p60975123
/
mindspore
forked from mindspore-Ecosystem/mindspore
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

!16052 fix numpy-native bugs 

From: @yanglf1121
Reviewed-by: @guoqi1024,@c_34
Signed-off-by: @c_34
This commit is contained in:
mindspore-ci-bot 2021-05-10 11:22:28 +08:00 committed by Gitee
parent cfc88d6fe6 56fbac9310
commit 889b9aab48
2 changed files with 13 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
mindspore/numpy/array_creations.py
View File

@ -83,7 +83,10 @@ def array(obj, dtype=None, copy=True, ndmin=0):
[1 2 3]
"""
res = asarray(obj, dtype)
if ndmin > res.ndim:
if res.size == 0:
_raise_value_error("Empty tensor cannot be expanded beyond the current dimension.")
res = _expand(res, ndmin)
if copy:
@ -253,6 +256,8 @@ def copy_(a):
"""
if not isinstance(a, Tensor):
a = asarray_const(a)
if a.size == 0:
return a
# The current implementation registers a new memory location for copied tensor by
# doing some reduandent operations.
origin_dtype = a.dtype

16
mindspore/numpy/math_ops.py
View File

@ -4335,7 +4335,7 @@ def searchsorted(a, v, side='left', sorter=None):
[0 5 1 2]
"""
if side not in ('left', 'right'):
_raise_value_error(f'{side} is an invalid value for keyword "side"')
_raise_value_error('invalid value for keyword "side"')
a = _to_tensor(a).astype(mstype.float32)
v = _to_tensor(v)
shape = F.shape(v)
@ -4705,14 +4705,14 @@ def histogram(a, bins=10, range=None, weights=None, density=False): # pylint: di
Examples:
>>> from mindspore import numpy as np
>>> print(np.histogram([1, 2, 1], bins=[0, 1, 2, 3]))
(Tensor(shape=[3], dtype=Int32, value= [0, 2, 1]),
(Tensor(shape=[3], dtype=Float32, value= [0, 2, 1]),
Tensor(shape=[4], dtype=Int32, value= [0, 1, 2, 3]))
>>> print(np.histogram(np.arange(4), bins=np.arange(5), density=True))
(Tensor(shape=[4], dtype=Float32, value=
[ 2.50000000e-01, 2.50000000e-01, 2.50000000e-01, 2.50000000e-01]),
Tensor(shape=[5], dtype=Int32, value= [0, 1, 2, 3, 4]))
>>> print(np.histogram([[1, 2, 1], [1, 0, 1]], bins=[0,1,2,3]))
(Tensor(shape=[3], dtype=Int32, value= [1, 4, 1]),
(Tensor(shape=[3], dtype=Float32, value= [1, 4, 1]),
Tensor(shape=[4], dtype=Int32, value= [0, 1, 2, 3]))
"""
a = _to_tensor(a).ravel()
@ -4726,7 +4726,7 @@ def histogram(a, bins=10, range=None, weights=None, density=False): # pylint: di
if density:
count = F.cast(count, mstype.float32)
count = count/diff(bin_edges)/F.reduce_sum(count)
return count.astype(mstype.int32), bin_edges
return count, bin_edges
@constexpr
@ -4794,7 +4794,7 @@ def histogramdd(sample, bins=10, range=None, weights=None, density=False): # pyl
[ 9 10 11]
[12 13 14]]
>>> print(np.histogramdd(sample, bins=(2, 3, 4)))
(Tensor(shape=[2, 3, 4], dtype=Int32, value=
(Tensor(shape=[2, 3, 4], dtype=Float32, value=
[[[1, 1, 0, 0],
[0, 0, 0, 0],
[0, 0, 0, 0]],
@ -4865,7 +4865,7 @@ def histogramdd(sample, bins=10, range=None, weights=None, density=False): # pyl
shape = _expanded_shape(ndim, dedges[i].size, i)
count /= _to_tensor(dedges[i]).reshape(shape)
count /= s
return count.astype(mstype.int32), bin_edges
return count, bin_edges
def histogram2d(x, y, bins=10, range=None, weights=None, density=False): # pylint: disable=redefined-builtin
@ -4917,7 +4917,7 @@ def histogram2d(x, y, bins=10, range=None, weights=None, density=False): # pylin
>>> x = np.arange(5)
>>> y = np.arange(2, 7)
>>> print(np.histogram2d(x, y, bins=(4, 6)))
(Tensor(shape=[4, 6], dtype=Int32, value=
(Tensor(shape=[4, 6], dtype=Float32, value=
[[1, 0, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0],
[0, 0, 0, 1, 0, 0]
@ -4929,7 +4929,7 @@ def histogram2d(x, y, bins=10, range=None, weights=None, density=False): # pylin
5.33333349e+00, 6.00000000e+00]))
"""
count, bin_edges = histogramdd((x, y), bins=bins, range=range, weights=weights, density=density)
return count.astype(mstype.int32), bin_edges[0], bin_edges[1]
return count, bin_edges[0], bin_edges[1]
def matrix_power(a, n):