A more elaborate example resampling both bond lengths and the bond angle of a water molecule is given on the <a class="reference internal" href="examples.html#examples"><span class="std std-ref">examples</span></a> page.</p>
<p>The dynamic sampler is then enabled with:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc hydra.run.dir=workdir_transferable_continous task=train_transferable task.seed=42 ansatz=transpsiformer task.molecule_batch_size=1 hamil/mol=LiH task.mols.directory=null task/sampler_factory/nuc_sampler=LiH_zmat task/sampler_factory/elec_warp_fn=nn_elec_warp +task.sampler_factory.update_nuc_period=10
</pre></div>
</div>
<p><code class="docutils literal notranslate"><span class="pre">update_nuc_period</span></code> sets how often, in training steps, a new geometry is drawn.
Since it is not part of the default sampler configuration it needs to be added with the <code class="docutils literal notranslate"><span class="pre">+</span></code> prefix.
<code class="docutils literal notranslate"><span class="pre">elec_warp_fn=nn_elec_warp</span></code> displaces each electron together with its nearest nucleus whenever the geometry changes, so that the electron walkers remain close to equilibrium and do not have to fully re-equilibrate from scratch after every geometry update.
If the geometry changes substantially between updates it can additionally help to set <code class="docutils literal notranslate"><span class="pre">+task.sampler_factory.elec_equilibration_steps</span></code>, which runs a number of extra electron sampling steps immediately after each nuclear update, before that geometry’s samples are used for training.</p>
<p>The trained wave function can now be evaluated on any geometry of interest similar to the previous example.
Typically interpolation within the training regime works very well, while extrapolation is much more challenging.
Note that, since the ansatz is not equivariant under rotations of the molecular geometry, the molecule should be aligned to a reference orientation before evaluation.</p>
</section>
</section>
<section id="ansatz">
<h3>Ansatz<a class="headerlink" href="#ansatz" title="Link to this heading">#</a></h3>
<p>The hyperparameters of the training and the wave function ansatz are specified through hydra config files. Predefined ansatzes can be found in <code class="docutils literal notranslate"><span class="pre">.../deepqmc/src/deepqmc/conf/ansatz</span></code> and selected via:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc ansatz=ferminet
</pre></div>
</div>
<p>The following predefined ansatzes are implemented:</p>
<ul class="simple">
<li><p><code class="docutils literal notranslate"><span class="pre">default</span></code>: DeepQMC’s general-purpose graph-neural-network-based wave function, from which the other predefined ansatzes are derived by adjusting hyperparameters</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">ferminet</span></code>: FermiNet <a class="reference internal" href="refs.html#pfauprr20" id="id15"><span>[PfauPRR20]</span></a></p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">psiformer</span></code>: Psiformer, a self-attention ansatz <a class="reference internal" href="refs.html#glehn23" id="id16"><span>[Glehn23]</span></a></p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">deeperwin</span></code>: DeepErwin <a class="reference internal" href="refs.html#gerard22" id="id17"><span>[Gerard22]</span></a></p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">lapnet</span></code>: LapNet, a Psiformer-based architecture designed for efficient forward-Laplacian evaluation <a class="reference internal" href="refs.html#li24" id="id18"><span>[Li24]</span></a></p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">transpsiformer</span></code>: a transferable extension of Psiformer that explicitly accounts for changes in the nuclear geometry <a class="reference internal" href="refs.html#schaetzle25" id="id19"><span>[Schaetzle25]</span></a></p></li>
</ul>
<p>The hyperparameters of such a predefined ansatz can also be overwritten at the command line:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span>$ deepqmc ansatz=psiformer ansatz.omni_factory.gnn_factory.n_interactions=2
</pre></div>
</div>
<p>For convenience the configuration of the <code class="docutils literal notranslate"><span class="pre">default</span></code> ansatz is reproduced here:</p>
<div class="highlight-yaml notranslate"><div class="highlight"><pre><span></span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.wf.NeuralNetworkWaveFunction</span>
<span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="nt">envelope</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.wf.env.ExponentialEnvelopes</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">isotropic</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">per_shell</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">per_orbital_exponent</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">spin_restricted</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">init_to_ones</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">softplus_zeta</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="nt">backflow_op</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.wf.nn_wave_function.BackflowOp</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">mult_act</span><span class="p">:</span><span class="w"> </span><span class="s">'${eval:"lambda</span><span class="nv"> </span><span class="s">x:</span><span class="nv"> </span><span class="s">x"}'</span>
<span class="nt">n_determinants</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">16</span>
<span class="nt">full_determinant</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="nt">cusp_electrons</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.wf.cusp.ElectronicCuspAsymptotic</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">same_scale</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">0.25</span>
<span class="w"> </span><span class="nt">anti_scale</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">0.5</span>
<span class="w"> </span><span class="nt">alpha</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">10.0</span>
<span class="w"> </span><span class="nt">trainable_alpha</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">cusp_function</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.wf.cusp.DeepQMCCusp</span>
<span class="nt">cusp_nuclei</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="nt">backflow_transform</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">mult</span>
<span class="nt">conf_coeff</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">haiku.Linear</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">with_bias</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">w_init</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">jax.numpy.ones</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="nt">omni_factory</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.wf.omni.OmniNet</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">embedding_dim</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">128</span>
<span class="w"> </span><span class="nt">jastrow_factory</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.wf.omni.Jastrow</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">sum_first</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">subnet_factory</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.hkext.MLP</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">hidden_layers</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">log</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">1</span>
<span class="w"> </span><span class="nt">bias</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">last_linear</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">activation</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">null</span>
<span class="w"> </span><span class="nt">init</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">default</span>
<span class="w"> </span><span class="nt">backflow_factory</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.wf.omni.Backflow</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">subnet_factory</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.hkext.MLP</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">hidden_layers</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">log</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">1</span>
<span class="w"> </span><span class="nt">bias</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">last_linear</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">activation</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">null</span>
<span class="w"> </span><span class="nt">init</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">default</span>
<span class="w"> </span><span class="nt">gnn_factory</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.ElectronGNN</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">n_interactions</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">3</span>
<span class="w"> </span><span class="nt">nuclei_embedding</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">null</span>
<span class="w"> </span><span class="nt">electron_embedding</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.electron_gnn.ElectronEmbedding</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">positional_embeddings</span><span class="p">:</span>
<span class="w"> </span><span class="nt">ne</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.edge_features.CombinedEdgeFeature</span>
<span class="w"> </span><span class="nt">features</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.edge_features.DistancePowerEdgeFeature</span>
<span class="w"> </span><span class="nt">powers</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">1</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.edge_features.DifferenceEdgeFeature</span>
<span class="w"> </span><span class="nt">use_spin</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">project_to_embedding_dim</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">two_particle_stream_dim</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">32</span>
<span class="w"> </span><span class="nt">self_interaction</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">edge_features</span><span class="p">:</span>
<span class="w"> </span><span class="nt">same</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.edge_features.CombinedEdgeFeature</span>
<span class="w"> </span><span class="nt">features</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.edge_features.DistancePowerEdgeFeature</span>
<span class="w"> </span><span class="nt">powers</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">1</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.edge_features.DifferenceEdgeFeature</span>
<span class="w"> </span><span class="nt">anti</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.edge_features.CombinedEdgeFeature</span>
<span class="w"> </span><span class="nt">features</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.edge_features.DistancePowerEdgeFeature</span>
<span class="w"> </span><span class="nt">powers</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">1</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.edge_features.DifferenceEdgeFeature</span>
<span class="w"> </span><span class="nt">layer_factory</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.electron_gnn.ElectronGNNLayer</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">subnet_factory</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.hkext.MLP</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">hidden_layers</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">log</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">2</span>
<span class="w"> </span><span class="nt">bias</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">last_linear</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">activation</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">jax.numpy.tanh</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">init</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">default</span>
<span class="w"> </span><span class="nt">subnet_factory_by_lbl</span><span class="p">:</span>
<span class="w"> </span><span class="nt">g</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.hkext.MLP</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">hidden_layers</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">log</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">1</span>
<span class="w"> </span><span class="nt">bias</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">last_linear</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">activation</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">jax.numpy.tanh</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">init</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">default</span>
<span class="w"> </span><span class="nt">electron_residual</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.hkext.ResidualConnection</span>
<span class="w"> </span><span class="nt">normalize</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">nucleus_residual</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">null</span>
<span class="w"> </span><span class="nt">two_particle_residual</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.hkext.ResidualConnection</span>
<span class="w"> </span><span class="nt">normalize</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">deep_features</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">shared</span>
<span class="w"> </span><span class="nt">update_rule</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">concatenate</span>
<span class="w"> </span><span class="nt">update_features</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.update_features.ResidualElectronUpdateFeature</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.update_features.NodeSumElectronUpdateFeature</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">node_types</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">up</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">down</span>
<span class="w"> </span><span class="nt">normalize</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.gnn.update_features.ConvolutionElectronUpdateFeature</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">edge_types</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">same</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">anti</span>
<span class="w"> </span><span class="nt">normalize</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">w_factory</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.hkext.MLP</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">hidden_layers</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">log</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">2</span>
<span class="w"> </span><span class="nt">bias</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">last_linear</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">activation</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">jax.numpy.tanh</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">init</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">default</span>
<span class="w"> </span><span class="nt">h_factory</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">deepqmc.hkext.MLP</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">hidden_layers</span><span class="p">:</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">log</span>
<span class="w"> </span><span class="p p-Indicator">-</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">2</span>
<span class="w"> </span><span class="nt">bias</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">last_linear</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">false</span>
<span class="w"> </span><span class="nt">activation</span><span class="p">:</span>
<span class="w"> </span><span class="nt">_target_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">jax.numpy.tanh</span>
<span class="w"> </span><span class="nt">_partial_</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">true</span>
<span class="w"> </span><span class="nt">init</span><span class="p">:</span><span class="w"> </span><span class="l l-Scalar l-Scalar-Plain">default</span>
</pre></div>
</div>
</section>
</section>
</section>
</article>
<footer class="prev-next-footer d-print-none">
<div class="prev-next-area">
<a class="left-prev"
href="installation.html"
title="previous page">
<i class="fa-solid fa-angle-left"></i>
<div class="prev-next-info">
<p class="prev-next-subtitle">previous</p>
<p class="prev-next-title">Installation</p>
</div>
</a>
<a class="right-next"
href="tutorial.html"
title="next page">
<div class="prev-next-info">
<p class="prev-next-subtitle">next</p>
<p class="prev-next-title">Tutorial</p>
</div>
<i class="fa-solid fa-angle-right"></i>
</a>
</div>
</footer>
</div>
</div>
<footer class="bd-footer-content">
</footer>
</main>
</div>
</div>
<!-- Scripts loaded after <body> so the DOM is not blocked -->
<script defer src="_static/scripts/bootstrap.js?digest=a95f357e85573c9b56d5"></script>
<script defer src="_static/scripts/pydata-sphinx-theme.js?digest=a95f357e85573c9b56d5"></script>
<footer class="bd-footer">
<div class="bd-footer__inner bd-page-width">
<div class="footer-items__start">
<div class="footer-item">
<p class="copyright">
© Copyright 2019-2026, Frank Noé and collaborators.
<br/>
</p>
</div>
</div>
<div class="footer-items__end">
<div class="footer-item">
<p class="theme-version">
<!-- # L10n: Setting the PST URL as an argument as this does not need to be localized -->
Built with the <a href="https://pydata-sphinx-theme.readthedocs.io/en/stable/index.html">PyData Sphinx Theme</a> 0.19.0.
</p></div>
</div>
</div>
</footer>
</body>
</html>