# Copyright 2022-2023 XProbe Inc.
# derived from copyright 1999-2021 Alibaba Group Holding Ltd.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import numpy as np
from ... import opcodes as OperandDef
from ...serialization.serializables import AnyField, TupleField
from .core import TensorArgReductionMixin, TensorReduction
class TensorArgmax(TensorReduction, TensorArgReductionMixin):
_op_type_ = OperandDef.ARGMAX
_func_name = "argmax"
_agg_func_name = "max"
_offset = AnyField("offset")
_total_shape = TupleField("total_shape")
def __init__(
self,
axis=None,
dtype=None,
combine_size=None,
offset=None,
total_shape=None,
stage=None,
**kw
):
if dtype is None:
dtype = np.dtype(int)
stage = self._rewrite_stage(stage)
super().__init__(
_axis=axis,
_combine_size=combine_size,
_offset=offset,
_total_shape=total_shape,
dtype=dtype,
stage=stage,
**kw
)
@property
def offset(self):
return getattr(self, "_offset", None)
@property
def total_shape(self):
return getattr(self, "_total_shape", None)
[docs]def argmax(a, axis=None, out=None, combine_size=None):
"""
Returns the indices of the maximum values along an axis.
Parameters
----------
a : array_like
Input tensor.
axis : int, optional
By default, the index is into the flattened tensor, otherwise
along the specified axis.
out : Tensor, optional
If provided, the result will be inserted into this tensor. It should
be of the appropriate shape and dtype.
combine_size: int, optional
The number of chunks to combine.
Returns
-------
index_array : Tensor of ints
Tensor of indices into the tensor. It has the same shape as `a.shape`
with the dimension along `axis` removed.
See Also
--------
Tensor.argmax, argmin
amax : The maximum value along a given axis.
unravel_index : Convert a flat index into an index tuple.
Notes
-----
In case of multiple occurrences of the maximum values, the indices
corresponding to the first occurrence are returned.
Examples
--------
>>> import mars.tensor as mt
>>> a = mt.arange(6).reshape(2,3)
>>> a.execute()
array([[0, 1, 2],
[3, 4, 5]])
>>> mt.argmax(a).execute()
5
>>> mt.argmax(a, axis=0).execute()
array([1, 1, 1])
>>> mt.argmax(a, axis=1).execute()
array([2, 2])
Indexes of the maximal elements of a N-dimensional tensor:
>>> ind = mt.unravel_index(mt.argmax(a, axis=None), a.shape)
>>> ind.execute()
(1, 2)
>>> a[ind].execute() # TODO(jisheng): accomplish when fancy index on tensor is supported
>>> b = mt.arange(6)
>>> b[1] = 5
>>> b.execute()
array([0, 5, 2, 3, 4, 5])
>>> mt.argmax(b).execute() # Only the first occurrence is returned.
1
"""
op = TensorArgmax(axis=axis, dtype=np.dtype(int), combine_size=combine_size)
return op(a, out=out)