507 lines
17 KiB
Python
507 lines
17 KiB
Python
|
"""Base class for sparse matrice with a .data attribute
|
||
|
|
|
||
|
|
subclasses must provide a _with_data() method that
|
||
|
|
creates a new matrix with the same sparsity pattern
|
||
|
|
as self but with a different data array
|
||
|
|
|
||
|
|
"""
|
||
|
|
|
||
|
|
import numpy as np
|
||
|
|
|
||
|
|
from ._base import _spbase, _ufuncs_with_fixed_point_at_zero
|
||
|
|
from ._sputils import isscalarlike, validateaxis
|
||
|
|
|
||
|
|
__all__ = []
|
||
|
|
|
||
|
|
|
||
|
|
# TODO implement all relevant operations
|
||
|
|
# use .data.__methods__() instead of /=, *=, etc.
|
||
|
|
class _data_matrix(_spbase):
|
||
|
|
def __init__(self):
|
||
|
|
_spbase.__init__(self)
|
||
|
|
|
||
|
|
@property
|
||
|
|
def dtype(self):
|
||
|
|
return self.data.dtype
|
||
|
|
|
||
|
|
@dtype.setter
|
||
|
|
def dtype(self, newtype):
|
||
|
|
self.data.dtype = newtype
|
||
|
|
|
||
|
|
def _deduped_data(self):
|
||
|
|
if hasattr(self, 'sum_duplicates'):
|
||
|
|
self.sum_duplicates()
|
||
|
|
return self.data
|
||
|
|
|
||
|
|
def __abs__(self):
|
||
|
|
return self._with_data(abs(self._deduped_data()))
|
||
|
|
|
||
|
|
def __round__(self, ndigits=0):
|
||
|
|
return self._with_data(np.around(self._deduped_data(), decimals=ndigits))
|
||
|
|
|
||
|
|
def _real(self):
|
||
|
|
return self._with_data(self.data.real)
|
||
|
|
|
||
|
|
def _imag(self):
|
||
|
|
return self._with_data(self.data.imag)
|
||
|
|
|
||
|
|
def __neg__(self):
|
||
|
|
if self.dtype.kind == 'b':
|
||
|
|
raise NotImplementedError('negating a boolean sparse array is not '
|
||
|
|
'supported')
|
||
|
|
return self._with_data(-self.data)
|
||
|
|
|
||
|
|
def __imul__(self, other): # self *= other
|
||
|
|
if isscalarlike(other):
|
||
|
|
self.data *= other
|
||
|
|
return self
|
||
|
|
else:
|
||
|
|
return NotImplemented
|
||
|
|
|
||
|
|
def __itruediv__(self, other): # self /= other
|
||
|
|
if isscalarlike(other):
|
||
|
|
recip = 1.0 / other
|
||
|
|
self.data *= recip
|
||
|
|
return self
|
||
|
|
else:
|
||
|
|
return NotImplemented
|
||
|
|
|
||
|
|
def astype(self, dtype, casting='unsafe', copy=True):
|
||
|
|
dtype = np.dtype(dtype)
|
||
|
|
if self.dtype != dtype:
|
||
|
|
matrix = self._with_data(
|
||
|
|
self.data.astype(dtype, casting=casting, copy=True),
|
||
|
|
copy=True
|
||
|
|
)
|
||
|
|
return matrix._with_data(matrix._deduped_data(), copy=False)
|
||
|
|
elif copy:
|
||
|
|
return self.copy()
|
||
|
|
else:
|
||
|
|
return self
|
||
|
|
|
||
|
|
astype.__doc__ = _spbase.astype.__doc__
|
||
|
|
|
||
|
|
def conjugate(self, copy=True):
|
||
|
|
if np.issubdtype(self.dtype, np.complexfloating):
|
||
|
|
return self._with_data(self.data.conjugate(), copy=copy)
|
||
|
|
elif copy:
|
||
|
|
return self.copy()
|
||
|
|
else:
|
||
|
|
return self
|
||
|
|
|
||
|
|
conjugate.__doc__ = _spbase.conjugate.__doc__
|
||
|
|
|
||
|
|
def copy(self):
|
||
|
|
return self._with_data(self.data.copy(), copy=True)
|
||
|
|
|
||
|
|
copy.__doc__ = _spbase.copy.__doc__
|
||
|
|
|
||
|
|
def count_nonzero(self):
|
||
|
|
return np.count_nonzero(self._deduped_data())
|
||
|
|
|
||
|
|
count_nonzero.__doc__ = _spbase.count_nonzero.__doc__
|
||
|
|
|
||
|
|
def power(self, n, dtype=None):
|
||
|
|
"""
|
||
|
|
This function performs element-wise power.
|
||
|
|
|
||
|
|
Parameters
|
||
|
|
----------
|
||
|
|
n : scalar
|
||
|
|
n is a non-zero scalar (nonzero avoids dense ones creation)
|
||
|
|
If zero power is desired, special case it to use `np.ones`
|
||
|
|
|
||
|
|
dtype : If dtype is not specified, the current dtype will be preserved.
|
||
|
|
|
||
|
|
Raises
|
||
|
|
------
|
||
|
|
NotImplementedError : if n is a zero scalar
|
||
|
|
If zero power is desired, special case it to use
|
||
|
|
`np.ones(A.shape, dtype=A.dtype)`
|
||
|
|
"""
|
||
|
|
if not isscalarlike(n):
|
||
|
|
raise NotImplementedError("input is not scalar")
|
||
|
|
if not n:
|
||
|
|
raise NotImplementedError(
|
||
|
|
"zero power is not supported as it would densify the matrix.\n"
|
||
|
|
"Use `np.ones(A.shape, dtype=A.dtype)` for this case."
|
||
|
|
)
|
||
|
|
|
||
|
|
data = self._deduped_data()
|
||
|
|
if dtype is not None:
|
||
|
|
data = data.astype(dtype)
|
||
|
|
return self._with_data(data ** n)
|
||
|
|
|
||
|
|
###########################
|
||
|
|
# Multiplication handlers #
|
||
|
|
###########################
|
||
|
|
|
||
|
|
def _mul_scalar(self, other):
|
||
|
|
return self._with_data(self.data * other)
|
||
|
|
|
||
|
|
|
||
|
|
# Add the numpy unary ufuncs for which func(0) = 0 to _data_matrix.
|
||
|
|
for npfunc in _ufuncs_with_fixed_point_at_zero:
|
||
|
|
name = npfunc.__name__
|
||
|
|
|
||
|
|
def _create_method(op):
|
||
|
|
def method(self):
|
||
|
|
result = op(self._deduped_data())
|
||
|
|
return self._with_data(result, copy=True)
|
||
|
|
|
||
|
|
method.__doc__ = (f"Element-wise {name}.\n\n"
|
||
|
|
f"See `numpy.{name}` for more information.")
|
||
|
|
method.__name__ = name
|
||
|
|
|
||
|
|
return method
|
||
|
|
|
||
|
|
setattr(_data_matrix, name, _create_method(npfunc))
|
||
|
|
|
||
|
|
|
||
|
|
def _find_missing_index(ind, n):
|
||
|
|
for k, a in enumerate(ind):
|
||
|
|
if k != a:
|
||
|
|
return k
|
||
|
|
|
||
|
|
k += 1
|
||
|
|
if k < n:
|
||
|
|
return k
|
||
|
|
else:
|
||
|
|
return -1
|
||
|
|
|
||
|
|
|
||
|
|
class _minmax_mixin:
|
||
|
|
"""Mixin for min and max methods.
|
||
|
|
|
||
|
|
These are not implemented for dia_matrix, hence the separate class.
|
||
|
|
"""
|
||
|
|
|
||
|
|
def _min_or_max_axis(self, axis, min_or_max):
|
||
|
|
N = self.shape[axis]
|
||
|
|
if N == 0:
|
||
|
|
raise ValueError("zero-size array to reduction operation")
|
||
|
|
M = self.shape[1 - axis]
|
||
|
|
idx_dtype = self._get_index_dtype(maxval=M)
|
||
|
|
|
||
|
|
mat = self.tocsc() if axis == 0 else self.tocsr()
|
||
|
|
mat.sum_duplicates()
|
||
|
|
|
||
|
|
major_index, value = mat._minor_reduce(min_or_max)
|
||
|
|
not_full = np.diff(mat.indptr)[major_index] < N
|
||
|
|
value[not_full] = min_or_max(value[not_full], 0)
|
||
|
|
|
||
|
|
mask = value != 0
|
||
|
|
major_index = np.compress(mask, major_index)
|
||
|
|
value = np.compress(mask, value)
|
||
|
|
|
||
|
|
if axis == 0:
|
||
|
|
return self._coo_container(
|
||
|
|
(value, (np.zeros(len(value), dtype=idx_dtype), major_index)),
|
||
|
|
dtype=self.dtype, shape=(1, M)
|
||
|
|
)
|
||
|
|
else:
|
||
|
|
return self._coo_container(
|
||
|
|
(value, (major_index, np.zeros(len(value), dtype=idx_dtype))),
|
||
|
|
dtype=self.dtype, shape=(M, 1)
|
||
|
|
)
|
||
|
|
|
||
|
|
def _min_or_max(self, axis, out, min_or_max):
|
||
|
|
if out is not None:
|
||
|
|
raise ValueError("Sparse arrays do not support an 'out' parameter.")
|
||
|
|
|
||
|
|
validateaxis(axis)
|
||
|
|
if self.ndim == 1:
|
||
|
|
if axis not in (None, 0, -1):
|
||
|
|
raise ValueError("axis out of range")
|
||
|
|
axis = None # avoid calling special axis case. no impact on 1d
|
||
|
|
|
||
|
|
if axis is None:
|
||
|
|
if 0 in self.shape:
|
||
|
|
raise ValueError("zero-size array to reduction operation")
|
||
|
|
|
||
|
|
zero = self.dtype.type(0)
|
||
|
|
if self.nnz == 0:
|
||
|
|
return zero
|
||
|
|
m = min_or_max.reduce(self._deduped_data().ravel())
|
||
|
|
if self.nnz != np.prod(self.shape):
|
||
|
|
m = min_or_max(zero, m)
|
||
|
|
return m
|
||
|
|
|
||
|
|
if axis < 0:
|
||
|
|
axis += 2
|
||
|
|
|
||
|
|
if (axis == 0) or (axis == 1):
|
||
|
|
return self._min_or_max_axis(axis, min_or_max)
|
||
|
|
else:
|
||
|
|
raise ValueError("axis out of range")
|
||
|
|
|
||
|
|
def _arg_min_or_max_axis(self, axis, argmin_or_argmax, compare):
|
||
|
|
if self.shape[axis] == 0:
|
||
|
|
raise ValueError("Cannot apply the operation along a zero-sized dimension.")
|
||
|
|
|
||
|
|
if axis < 0:
|
||
|
|
axis += 2
|
||
|
|
|
||
|
|
zero = self.dtype.type(0)
|
||
|
|
|
||
|
|
mat = self.tocsc() if axis == 0 else self.tocsr()
|
||
|
|
mat.sum_duplicates()
|
||
|
|
|
||
|
|
ret_size, line_size = mat._swap(mat.shape)
|
||
|
|
ret = np.zeros(ret_size, dtype=int)
|
||
|
|
|
||
|
|
nz_lines, = np.nonzero(np.diff(mat.indptr))
|
||
|
|
for i in nz_lines:
|
||
|
|
p, q = mat.indptr[i:i + 2]
|
||
|
|
data = mat.data[p:q]
|
||
|
|
indices = mat.indices[p:q]
|
||
|
|
extreme_index = argmin_or_argmax(data)
|
||
|
|
extreme_value = data[extreme_index]
|
||
|
|
if compare(extreme_value, zero) or q - p == line_size:
|
||
|
|
ret[i] = indices[extreme_index]
|
||
|
|
else:
|
||
|
|
zero_ind = _find_missing_index(indices, line_size)
|
||
|
|
if extreme_value == zero:
|
||
|
|
ret[i] = min(extreme_index, zero_ind)
|
||
|
|
else:
|
||
|
|
ret[i] = zero_ind
|
||
|
|
|
||
|
|
if axis == 1:
|
||
|
|
ret = ret.reshape(-1, 1)
|
||
|
|
|
||
|
|
return self._ascontainer(ret)
|
||
|
|
|
||
|
|
def _arg_min_or_max(self, axis, out, argmin_or_argmax, compare):
|
||
|
|
if out is not None:
|
||
|
|
raise ValueError("Sparse types do not support an 'out' parameter.")
|
||
|
|
|
||
|
|
validateaxis(axis)
|
||
|
|
|
||
|
|
if self.ndim == 1:
|
||
|
|
if axis not in (None, 0, -1):
|
||
|
|
raise ValueError("axis out of range")
|
||
|
|
axis = None # avoid calling special axis case. no impact on 1d
|
||
|
|
|
||
|
|
if axis is not None:
|
||
|
|
return self._arg_min_or_max_axis(axis, argmin_or_argmax, compare)
|
||
|
|
|
||
|
|
if 0 in self.shape:
|
||
|
|
raise ValueError("Cannot apply the operation to an empty matrix.")
|
||
|
|
|
||
|
|
if self.nnz == 0:
|
||
|
|
return 0
|
||
|
|
|
||
|
|
zero = self.dtype.type(0)
|
||
|
|
mat = self.tocoo()
|
||
|
|
# Convert to canonical form: no duplicates, sorted indices.
|
||
|
|
mat.sum_duplicates()
|
||
|
|
extreme_index = argmin_or_argmax(mat.data)
|
||
|
|
extreme_value = mat.data[extreme_index]
|
||
|
|
num_col = mat.shape[-1]
|
||
|
|
|
||
|
|
# If the min value is less than zero, or max is greater than zero,
|
||
|
|
# then we do not need to worry about implicit zeros.
|
||
|
|
if compare(extreme_value, zero):
|
||
|
|
# cast to Python int to avoid overflow and RuntimeError
|
||
|
|
return int(mat.row[extreme_index]) * num_col + int(mat.col[extreme_index])
|
||
|
|
|
||
|
|
# Cheap test for the rare case where we have no implicit zeros.
|
||
|
|
size = np.prod(self.shape)
|
||
|
|
if size == mat.nnz:
|
||
|
|
return int(mat.row[extreme_index]) * num_col + int(mat.col[extreme_index])
|
||
|
|
|
||
|
|
# At this stage, any implicit zero could be the min or max value.
|
||
|
|
# After sum_duplicates(), the `row` and `col` arrays are guaranteed to
|
||
|
|
# be sorted in C-order, which means the linearized indices are sorted.
|
||
|
|
linear_indices = mat.row * num_col + mat.col
|
||
|
|
first_implicit_zero_index = _find_missing_index(linear_indices, size)
|
||
|
|
if extreme_value == zero:
|
||
|
|
return min(first_implicit_zero_index, extreme_index)
|
||
|
|
return first_implicit_zero_index
|
||
|
|
|
||
|
|
def max(self, axis=None, out=None):
|
||
|
|
"""
|
||
|
|
Return the maximum of the array/matrix or maximum along an axis.
|
||
|
|
This takes all elements into account, not just the non-zero ones.
|
||
|
|
|
||
|
|
Parameters
|
||
|
|
----------
|
||
|
|
axis : {-2, -1, 0, 1, None} optional
|
||
|
|
Axis along which the sum is computed. The default is to
|
||
|
|
compute the maximum over all elements, returning
|
||
|
|
a scalar (i.e., `axis` = `None`).
|
||
|
|
|
||
|
|
out : None, optional
|
||
|
|
This argument is in the signature *solely* for NumPy
|
||
|
|
compatibility reasons. Do not pass in anything except
|
||
|
|
for the default value, as this argument is not used.
|
||
|
|
|
||
|
|
Returns
|
||
|
|
-------
|
||
|
|
amax : coo_matrix or scalar
|
||
|
|
Maximum of `a`. If `axis` is None, the result is a scalar value.
|
||
|
|
If `axis` is given, the result is a sparse.coo_matrix of dimension
|
||
|
|
``a.ndim - 1``.
|
||
|
|
|
||
|
|
See Also
|
||
|
|
--------
|
||
|
|
min : The minimum value of a sparse array/matrix along a given axis.
|
||
|
|
numpy.matrix.max : NumPy's implementation of 'max' for matrices
|
||
|
|
|
||
|
|
"""
|
||
|
|
return self._min_or_max(axis, out, np.maximum)
|
||
|
|
|
||
|
|
def min(self, axis=None, out=None):
|
||
|
|
"""
|
||
|
|
Return the minimum of the array/matrix or maximum along an axis.
|
||
|
|
This takes all elements into account, not just the non-zero ones.
|
||
|
|
|
||
|
|
Parameters
|
||
|
|
----------
|
||
|
|
axis : {-2, -1, 0, 1, None} optional
|
||
|
|
Axis along which the sum is computed. The default is to
|
||
|
|
compute the minimum over all elements, returning
|
||
|
|
a scalar (i.e., `axis` = `None`).
|
||
|
|
|
||
|
|
out : None, optional
|
||
|
|
This argument is in the signature *solely* for NumPy
|
||
|
|
compatibility reasons. Do not pass in anything except for
|
||
|
|
the default value, as this argument is not used.
|
||
|
|
|
||
|
|
Returns
|
||
|
|
-------
|
||
|
|
amin : coo_matrix or scalar
|
||
|
|
Minimum of `a`. If `axis` is None, the result is a scalar value.
|
||
|
|
If `axis` is given, the result is a sparse.coo_matrix of dimension
|
||
|
|
``a.ndim - 1``.
|
||
|
|
|
||
|
|
See Also
|
||
|
|
--------
|
||
|
|
max : The maximum value of a sparse array/matrix along a given axis.
|
||
|
|
numpy.matrix.min : NumPy's implementation of 'min' for matrices
|
||
|
|
|
||
|
|
"""
|
||
|
|
return self._min_or_max(axis, out, np.minimum)
|
||
|
|
|
||
|
|
def nanmax(self, axis=None, out=None):
|
||
|
|
"""
|
||
|
|
Return the maximum of the array/matrix or maximum along an axis, ignoring any
|
||
|
|
NaNs. This takes all elements into account, not just the non-zero
|
||
|
|
ones.
|
||
|
|
|
||
|
|
.. versionadded:: 1.11.0
|
||
|
|
|
||
|
|
Parameters
|
||
|
|
----------
|
||
|
|
axis : {-2, -1, 0, 1, None} optional
|
||
|
|
Axis along which the maximum is computed. The default is to
|
||
|
|
compute the maximum over all elements, returning
|
||
|
|
a scalar (i.e., `axis` = `None`).
|
||
|
|
|
||
|
|
out : None, optional
|
||
|
|
This argument is in the signature *solely* for NumPy
|
||
|
|
compatibility reasons. Do not pass in anything except
|
||
|
|
for the default value, as this argument is not used.
|
||
|
|
|
||
|
|
Returns
|
||
|
|
-------
|
||
|
|
amax : coo_matrix or scalar
|
||
|
|
Maximum of `a`. If `axis` is None, the result is a scalar value.
|
||
|
|
If `axis` is given, the result is a sparse.coo_matrix of dimension
|
||
|
|
``a.ndim - 1``.
|
||
|
|
|
||
|
|
See Also
|
||
|
|
--------
|
||
|
|
nanmin : The minimum value of a sparse array/matrix along a given axis,
|
||
|
|
ignoring NaNs.
|
||
|
|
max : The maximum value of a sparse array/matrix along a given axis,
|
||
|
|
propagating NaNs.
|
||
|
|
numpy.nanmax : NumPy's implementation of 'nanmax'.
|
||
|
|
|
||
|
|
"""
|
||
|
|
return self._min_or_max(axis, out, np.fmax)
|
||
|
|
|
||
|
|
def nanmin(self, axis=None, out=None):
|
||
|
|
"""
|
||
|
|
Return the minimum of the array/matrix or minimum along an axis, ignoring any
|
||
|
|
NaNs. This takes all elements into account, not just the non-zero
|
||
|
|
ones.
|
||
|
|
|
||
|
|
.. versionadded:: 1.11.0
|
||
|
|
|
||
|
|
Parameters
|
||
|
|
----------
|
||
|
|
axis : {-2, -1, 0, 1, None} optional
|
||
|
|
Axis along which the minimum is computed. The default is to
|
||
|
|
compute the minimum over all elements, returning
|
||
|
|
a scalar (i.e., `axis` = `None`).
|
||
|
|
|
||
|
|
out : None, optional
|
||
|
|
This argument is in the signature *solely* for NumPy
|
||
|
|
compatibility reasons. Do not pass in anything except for
|
||
|
|
the default value, as this argument is not used.
|
||
|
|
|
||
|
|
Returns
|
||
|
|
-------
|
||
|
|
amin : coo_matrix or scalar
|
||
|
|
Minimum of `a`. If `axis` is None, the result is a scalar value.
|
||
|
|
If `axis` is given, the result is a sparse.coo_matrix of dimension
|
||
|
|
``a.ndim - 1``.
|
||
|
|
|
||
|
|
See Also
|
||
|
|
--------
|
||
|
|
nanmax : The maximum value of a sparse array/matrix along a given axis,
|
||
|
|
ignoring NaNs.
|
||
|
|
min : The minimum value of a sparse array/matrix along a given axis,
|
||
|
|
propagating NaNs.
|
||
|
|
numpy.nanmin : NumPy's implementation of 'nanmin'.
|
||
|
|
|
||
|
|
"""
|
||
|
|
return self._min_or_max(axis, out, np.fmin)
|
||
|
|
|
||
|
|
def argmax(self, axis=None, out=None):
|
||
|
|
"""Return indices of maximum elements along an axis.
|
||
|
|
|
||
|
|
Implicit zero elements are also taken into account. If there are
|
||
|
|
several maximum values, the index of the first occurrence is returned.
|
||
|
|
|
||
|
|
Parameters
|
||
|
|
----------
|
||
|
|
axis : {-2, -1, 0, 1, None}, optional
|
||
|
|
Axis along which the argmax is computed. If None (default), index
|
||
|
|
of the maximum element in the flatten data is returned.
|
||
|
|
out : None, optional
|
||
|
|
This argument is in the signature *solely* for NumPy
|
||
|
|
compatibility reasons. Do not pass in anything except for
|
||
|
|
the default value, as this argument is not used.
|
||
|
|
|
||
|
|
Returns
|
||
|
|
-------
|
||
|
|
ind : numpy.matrix or int
|
||
|
|
Indices of maximum elements. If matrix, its size along `axis` is 1.
|
||
|
|
"""
|
||
|
|
return self._arg_min_or_max(axis, out, np.argmax, np.greater)
|
||
|
|
|
||
|
|
def argmin(self, axis=None, out=None):
|
||
|
|
"""Return indices of minimum elements along an axis.
|
||
|
|
|
||
|
|
Implicit zero elements are also taken into account. If there are
|
||
|
|
several minimum values, the index of the first occurrence is returned.
|
||
|
|
|
||
|
|
Parameters
|
||
|
|
----------
|
||
|
|
axis : {-2, -1, 0, 1, None}, optional
|
||
|
|
Axis along which the argmin is computed. If None (default), index
|
||
|
|
of the minimum element in the flatten data is returned.
|
||
|
|
out : None, optional
|
||
|
|
This argument is in the signature *solely* for NumPy
|
||
|
|
compatibility reasons. Do not pass in anything except for
|
||
|
|
the default value, as this argument is not used.
|
||
|
|
|
||
|
|
Returns
|
||
|
|
-------
|
||
|
|
ind : numpy.matrix or int
|
||
|
|
Indices of minimum elements. If matrix, its size along `axis` is 1.
|
||
|
|
"""
|
||
|
|
return self._arg_min_or_max(axis, out, np.argmin, np.less)
|