-
Notifications
You must be signed in to change notification settings - Fork 8
/
Copy pathnep-0020-gufunc-signature-enhancement.html
848 lines (664 loc) · 53 KB
/
nep-0020-gufunc-signature-enhancement.html
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
<!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>NEP 20 — Expansion of generalized universal function signatures — NumPy Enhancement Proposals</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=8878045cc6db502f8baf" rel="stylesheet" />
<link href="_static/styles/pydata-sphinx-theme.css?digest=8878045cc6db502f8baf" rel="stylesheet" />
<link rel="stylesheet" type="text/css" href="_static/pygments.css?v=03e43079" />
<!-- So that users can add custom icons -->
<script src="_static/scripts/fontawesome.js?digest=8878045cc6db502f8baf"></script>
<!-- Pre-loaded scripts that we'll load fully later -->
<link rel="preload" as="script" href="_static/scripts/bootstrap.js?digest=8878045cc6db502f8baf" />
<link rel="preload" as="script" href="_static/scripts/pydata-sphinx-theme.js?digest=8878045cc6db502f8baf" />
<script src="_static/documentation_options.js?v=7f41d439"></script>
<script src="_static/doctools.js?v=888ff710"></script>
<script src="_static/sphinx_highlight.js?v=dc90522c"></script>
<script>DOCUMENTATION_OPTIONS.pagename = 'nep-0020-gufunc-signature-enhancement';</script>
<link rel="icon" href="_static/favicon.ico"/>
<link rel="index" title="Index" href="genindex.html" />
<link rel="search" title="Search" href="search.html" />
<link rel="next" title="NEP 22 — Duck typing for NumPy arrays – high level overview" href="nep-0022-ndarray-duck-typing-overview.html" />
<link rel="prev" title="NEP 19 — Random number generator policy" href="nep-0019-rng-policy.html" />
<meta name="viewport" content="width=device-width, initial-scale=1"/>
<meta name="docsearch:language" content="en"/>
<meta name="docsearch:version" content="" />
<meta name="docbuild:last-update" content="Jan 25, 2025"/>
</head>
<body data-bs-spy="scroll" data-bs-target=".bd-toc-nav" data-offset="180" data-bs-root-margin="0px 0px -60%" 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 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="content.html">
<img src="_static/numpylogo.svg" class="logo__image only-light" alt="NumPy Enhancement Proposals - Home"/>
<img src="_static/numpylogo_dark.svg" class="logo__image only-dark pst-js-only" alt="NumPy Enhancement Proposals - Home"/>
</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 current active">
<a class="nav-link nav-internal" href="index.html">
Index
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="scope.html">
The Scope of NumPy
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="roadmap.html">
Current roadmap
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-external" href="https://github.com/numpy/numpy/issues?q=is%3Aopen+is%3Aissue+label%3A%2223+-+Wish+List%22">
Wishlist
</a>
</li>
</ul>
</nav></div>
</div>
<div class="navbar-header-items__end">
<div class="navbar-item navbar-persistent--container">
<button class="btn search-button-field 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"></i>
<span class="search-button__default-text">Search</span>
<span class="search-button__kbd-shortcut"><kbd class="kbd-shortcut__modifier">Ctrl</kbd>+<kbd class="kbd-shortcut__modifier">K</kbd></span>
</button>
</div>
<div class="navbar-item">
<button class="btn btn-sm nav-link pst-navbar-icon theme-switch-button pst-js-only" aria-label="Color mode" data-bs-title="Color mode" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="theme-switch fa-solid fa-sun fa-lg" data-mode="light" title="Light"></i>
<i class="theme-switch fa-solid fa-moon fa-lg" data-mode="dark" title="Dark"></i>
<i class="theme-switch fa-solid fa-circle-half-stroke fa-lg" data-mode="auto" title="System Settings"></i>
</button></div>
<div class="navbar-item"><ul class="navbar-icon-links"
aria-label="Icon Links">
<li class="nav-item">
<a href="https://github.com/numpy/numpy" 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-square-github fa-lg" aria-hidden="true"></i>
<span class="sr-only">GitHub</span></a>
</li>
</ul></div>
</div>
</div>
<div class="navbar-persistent--mobile">
<button class="btn search-button-field 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"></i>
<span class="search-button__default-text">Search</span>
<span class="search-button__kbd-shortcut"><kbd class="kbd-shortcut__modifier">Ctrl</kbd>+<kbd class="kbd-shortcut__modifier">K</kbd></span>
</button>
</div>
<button class="pst-navbar-icon sidebar-toggle secondary-toggle" aria-label="On this page">
<span class="fa-solid fa-outdent"></span>
</button>
</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 current active">
<a class="nav-link nav-internal" href="index.html">
Index
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="scope.html">
The Scope of NumPy
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="roadmap.html">
Current roadmap
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-external" href="https://github.com/numpy/numpy/issues?q=is%3Aopen+is%3Aissue+label%3A%2223+-+Wish+List%22">
Wishlist
</a>
</li>
</ul>
</nav></div>
</div>
<div class="sidebar-header-items__end">
<div class="navbar-item">
<button class="btn btn-sm nav-link pst-navbar-icon theme-switch-button pst-js-only" aria-label="Color mode" data-bs-title="Color mode" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="theme-switch fa-solid fa-sun fa-lg" data-mode="light" title="Light"></i>
<i class="theme-switch fa-solid fa-moon fa-lg" data-mode="dark" title="Dark"></i>
<i class="theme-switch fa-solid fa-circle-half-stroke fa-lg" data-mode="auto" title="System Settings"></i>
</button></div>
<div class="navbar-item"><ul class="navbar-icon-links"
aria-label="Icon Links">
<li class="nav-item">
<a href="https://github.com/numpy/numpy" 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-square-github fa-lg" aria-hidden="true"></i>
<span class="sr-only">GitHub</span></a>
</li>
</ul></div>
</div>
</div>
<div class="sidebar-primary-items__start sidebar-primary__section">
<div class="sidebar-primary-item">
<nav class="bd-docs-nav bd-links"
aria-label="Section Navigation">
<p class="bd-links__title" role="heading" aria-level="1">Section Navigation</p>
<div class="bd-toc-item navbar-nav"><ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="scope.html">The Scope of NumPy</a></li>
<li class="toctree-l1"><a class="reference internal" href="roadmap.html">Current roadmap</a></li>
</ul>
<ul class="current nav bd-sidenav">
<li class="toctree-l1 has-children"><a class="reference internal" href="meta.html">Meta-NEPs (NEPs about NEPs or active Processes)</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="nep-0000.html">NEP 0 — Purpose and process</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0023-backwards-compatibility.html">NEP 23 — Backwards compatibility and deprecation policy</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0036-fair-play.html">NEP 36 — Fair play</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0045-c_style_guide.html">NEP 45 — C style guide</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0046-sponsorship-guidelines.html">NEP 46 — NumPy sponsorship guidelines</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0048-spending-project-funds.html">NEP 48 — Spending NumPy project funds</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-template.html">NEP X — Template and instructions</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="provisional.html">Provisional NEPs (provisionally accepted; interface may change)</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul class="simple">
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="accepted.html">Accepted NEPs (implementation in progress)</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="nep-0041-improved-dtype-support.html">NEP 41 — First step towards a new datatype system</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0042-new-dtypes.html">NEP 42 — New and extensible DTypes</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0044-restructuring-numpy-docs.html">NEP 44 — Restructuring the NumPy documentation</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0051-scalar-representation.html">NEP 51 — Changing the representation of NumPy scalars</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="open.html">Open NEPs (under consideration)</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="nep-0043-extensible-ufuncs.html">NEP 43 — Enhancing the extensibility of UFuncs</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0053-c-abi-evolution.html">NEP 53 — Evolving the NumPy C-API for NumPy 2.0</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0054-simd-cpp-highway.html">NEP 54 — SIMD infrastructure evolution: adopting Google Highway when moving to C++?</a></li>
</ul>
</details></li>
<li class="toctree-l1 current active has-children"><a class="reference internal" href="finished.html">Finished NEPs</a><details open="open"><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul class="current">
<li class="toctree-l2"><a class="reference internal" href="nep-0001-npy-format.html">NEP 1 — A simple file format for NumPy arrays</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0005-generalized-ufuncs.html">NEP 5 — Generalized universal functions</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0007-datetime-proposal.html">NEP 7 — A proposal for implementing some date/time types in NumPy</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0010-new-iterator-ufunc.html">NEP 10 — Optimizing iterator/UFunc performance</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0013-ufunc-overrides.html">NEP 13 — A mechanism for overriding Ufuncs</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0014-dropping-python2.7-proposal.html">NEP 14 — Plan for dropping Python 2.7 support</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0015-merge-multiarray-umath.html">NEP 15 — Merging multiarray and umath</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0018-array-function-protocol.html">NEP 18 — A dispatch mechanism for NumPy's high level array functions</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0019-rng-policy.html">NEP 19 — Random number generator policy</a></li>
<li class="toctree-l2 current active"><a class="current reference internal" href="#">NEP 20 — Expansion of generalized universal function signatures</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0022-ndarray-duck-typing-overview.html">NEP 22 — Duck typing for NumPy arrays – high level overview</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0027-zero-rank-arrarys.html">NEP 27 — Zero rank arrays</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0028-website-redesign.html">NEP 28 — numpy.org website redesign</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0029-deprecation_policy.html">NEP 29 — Recommend Python and NumPy version support as a community policy standard</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0032-remove-financial-functions.html">NEP 32 — Remove the financial functions from NumPy</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0034-infer-dtype-is-object.html">NEP 34 — Disallow inferring ``dtype=object`` from sequences</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0035-array-creation-dispatch-with-array-function.html">NEP 35 — Array creation dispatching with __array_function__</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0038-SIMD-optimizations.html">NEP 38 — Using SIMD optimization instructions for performance</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0040-legacy-datatype-impl.html">NEP 40 — Legacy datatype implementation in NumPy</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0049.html">NEP 49 — Data allocation strategies</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0050-scalar-promotion.html">NEP 50 — Promotion rules for Python scalars</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0052-python-api-cleanup.html">NEP 52 — Python API cleanup for NumPy 2.0</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0055-string_dtype.html">NEP 55 — Add a UTF-8 variable-width string DType to NumPy</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0056-array-api-main-namespace.html">NEP 56 — Array API standard support in NumPy's main namespace</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="deferred.html">Deferred and Superseded NEPs</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="nep-0002-warnfix.html">NEP 2 — A proposal to build numpy without warning with a big set of warning flags</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0003-math_config_clean.html">NEP 3 — Cleaning the math configuration of numpy.core</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0004-datetime-proposal3.html">NEP 4 — A (third) proposal for implementing some date/time types in NumPy</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0006-newbugtracker.html">NEP 6 — Replacing Trac with a different bug tracker</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0008-groupby_additions.html">NEP 8 — A proposal for adding groupby functionality to NumPy</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0009-structured_array_extensions.html">NEP 9 — Structured array extensions</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0011-deferred-ufunc-evaluation.html">NEP 11 — Deferred UFunc evaluation</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0012-missing-data.html">NEP 12 — Missing data functionality in NumPy</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0021-advanced-indexing.html">NEP 21 — Simplified and explicit advanced indexing</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0024-missing-data-2.html">NEP 24 — Missing data functionality - alternative 1 to NEP 12</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0025-missing-data-3.html">NEP 25 — NA support via special dtypes</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0026-missing-data-summary.html">NEP 26 — Summary of missing data NEPs and discussion</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0030-duck-array-protocol.html">NEP 30 — Duck typing for NumPy arrays - implementation</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0031-uarray.html">NEP 31 — Context-local and global overrides of the NumPy API</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0037-array-module.html">NEP 37 — A dispatch protocol for NumPy-like modules</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0047-array-api-standard.html">NEP 47 — Adopting the array API standard</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="rejected.html">Rejected and Withdrawn NEPs</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="nep-0016-abstract-array.html">NEP 16 — An abstract base class for identifying "duck arrays"</a></li>
<li class="toctree-l2"><a class="reference internal" href="nep-0017-split-out-maskedarray.html">NEP 17 — Split out masked arrays</a></li>
</ul>
</details></li>
</ul>
</div>
</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="content.html" class="nav-link" aria-label="Home">
<i class="fa-solid fa-home"></i>
</a>
</li>
<li class="breadcrumb-item"><a href="index.html" class="nav-link">Roadmap & NumPy enhancement proposals</a></li>
<li class="breadcrumb-item"><a href="finished.html" class="nav-link">Finished NEPs</a></li>
<li class="breadcrumb-item active" aria-current="page"><span class="ellipsis">NEP 20 — Expansion of generalized universal function signatures</span></li>
</ul>
</nav>
</div>
</div>
</div>
</div>
<div id="searchbox"></div>
<article class="bd-article">
<section id="nep-20-expansion-of-generalized-universal-function-signatures">
<span id="nep20"></span><h1>NEP 20 — Expansion of generalized universal function signatures<a class="headerlink" href="#nep-20-expansion-of-generalized-universal-function-signatures" title="Link to this heading">#</a></h1>
<dl class="field-list simple">
<dt class="field-odd">Author<span class="colon">:</span></dt>
<dd class="field-odd"><p>Marten van Kerkwijk <<a class="reference external" href="mailto:mhvk%40astro.utoronto.ca">mhvk<span>@</span>astro<span>.</span>utoronto<span>.</span>ca</a>></p>
</dd>
<dt class="field-even">Status<span class="colon">:</span></dt>
<dd class="field-even"><p>Final</p>
</dd>
<dt class="field-odd">Type<span class="colon">:</span></dt>
<dd class="field-odd"><p>Standards Track</p>
</dd>
<dt class="field-even">Created<span class="colon">:</span></dt>
<dd class="field-even"><p>2018-06-10</p>
</dd>
<dt class="field-odd">Resolution<span class="colon">:</span></dt>
<dd class="field-odd"><p><a class="reference external" href="https://mail.python.org/pipermail/numpy-discussion/2018-April/077959.html">https://mail.python.org/pipermail/numpy-discussion/2018-April/077959.html</a>,
<a class="reference external" href="https://mail.python.org/pipermail/numpy-discussion/2018-May/078078.html">https://mail.python.org/pipermail/numpy-discussion/2018-May/078078.html</a></p>
</dd>
</dl>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>The proposal to add fixed (i) and flexible (ii) dimensions
was accepted, while that to add broadcastable (iii) ones was deferred.</p>
</div>
<section id="abstract">
<h2>Abstract<a class="headerlink" href="#abstract" title="Link to this heading">#</a></h2>
<p>Generalized universal functions are, as their name indicates, generalization
of universal functions: they operate on non-scalar elements. Their signature
describes the structure of the elements they operate on, with names linking
dimensions of the operands that should be the same. Here, it is proposed to
extend the signature to allow the signature to indicate that a dimension (i)
has fixed size; (ii) can be absent; and (iii) can be broadcast.</p>
</section>
<section id="detailed-description">
<h2>Detailed description<a class="headerlink" href="#detailed-description" title="Link to this heading">#</a></h2>
<p>Each part of the proposal is driven by specific needs <a class="footnote-reference brackets" href="#id12" id="id1" role="doc-noteref"><span class="fn-bracket">[</span>1<span class="fn-bracket">]</span></a>.</p>
<ol class="arabic">
<li><p>Fixed-size dimensions. Code working with spatial vectors often explicitly
is for 2 or 3-dimensional space (e.g., the code from the <a class="reference external" href="http://www.iausofa.org/">Standards Of
Fundamental Astronomy</a>, which the author hopes
to wrap using gufuncs for astropy <a class="footnote-reference brackets" href="#id13" id="id2" role="doc-noteref"><span class="fn-bracket">[</span>2<span class="fn-bracket">]</span></a>). The signature should be able to
indicate that. E.g., the signature of a function that converts a polar
angle to a two-dimensional cartesian unit vector would currently have to be
<code class="docutils literal notranslate"><span class="pre">()->(n)</span></code>, with there being no way to indicate that <code class="docutils literal notranslate"><span class="pre">n</span></code> has to equal 2.
Indeed, this signature is particularly annoying since without putting in an
output argument, the current gufunc wrapper code fails because it cannot
determine <code class="docutils literal notranslate"><span class="pre">n</span></code>. Similarly, the signature for an cross product of two
3-dimensional vectors has to be <code class="docutils literal notranslate"><span class="pre">(n),(n)->(n)</span></code>, with again no way to
indicate that <code class="docutils literal notranslate"><span class="pre">n</span></code> has to equal 3. Hence, the proposal here to allow one
to give numerical values in addition to variable names. Thus, angle to
two-dimensional unit vector would be <code class="docutils literal notranslate"><span class="pre">()->(2)</span></code>; two angles to
three-dimensional unit vector <code class="docutils literal notranslate"><span class="pre">(),()->(3)</span></code>; and that for the cross
product of two three-dimensional vectors would be <code class="docutils literal notranslate"><span class="pre">(3),(3)->(3)</span></code>.</p></li>
<li><p>Possibly missing dimensions. This part is almost entirely driven by the
wish to wrap <code class="docutils literal notranslate"><span class="pre">matmul</span></code> in a gufunc. <code class="docutils literal notranslate"><span class="pre">matmul</span></code> stands for matrix
multiplication, and if it did only that, it could be covered with the
signature <code class="docutils literal notranslate"><span class="pre">(m,n),(n,p)->(m,p)</span></code>. However, it has special cases for when a
dimension is missing, allowing either argument to be treated as a single
vector, with the function thus becoming, effectively, vector-matrix,
matrix-vector, or vector-vector multiplication (but with no
broadcasting). To support this, it is suggested to allow postfixing a
dimension name with a question mark to indicate that the dimension does not
necessarily have to be present.</p>
<p>With this addition, the signature for <code class="docutils literal notranslate"><span class="pre">matmul</span></code> can be expressed as
<code class="docutils literal notranslate"><span class="pre">(m?,n),(n,p?)->(m?,p?)</span></code>. This indicates that if, e.g., the second
operand has only one dimension, for the purposes of the elementary function
it will be treated as if that input has core shape <code class="docutils literal notranslate"><span class="pre">(n,</span> <span class="pre">1)</span></code>, and the
output has the corresponding core shape of <code class="docutils literal notranslate"><span class="pre">(m,</span> <span class="pre">1)</span></code>. The actual output
array, however, has the flexible dimension removed, i.e., it will have
shape <code class="docutils literal notranslate"><span class="pre">(...,</span> <span class="pre">m)</span></code>. Similarly, if both arguments have only a single
dimension, the inputs will be presented as having shapes <code class="docutils literal notranslate"><span class="pre">(1,</span> <span class="pre">n)</span></code> and
<code class="docutils literal notranslate"><span class="pre">(n,</span> <span class="pre">1)</span></code> to the elementary function, and the output as <code class="docutils literal notranslate"><span class="pre">(1,</span> <span class="pre">1)</span></code>, while
the actual output array returned will have shape <code class="docutils literal notranslate"><span class="pre">()</span></code>. In this way, the
signature allows one to use a single elementary function for four related
but different signatures, <code class="docutils literal notranslate"><span class="pre">(m,n),(n,p)->(m,p)</span></code>, <code class="docutils literal notranslate"><span class="pre">(n),(n,p)->(p)</span></code>,
<code class="docutils literal notranslate"><span class="pre">(m,n),(n)->(m)</span></code> and <code class="docutils literal notranslate"><span class="pre">(n),(n)->()</span></code>.</p>
</li>
<li><p>Dimensions that can be broadcast. For some applications, broadcasting
between operands makes sense. For instance, an <code class="docutils literal notranslate"><span class="pre">all_equal</span></code> function that
compares vectors in arrays could have a signature <code class="docutils literal notranslate"><span class="pre">(n),(n)->()</span></code>, but this
forces both operands to be arrays, while it would be useful also to check
that, e.g., all parts of a vector are constant (maybe zero). The proposal
is to allow the implementer of a gufunc to indicate that a dimension can be
broadcast by post-fixing the dimension name with <code class="docutils literal notranslate"><span class="pre">|1</span></code>. Hence, the
signature for <code class="docutils literal notranslate"><span class="pre">all_equal</span></code> would become <code class="docutils literal notranslate"><span class="pre">(n|1),(n|1)->()</span></code>. The
signature seems handy more generally for “chained ufuncs”; e.g., another
application might be in a putative ufunc implementing <code class="docutils literal notranslate"><span class="pre">sumproduct</span></code>.</p>
<p>Another example that arose in the discussion, is of a weighted mean, which
might look like <code class="docutils literal notranslate"><span class="pre">weighted_mean(y,</span> <span class="pre">sigma[,</span> <span class="pre">axis,</span> <span class="pre">...])</span></code>, returning the
mean and its uncertainty. With a signature of <code class="docutils literal notranslate"><span class="pre">(n),(n)->(),()</span></code>, one
would be forced to always give as many sigmas as there are data points,
while broadcasting would allow one to give a single sigma for all points
(which is still useful to calculate the uncertainty on the mean).</p>
</li>
</ol>
</section>
<section id="implementation">
<h2>Implementation<a class="headerlink" href="#implementation" title="Link to this heading">#</a></h2>
<p>The proposed changes have all been implemented <a class="footnote-reference brackets" href="#id15" id="id3" role="doc-noteref"><span class="fn-bracket">[</span>3<span class="fn-bracket">]</span></a>, <a class="footnote-reference brackets" href="#id16" id="id4" role="doc-noteref"><span class="fn-bracket">[</span>4<span class="fn-bracket">]</span></a>, <a class="footnote-reference brackets" href="#id17" id="id5" role="doc-noteref"><span class="fn-bracket">[</span>5<span class="fn-bracket">]</span></a>. These PRs
extend the ufunc structure with two new fields, each of size equal to the
number of distinct dimensions, with <code class="docutils literal notranslate"><span class="pre">core_dim_sizes</span></code> holding possibly fixed
sizes, and <code class="docutils literal notranslate"><span class="pre">core_dim_flags</span></code> holding flags indicating whether a dimension can
be missing or broadcast. To ensure we can distinguish between this new
version and previous versions, an unused entry <code class="docutils literal notranslate"><span class="pre">reserved1</span></code> is repurposed as
a version number.</p>
<p>In the implementation, care is taken that to the elementary function flagged
dimensions are not treated any differently than non-flagged ones: for
instance, sizes of fixed-size dimensions are still passed on to the elementary
function (but the loop can now count on that size being equal to the fixed one
given in the signature).</p>
<p>An implementation detail to be decided upon is whether it might be handy to
have a summary of all flags. This could possibly be stored in <code class="docutils literal notranslate"><span class="pre">core_enabled</span></code>
(which currently is a bool), with non-zero continuing to indicate a gufunc,
but specific flags indicating whether or not a gufunc uses fixed, flexible, or
broadcastable dimensions.</p>
<p>With the above, the formal definition of the syntax would become <a class="footnote-reference brackets" href="#id16" id="id6" role="doc-noteref"><span class="fn-bracket">[</span>4<span class="fn-bracket">]</span></a>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="o"><</span><span class="n">Signature</span><span class="o">></span> <span class="p">:</span><span class="o">:=</span> <span class="o"><</span><span class="n">Input</span> <span class="n">arguments</span><span class="o">></span> <span class="s2">"->"</span> <span class="o"><</span><span class="n">Output</span> <span class="n">arguments</span><span class="o">></span>
<span class="o"><</span><span class="n">Input</span> <span class="n">arguments</span><span class="o">></span> <span class="p">:</span><span class="o">:=</span> <span class="o"><</span><span class="n">Argument</span> <span class="nb">list</span><span class="o">></span>
<span class="o"><</span><span class="n">Output</span> <span class="n">arguments</span><span class="o">></span> <span class="p">:</span><span class="o">:=</span> <span class="o"><</span><span class="n">Argument</span> <span class="nb">list</span><span class="o">></span>
<span class="o"><</span><span class="n">Argument</span> <span class="nb">list</span><span class="o">></span> <span class="p">:</span><span class="o">:=</span> <span class="n">nil</span> <span class="o">|</span> <span class="o"><</span><span class="n">Argument</span><span class="o">></span> <span class="o">|</span> <span class="o"><</span><span class="n">Argument</span><span class="o">></span> <span class="s2">","</span> <span class="o"><</span><span class="n">Argument</span> <span class="nb">list</span><span class="o">></span>
<span class="o"><</span><span class="n">Argument</span><span class="o">></span> <span class="p">:</span><span class="o">:=</span> <span class="s2">"("</span> <span class="o"><</span><span class="n">Core</span> <span class="n">dimension</span> <span class="nb">list</span><span class="o">></span> <span class="s2">")"</span>
<span class="o"><</span><span class="n">Core</span> <span class="n">dimension</span> <span class="nb">list</span><span class="o">></span> <span class="p">:</span><span class="o">:=</span> <span class="n">nil</span> <span class="o">|</span> <span class="o"><</span><span class="n">Core</span> <span class="n">dimension</span><span class="o">></span> <span class="o">|</span>
<span class="o"><</span><span class="n">Core</span> <span class="n">dimension</span><span class="o">></span> <span class="s2">","</span> <span class="o"><</span><span class="n">Core</span> <span class="n">dimension</span> <span class="nb">list</span><span class="o">></span>
<span class="o"><</span><span class="n">Core</span> <span class="n">dimension</span><span class="o">></span> <span class="p">:</span><span class="o">:=</span> <span class="o"><</span><span class="n">Dimension</span> <span class="n">name</span><span class="o">></span> <span class="o"><</span><span class="n">Dimension</span> <span class="n">modifier</span><span class="o">></span>
<span class="o"><</span><span class="n">Dimension</span> <span class="n">name</span><span class="o">></span> <span class="p">:</span><span class="o">:=</span> <span class="n">valid</span> <span class="n">Python</span> <span class="n">variable</span> <span class="n">name</span> <span class="o">|</span> <span class="n">valid</span> <span class="n">integer</span>
<span class="o"><</span><span class="n">Dimension</span> <span class="n">modifier</span><span class="o">></span> <span class="p">:</span><span class="o">:=</span> <span class="n">nil</span> <span class="o">|</span> <span class="s2">"|1"</span> <span class="o">|</span> <span class="s2">"?"</span>
</pre></div>
</div>
<ol class="arabic simple">
<li><p>All quotes are for clarity.</p></li>
<li><p>Unmodified core dimensions that share the same name must have the same size.
Each dimension name typically corresponds to one level of looping in the
elementary function’s implementation.</p></li>
<li><p>White spaces are ignored.</p></li>
<li><p>An integer as a dimension name freezes that dimension to the value.</p></li>
<li><p>If a name if suffixed with the <code class="docutils literal notranslate"><span class="pre">|1</span></code> modifier, it is allowed to broadcast
against other dimensions with the same name. All input dimensions
must share this modifier, while no output dimensions should have it.</p></li>
<li><p>If the name is suffixed with the <code class="docutils literal notranslate"><span class="pre">?</span></code> modifier, the dimension is a core
dimension only if it exists on all inputs and outputs that share it;
otherwise it is ignored (and replaced by a dimension of size 1 for the
elementary function).</p></li>
</ol>
<p>Examples of signatures <a class="footnote-reference brackets" href="#id16" id="id7" role="doc-noteref"><span class="fn-bracket">[</span>4<span class="fn-bracket">]</span></a>:</p>
<div class="pst-scrollable-table-container"><table class="table">
<tbody>
<tr class="row-odd"><td><p>Signature</p></td>
<td><p>Possible use</p></td>
</tr>
<tr class="row-even"><td><p><code class="docutils literal notranslate"><span class="pre">(),()->()</span></code></p></td>
<td><p>Addition</p></td>
</tr>
<tr class="row-odd"><td><p><code class="docutils literal notranslate"><span class="pre">(i)->()</span></code></p></td>
<td><p>Sum over last axis</p></td>
</tr>
<tr class="row-even"><td><p><code class="docutils literal notranslate"><span class="pre">(i|1),(i|1)->()</span></code></p></td>
<td><p>Test for equality along axis,
allowing comparison with a scalar</p></td>
</tr>
<tr class="row-odd"><td><p><code class="docutils literal notranslate"><span class="pre">(i),(i)->()</span></code></p></td>
<td><p>inner vector product</p></td>
</tr>
<tr class="row-even"><td><p><code class="docutils literal notranslate"><span class="pre">(m,n),(n,p)->(m,p)</span></code></p></td>
<td><p>matrix multiplication</p></td>
</tr>
<tr class="row-odd"><td><p><code class="docutils literal notranslate"><span class="pre">(n),(n,p)->(p)</span></code></p></td>
<td><p>vector-matrix multiplication</p></td>
</tr>
<tr class="row-even"><td><p><code class="docutils literal notranslate"><span class="pre">(m,n),(n)->(m)</span></code></p></td>
<td><p>matrix-vector multiplication</p></td>
</tr>
<tr class="row-odd"><td><p><code class="docutils literal notranslate"><span class="pre">(m?,n),(n,p?)->(m?,p?)</span></code></p></td>
<td><p>all four of the above at once,
except vectors cannot have loop
dimensions (ie, like <code class="docutils literal notranslate"><span class="pre">matmul</span></code>)</p></td>
</tr>
<tr class="row-even"><td><p><code class="docutils literal notranslate"><span class="pre">(3),(3)->(3)</span></code></p></td>
<td><p>cross product for 3-vectors</p></td>
</tr>
<tr class="row-odd"><td><p><code class="docutils literal notranslate"><span class="pre">(i,t),(j,t)->(i,j)</span></code></p></td>
<td><p>inner over the last dimension,
outer over the second to last,
and loop/broadcast over the rest.</p></td>
</tr>
</tbody>
</table>
</div>
</section>
<section id="backward-compatibility">
<h2>Backward compatibility<a class="headerlink" href="#backward-compatibility" title="Link to this heading">#</a></h2>
<p>One possible worry is the change in ufunc structure. For most applications,
which call <code class="docutils literal notranslate"><span class="pre">PyUFunc_FromDataAndSignature</span></code>, this is entirely transparent.
Furthermore, by repurposing <code class="docutils literal notranslate"><span class="pre">reserved1</span></code> as a version number, code compiled
against older versions of numpy will continue to work (though one will get a
warning upon import of that code with a newer version of numpy), except if
code explicitly changes the <code class="docutils literal notranslate"><span class="pre">reserved1</span></code> entry.</p>
</section>
<section id="alternatives">
<h2>Alternatives<a class="headerlink" href="#alternatives" title="Link to this heading">#</a></h2>
<p>It was suggested instead of extending the signature, to have multiple
dispatch, so that, e.g., <code class="docutils literal notranslate"><span class="pre">matmul</span></code> would simply have the multiple signatures
it supports, i.e., instead of <code class="docutils literal notranslate"><span class="pre">(m?,n),(n,p?)->(m?,p?)</span></code> one would have
<code class="docutils literal notranslate"><span class="pre">(m,n),(n,p)->(m,p)</span> <span class="pre">|</span> <span class="pre">(n),(n,p)->(p)</span> <span class="pre">|</span> <span class="pre">(m,n),(n)->(m)</span> <span class="pre">|</span> <span class="pre">(n),(n)->()</span></code>. A
disadvantage of this is that the developer now has to make sure that the
elementary function can deal with these different signatures. Furthermore,
the expansion quickly becomes cumbersome. For instance, for the <code class="docutils literal notranslate"><span class="pre">all_equal</span></code>
signature of <code class="docutils literal notranslate"><span class="pre">(n|1),(n|1)->()</span></code>, one would have to have five entries:
<code class="docutils literal notranslate"><span class="pre">(n),(n)->()</span> <span class="pre">|</span> <span class="pre">(n),(1)->()</span> <span class="pre">|</span> <span class="pre">(1),(n)->()</span> <span class="pre">|</span> <span class="pre">(n),()->()</span> <span class="pre">|</span> <span class="pre">(),(n)->()</span></code>. For
signatures like <code class="docutils literal notranslate"><span class="pre">(m|1,n|1,o|1),(m|1,n|1,o|1)->()</span></code> (from the <code class="docutils literal notranslate"><span class="pre">cube_equal</span></code>
test case in <a class="footnote-reference brackets" href="#id16" id="id8" role="doc-noteref"><span class="fn-bracket">[</span>4<span class="fn-bracket">]</span></a>), it is not even worth writing out the expansion.</p>
<p>For broadcasting, the alternative suffix of <code class="docutils literal notranslate"><span class="pre">^</span></code> was suggested (as
broadcasting can be thought of as increasing the size of the array). This
seems less clear. Furthermore, it was wondered whether it should not just be
an all-or-nothing flag. This could be the case, though given the postfix
for flexible dimensions, arguably another postfix is clearer (as is the
implementation).</p>
</section>
<section id="discussion">
<h2>Discussion<a class="headerlink" href="#discussion" title="Link to this heading">#</a></h2>
<p>The proposals here were discussed at fair length on the mailing list <a class="footnote-reference brackets" href="#id18" id="id9" role="doc-noteref"><span class="fn-bracket">[</span>6<span class="fn-bracket">]</span></a>,
<a class="footnote-reference brackets" href="#id19" id="id10" role="doc-noteref"><span class="fn-bracket">[</span>7<span class="fn-bracket">]</span></a>. The main points of contention were whether the use cases were
sufficiently strong. In particular, for frozen dimensions, it was argued that
checks on the right number could be put in loop selection code. This seems
much less clear for no benefit.</p>
<p>For broadcasting, the lack of examples of elementary functions that might need
it was noted, with it being questioned whether something like <code class="docutils literal notranslate"><span class="pre">all_equal</span></code>
was best done with a gufunc rather than as a special method on <code class="docutils literal notranslate"><span class="pre">np.equal</span></code>.
One counter-argument to this would be that there is an actual PR for
<code class="docutils literal notranslate"><span class="pre">all_equal</span></code> <a class="footnote-reference brackets" href="#id20" id="id11" role="doc-noteref"><span class="fn-bracket">[</span>8<span class="fn-bracket">]</span></a>. Another that even if one were to use a method, it would
be good to be able to express their signature (just as is possible at least
for <code class="docutils literal notranslate"><span class="pre">reduce</span></code> and <code class="docutils literal notranslate"><span class="pre">accumulate</span></code>).</p>
<p>A final argument was that we were making the gufuncs too complex. This
arguably holds for the dimensions that can be omitted, but that also has the
strongest use case. The frozen dimensions has a very simple implementation and
its meaning is obvious. The ability to broadcast is simple too, once the
flexible dimensions are supported.</p>
</section>
<section id="references-and-footnotes">
<h2>References and footnotes<a class="headerlink" href="#references-and-footnotes" title="Link to this heading">#</a></h2>
<aside class="footnote-list brackets">
<aside class="footnote brackets" id="id12" role="doc-footnote">
<span class="label"><span class="fn-bracket">[</span><a role="doc-backlink" href="#id1">1</a><span class="fn-bracket">]</span></span>
<p>Identified needs and suggestions for the implementation are not all by
the author. In particular, the suggestion for fixed dimensions and
initial implementation was by Jaime Frio (<a class="reference external" href="https://github.com/numpy/numpy/pull/5015">gh-5015</a>), the suggestion of <code class="docutils literal notranslate"><span class="pre">?</span></code>
to indicate dimensions can be omitted was by Nathaniel Smith, and the
initial implementation of that by Matti Picus (<a class="reference external" href="https://github.com/numpy/numpy/pull/11132">gh-11132</a>).</p>
</aside>
<aside class="footnote brackets" id="id13" role="doc-footnote">
<span class="label"><span class="fn-bracket">[</span><a role="doc-backlink" href="#id2">2</a><span class="fn-bracket">]</span></span>
<p><a class="reference external" href="https://github.com/astropy/astropy/pull/7502">wrap ERFA functions in gufuncs</a> (<a class="reference external" href="https://github.com/liberfa/erfa">ERFA</a>) is the less stringently licensed
version of <a class="reference external" href="http://www.iausofa.org/">Standards Of Fundamental Astronomy</a></p>
</aside>
<aside class="footnote brackets" id="id15" role="doc-footnote">
<span class="label"><span class="fn-bracket">[</span><a role="doc-backlink" href="#id3">3</a><span class="fn-bracket">]</span></span>
<p><a class="reference external" href="https://github.com/numpy/numpy/pull/11175">fixed-size and flexible dimensions</a></p>
</aside>
<aside class="footnote brackets" id="id16" role="doc-footnote">
<span class="label"><span class="fn-bracket">[</span>4<span class="fn-bracket">]</span></span>
<span class="backrefs">(<a role="doc-backlink" href="#id4">1</a>,<a role="doc-backlink" href="#id6">2</a>,<a role="doc-backlink" href="#id7">3</a>,<a role="doc-backlink" href="#id8">4</a>)</span>
<p><a class="reference external" href="https://github.com/numpy/numpy/pull/11179">broadcastable dimensions</a></p>
</aside>
<aside class="footnote brackets" id="id17" role="doc-footnote">
<span class="label"><span class="fn-bracket">[</span><a role="doc-backlink" href="#id5">5</a><span class="fn-bracket">]</span></span>
<p><a class="reference external" href="https://github.com/numpy/numpy/pull/11133">use in matmul</a></p>
</aside>
<aside class="footnote brackets" id="id18" role="doc-footnote">
<span class="label"><span class="fn-bracket">[</span><a role="doc-backlink" href="#id9">6</a><span class="fn-bracket">]</span></span>
<p>Discusses implementations for <code class="docutils literal notranslate"><span class="pre">matmul</span></code>:
<a class="reference external" href="https://mail.python.org/pipermail/numpy-discussion/2018-May/077972.html">https://mail.python.org/pipermail/numpy-discussion/2018-May/077972.html</a>,
<a class="reference external" href="https://mail.python.org/pipermail/numpy-discussion/2018-May/078021.html">https://mail.python.org/pipermail/numpy-discussion/2018-May/078021.html</a></p>
</aside>
<aside class="footnote brackets" id="id19" role="doc-footnote">
<span class="label"><span class="fn-bracket">[</span><a role="doc-backlink" href="#id10">7</a><span class="fn-bracket">]</span></span>
<p>Broadcasting:
<a class="reference external" href="https://mail.python.org/pipermail/numpy-discussion/2018-May/078078.html">https://mail.python.org/pipermail/numpy-discussion/2018-May/078078.html</a></p>
</aside>
<aside class="footnote brackets" id="id20" role="doc-footnote">
<span class="label"><span class="fn-bracket">[</span><a role="doc-backlink" href="#id11">8</a><span class="fn-bracket">]</span></span>
<p><a class="reference external" href="https://github.com/numpy/numpy/pull/8528">Logical gufuncs</a> (includes
<code class="docutils literal notranslate"><span class="pre">all_equal</span></code>)</p>
</aside>
</aside>
</section>
<section id="copyright">
<h2>Copyright<a class="headerlink" href="#copyright" title="Link to this heading">#</a></h2>
<p>This document has been placed in the public domain.</p>
</section>
</section>
</article>
</div>
<dialog id="pst-secondary-sidebar-modal"></dialog>
<div id="pst-secondary-sidebar" class="bd-sidebar-secondary bd-toc"><div class="sidebar-secondary-items sidebar-secondary__inner">
<div class="sidebar-secondary-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 class="bd-toc-nav page-toc" aria-labelledby="pst-page-navigation-heading-2">
<ul class="visible nav section-nav flex-column">
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#abstract">Abstract</a></li>
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#detailed-description">Detailed description</a></li>
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#implementation">Implementation</a></li>
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#backward-compatibility">Backward compatibility</a></li>
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#alternatives">Alternatives</a></li>
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#discussion">Discussion</a></li>
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#references-and-footnotes">References and footnotes</a></li>
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#copyright">Copyright</a></li>
</ul>
</nav></div>
</div></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=8878045cc6db502f8baf"></script>
<script defer src="_static/scripts/pydata-sphinx-theme.js?digest=8878045cc6db502f8baf"></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 2017-2025, NumPy Developers.
<br/>
</p>
</div>
<div class="footer-item">
<p class="sphinx-version">
Created using <a href="https://www.sphinx-doc.org/">Sphinx</a> 7.2.6.
<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> 0.16.1.
</p></div>
</div>
</div>
</footer>
</body>
</html>