E408 BUG: Inplace updating not working as expected with advanced indexing. · Issue #30732 · numpy/numpy · GitHub
[go: up one dir, main page]
Skip to content

BUG: Inplace updating not working as expected with advanced indexing. #30732

New issue
New issue
Open
Open
BUG: Inplace updating not working as expected with advanced indexing.#30732
Labels
00 - Bug
@rolfws

Description

@rolfws
rolfws
opened on Jan 27, 2026

Describe the issue:

I have an int64 array, mesh tuple (as constructed by ix_) and a boolean array.
The shape of the integer array indexed by the mesh is the same as the boolean array.
Then we want to add 1 to all places where the boolean array.

So:

>>> int_array[mesh][bool_array]
array([3])
>>> int_array[mesh][bool_array] += 1
>>> int_array[mesh][bool_array]
array([3])

In the example, I have tried multiple ways and the update seems inconsistent to me.

Reproduce the code example:

import numpy as np

int_array = np.array(
    [
        [1, 8, 12, 9],
        [3, 5, 10, 12],
        [3, 8, 8, 11],
        [0, 5, 14, 11],
        [0, 6, 6, 18],
        [0, 4, 12, 14],
        [2, 5, 8, 15],
        [3, 9, 7, 11],
        [2, 8, 7, 13],
        [3, 9, 8, 10],
        [2, 8, 9, 11],
        [0, 0, 0, 0],
        [3, 8, 8, 11],
        [0, 0, 0, 0],
        [2, 9, 8, 11],
        [0, 0, 0, 0],
        [2, 8, 9, 11],
        [3, 6, 12, 9],
        [0, 0, 0, 0],
        [3, 7, 7, 13],
        [3, 8, 8, 11],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [3, 7, 10, 10],
        [2, 8, 8, 12],
        [2, 8, 8, 12],
        [4, 8, 7, 11],
        [2, 6, 7, 15],
        [0, 0, 0, 0],
        [3, 7, 9, 11],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [2, 9, 7, 12],
        [0, 0, 0, 0],
        [0, 8, 8, 14],
        [4, 8, 7, 11],
        [2, 7, 7, 14],
        [3, 8, 7, 12],
        [0, 0, 0, 0],
        [2, 7, 10, 11],
        [3, 8, 8, 11],
        [0, 0, 0, 0],
        [3, 8, 8, 11],
        [3, 9, 7, 11],
        [0, 1, 10, 19],
        [0, 0, 0, 0],
        [4, 9, 7, 10],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [3, 8, 8, 11],
        [3, 7, 8, 12],
        [3, 7, 7, 13],
        [3, 7, 7, 13],
        [2, 6, 8, 14],
        [3, 6, 10, 11],
        [0, 0, 0, 0],
        [3, 8, 7, 12],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [3, 6, 9, 12],
        [0, 0, 0, 0],
        [4, 7, 8, 11],
        [3, 7, 10, 10],
        [0, 0, 0, 0],
        [3, 10, 7, 10],
        [2, 6, 6, 16],
        [0, 0, 0, 0],
        [2, 7, 8, 13],
        [0, 0, 0, 0],
        [3, 9, 7, 11],
        [0, 0, 0, 0],
        [0, 8, 8, 14],
        [0, 0, 0, 0],
        [0, 3, 5, 22],
        [3, 6, 9, 12],
        [0, 0, 0, 0],
        [4, 7, 8, 11],
        [3, 8, 8, 11],
        [0, 0, 0, 0],
        [3, 7, 8, 12],
        [3, 9, 8, 10],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [4, 8, 7, 11],
        [0, 4, 9, 17],
        [0, 0, 0, 0],
        [3, 6, 8, 13],
        [3, 8, 9, 10],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [2, 6, 7, 15],
        [3, 9, 7, 11],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [2, 8, 8, 12],
        [4, 7, 7, 12],
        [1, 7, 9, 13],
        [2, 6, 7, 15],
        [3, 8, 8, 11],
        [0, 0, 0, 0],
        [2, 7, 9, 12],
        [0, 0, 0, 0],
        [2, 6, 11, 11],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [3, 7, 7, 12],
        [4, 8, 8, 10],
        [3, 8, 8, 11],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [3, 8, 7, 12],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [3, 8, 7, 12],
        [2, 7, 6, 15],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [3, 8, 7, 12],
        [0, 4, 7, 19],
        [2, 6, 8, 14],
        [2, 8, 8, 12],
        [0, 0, 0, 0],
        [3, 7, 9, 11],
        [0, 0, 0, 0],
        [0, 7, 11, 12],
        [0, 0, 0, 0],
        [3, 8, 7, 12],
        [0, 0, 0, 0],
        [2, 8, 12, 8],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [2, 11, 9, 8],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [4, 8, 7, 11],
        [2, 8, 7, 13],
        [3, 8, 8, 11],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [3, 10, 7, 10],
        [2, 8, 10, 10],
        [2, 6, 9, 13],
        [4, 8, 6, 12],
        [2, 9, 7, 12],
        [2, 5, 7, 16],
        [0, 0, 0, 0],
        [0, 5, 14, 11],
        [3, 8, 8, 11],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [1, 8, 11, 10],
        [2, 8, 10, 10],
        [2, 7, 6, 15],
        [0, 0, 0, 0],
        [3, 8, 7, 12],
        [0, 0, 0, 0],
        [0, 6, 7, 17],
        [4, 11, 6, 9],
        [3, 6, 9, 12],
        [3, 7, 9, 11],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 4, 8, 18],
        [1, 7, 13, 9],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 4, 8, 18],
        [0, 0, 0, 0],
        [3, 8, 5, 14],
        [3, 8, 7, 12],
        [0, 0, 0, 0],
        [0, 3, 3, 24],
        [3, 7, 9, 11],
        [2, 7, 13, 8],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [3, 6, 7, 14],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [4, 8, 7, 11],
        [4, 8, 7, 11],
        [0, 0, 0, 0],
        [4, 1, 2, 23],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
        [0, 0, 0, 0],
    ],
    dtype=np.int64,
)
bool_array = np.array(
    [
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [True, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
        [False, False, False, False],
    ],
    dtype=bool,
)
mesh = (
    np.array(
        [
            [0],
            [1],
            [2],
            [3],
            [4],
            [5],
            [6],
            [7],
            [8],
            [9],
            [10],
            [12],
            [14],
            [16],
            [17],
            [19],
            [20],
            [23],
            [24],
            [25],
            [26],
            [27],
            [29],
            [32],
            [34],
            [35],
            [36],
            [37],
            [39],
            [40],
            [42],
            [43],
            [44],
            [46],
            [51],
            [52],
            [53],
            [54],
            [55],
            [56],
            [58],
            [62],
            [64],
            [65],
            [67],
            [68],
            [70],
            [72],
            [74],
            [76],
            [77],
            [80],
            [82],
            [83],
            [86],
            [87],
            [89],
            [90],
            [95],
            [96],
            [101],
            [102],
            [103],
            [104],
            [105],
            [107],
            [109],
            [112],
            [113],
            [114],
            [117],
            [122],
            [123],
            [127],
            [128],
            [129],
            [130],
            [132],
            [134],
            [136],
            [138],
            [141],
            [147],
            [148],
            [149],
            [154],
            [155],
            [156],
            [157],
            [158],
            [159],
            [161],
            [162],
            [165],
            [166],
            [167],
            [169],
            [171],
            [172],
            [173],
            [174],
            [177],
            [178],
            [181],
            [183],
            [184],
            [186],
            [187],
            [188],
            [196],
            [197],
            [199],
            [79],
            [191],
            [0],
            [1],
            [2],
            [3],
            [4],
            [5],
            [6],
            [7],
            [8],
            [9],
            [10],
            [12],
            [14],
            [16],
            [17],
            [19],
            [20],
            [23],
            [24],
            [25],
            [26],
            [27],
            [29],
            [32],
            [34],
            [35],
            [36],
            [37],
            [39],
            [40],
            [42],
            [43],
            [44],
            [46],
            [51],
            [52],
            [53],
            [54],
            [55],
            [56],
            [58],
            [62],
            [64],
            [65],
            [67],
            [68],
            [70],
            [72],
            [74],
            [76],
            [77],
            [79],
            [80],
            [82],
            [83],
            [86],
            [87],
            [89],
            [90],
            [95],
            [96],
            [101],
            [102],
            [103],
            [104],
            [105],
            [107],
            [109],
            [112],
            [113],
            [114],
            [117],
            [122],
            [123],
            [127],
            [128],
            [129],
            [130],
            [132],
            [134],
            [136],
            [138],
            [141],
            [147],
            [148],
            [149],
            [154],
            [155],
            [156],
            [157],
            [158],
            [159],
            [161],
            [162],
            [165],
            [166],
            [167],
            [169],
            [171],
            [172],
            [173],
            [174],
            [177],
            [178],
            [181],
            [183],
            [184],
            [186],
            [187],
            [188],
            [191],
            [196],
            [197],
            [199],
        ],
        dtype=int,
    ),
    np.array([[0, 1, 2, 3]], dtype=int),
)

print(int_array.shape, int_array.dtype)
print(bool_array.shape, bool_array.dtype)
print(mesh[0].shape, mesh[0].dtype)
print(mesh[1].shape, mesh[1].dtype)

# Does not work
print("double indexing inplace addition")
print(int_array[mesh][bool_array])
int_array[mesh][bool_array] += 1
print(int_array[mesh][bool_array])

print("double indexing overwrite")
print(int_array[mesh][bool_array])
int_array[mesh][bool_array] = 10
print(int_array[mesh][bool_array])

print("Addition with array of same shape")
print(int_array[mesh][bool_array])
int_array[mesh] += np.where(bool_array, 1, 0)
print(int_array[mesh][bool_array])

bool_array[0, 0] = True
# Does also not work
print("Changing the boolean array")
print(int_array[mesh][bool_array])
int_array[mesh][bool_array] = 0
print(int_array[mesh][bool_array])

# Works
print("Adding a (broadcasted) array works")
print(int_array[:5])
int_array[mesh] += np.array([100, 200, 300, 400])
print(int_array[:5])

Error message:

not applicable

Python and NumPy Versions:

I tried multiple versions with the same result:
2.4.1
3.13.9 (main, Nov 19 2025, 22:47:49) [Clang 21.1.4 ]

2.3.0
3.13.9 (main, Nov 19 2025, 22:47:49) [Clang 21.1.4 ]

2.2.0
3.13.9 (main, Nov 19 2025, 22:47:49) [Clang 21.1.4 ]

Runtime Environment:

2.4.1:
[{'numpy_version': '2.4.1',
'python': '3.13.9 (main, Nov 19 2025, 22:47:49) [Clang 21.1.4 ]',
'uname': uname_result(system='Linux', node='rolfDebianlt', release='6.12.63+deb13-amd64', version='#1 SMP PREEMPT_DYNAMIC Debian 6.12.63-1 (2025-12-30)', machine='x86_64')},
{'simd_extensions': {'baseline': ['X86_V2'],
'found': ['X86_V3'],
'not_found': ['X86_V4', 'AVX512_ICL', 'AVX512_SPR']}},
{'ignore_floating_point_errors_in_matmul': False},
{'architecture': 'Haswell',
'filepath': '/home/rolf/thunderstock/test/.venv/lib/python3.13/site-packages/numpy.libs/libscipy_openblas64_-fdde5778.so',
'internal_api': 'openblas',
'num_threads': 20,
'prefix': 'libscipy_openblas',
'threading_layer': 'pthreads',
'user_api': 'blas',
'version': '0.3.30'}]
2.3.0:
[{'numpy_version': '2.3.0',
'python': '3.13.9 (main, Nov 19 2025, 22:47:49) [Clang 21.1.4 ]',
'uname': uname_result(system='Linux', node='rolfDebianlt', release='6.12.63+deb13-amd64', version='#1 SMP PREEMPT_DYNAMIC Debian 6.12.63-1 (2025-12-30)', machine='x86_64')},
{'simd_extensions': {'baseline': ['SSE', 'SSE2', 'SSE3'], 6549
'found': ['SSSE3',
'SSE41',
'POPCNT',
'SSE42',
'AVX',
'F16C',
'FMA3',
'AVX2'],
'not_found': ['AVX512F',
'AVX512CD',
'AVX512_KNL',
'AVX512_KNM',
'AVX512_SKX',
'AVX512_CLX',
'AVX512_CNL',
'AVX512_ICL',
'AVX512_SPR']}},
{'architecture': 'Haswell',
'filepath': '/home/rolf/thunderstock/test/.venv/lib/python3.13/site-packages/numpy.libs/libscipy_openblas64_-56d6093b.so',
'internal_api': 'openblas',
'num_threads': 20,
'prefix': 'libscipy_openblas',
'threading_layer': 'pthreads',
'user_api': 'blas',
'version': '0.3.29'}]
2.2.0:
[{'numpy_version': '2.2.0',
'python': '3.13.9 (main, Nov 19 2025, 22:47:49) [Clang 21.1.4 ]',
'uname': uname_result(system='Linux', node='rolfDebianlt', release='6.12.63+deb13-amd64', version='#1 SMP PREEMPT_DYNAMIC Debian 6.12.63-1 (2025-12-30)', machine='x86_64')},
{'simd_extensions': {'baseline': ['SSE', 'SSE2', 'SSE3'],
'found': ['SSSE3',
'SSE41',
'POPCNT',
'SSE42',
'AVX',
'F16C',
'FMA3',
'AVX2'],
'not_found': ['AVX512F',
'AVX512CD',
'AVX512_KNL',
'AVX512_KNM',
'AVX512_SKX',
'AVX512_CLX',
'AVX512_CNL',
'AVX512_ICL']}},
{'architecture': 'Haswell',
'filepath': '/home/rolf/thunderstock/test/.venv/lib/python3.13/site-packages/numpy.libs/libscipy_openblas64_-6bb31eeb.so',
'internal_api': 'openblas',
'num_threads': 20,
'prefix': 'libscipy_openblas',
'threading_layer': 'pthreads',
'user_api': 'blas',
'version': '0.3.28'}]

How does this issue affect you or how did you find it:

The code does work for my colleague on his MacBook, but fails on my debian linux system, and also seems to fail on our cloud (AWS Lambda x86).
I found it because our lambda was timing out, and testing locally I could find an infinite loop.
The above operation was assumed to work, and it does work as expected till we found this specific example.
My colleague could not find the bug locally, as it was not reproducable on his machine, it was reproducable on mine.
So in this case we had an impact, but for now an easy work around is in place.
But the above structure might also be in place in other calculations, which do not create infinite loops, but gives wrong calculations.
So I cannot for sure say how much it does impact us right now.

Metadata

Metadata

Assignees

No one assigned

    Labels

    00 - Bug

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions
      0