"""Compressed Sparse Row matrix format"""
from __future__ import division, print_function, absolute_import
__docformat__ = "restructuredtext en"
__all__ = ['csr_matrix', 'isspmatrix_csr']
import numpy as np
from scipy._lib.six import xrange
from .base import spmatrix
from ._sparsetools import csr_tocsc, csr_tobsr, csr_count_blocks, \
get_csr_submatrix, csr_sample_values
from .sputils import (upcast, isintlike, IndexMixin, issequence,
get_index_dtype, ismatrix)
from .compressed import _cs_matrix
class csr_matrix(_cs_matrix, IndexMixin):
"""
Compressed Sparse Row matrix
This can be instantiated in several ways:
csr_matrix(D)
with a dense matrix or rank-2 ndarray D
csr_matrix(S)
with another sparse matrix S (equivalent to S.tocsr())
csr_matrix((M, N), [dtype])
to construct an empty matrix with shape (M, N)
dtype is optional, defaulting to dtype='d'.
csr_matrix((data, (row_ind, col_ind)), [shape=(M, N)])
where ``data``, ``row_ind`` and ``col_ind`` satisfy the
relationship ``a[row_ind[k], col_ind[k]] = data[k]``.
csr_matrix((data, indices, indptr), [shape=(M, N)])
is the standard CSR representation where the column indices for
row i are stored in ``indices[indptr[i]:indptr[i+1]]`` and their
corresponding values are stored in ``data[indptr[i]:indptr[i+1]]``.
If the shape parameter is not supplied, the matrix dimensions
are inferred from the index arrays.
Attributes
----------
dtype : dtype
Data type of the matrix
shape : 2-tuple
Shape of the matrix
ndim : int
Number of dimensions (this is always 2)
nnz
Number of nonzero elements
data
CSR format data array of the matrix
indices
CSR format index array of the matrix
indptr
CSR format index pointer array of the matrix
has_sorted_indices
Whether indices are sorted
Notes
-----
Sparse matrices can be used in arithmetic operations: they support
addition, subtraction, multiplication, division, and matrix power.
Advantages of the CSR format
- efficient arithmetic operations CSR + CSR, CSR * CSR, etc.
- efficient row slicing
- fast matrix vector products
Disadvantages of the CSR format
- slow column slicing operations (consider CSC)
- changes to the sparsity structure are expensive (consider LIL or DOK)
Examples
--------
>>> import numpy as np
>>> from scipy.sparse import csr_matrix
>>> csr_matrix((3, 4), dtype=np.int8).toarray()
array([[0, 0, 0, 0],
[0, 0, 0, 0],
[0, 0, 0, 0]], dtype=int8)
>>> row = np.array([0, 0, 1, 2, 2, 2])
>>> col = np.array([0, 2, 2, 0, 1, 2])
>>> data = np.array([1, 2, 3, 4, 5, 6])
>>> csr_matrix((data, (row, col)), shape=(3, 3)).toarray()
array([[1, 0, 2],
[0, 0, 3],
[4, 5, 6]])
>>> indptr = np.array([0, 2, 3, 6])
>>> indices = np.array([0, 2, 2, 0, 1, 2])
>>> data = np.array([1, 2, 3, 4, 5, 6])
>>> csr_matrix((data, indices, indptr), shape=(3, 3)).toarray()
array([[1, 0, 2],
[0, 0, 3],
[4, 5, 6]])
As an example of how to construct a CSR matrix incrementally,
the following snippet builds a term-document matrix from texts:
>>> docs = [["hello", "world", "hello"], ["goodbye", "cruel", "world"]]
>>> indptr = [0]
>>> indices = []
>>> data = []
>>> vocabulary = {}
>>> for d in docs:
... for term in d:
... index = vocabulary.setdefault(term, len(vocabulary))
... indices.append(index)
... data.append(1)
... indptr.append(len(indices))
...
>>> csr_matrix((data, indices, indptr), dtype=int).toarray()
array([[2, 1, 0, 0],
[0, 1, 1, 1]])
"""
format = 'csr'
def transpose(self, axes=None, copy=False):
if axes is not None:
raise ValueError(("Sparse matrices do not support "
"an 'axes' parameter because swapping "
"dimensions is the only logical permutation."))
M, N = self.shape
from .csc import csc_matrix
return csc_matrix((self.data, self.indices,
self.indptr), shape=(N, M), copy=copy)
transpose.__doc__ = spmatrix.transpose.__doc__
def tolil(self, copy=False):
from .lil import lil_matrix
lil = lil_matrix(self.shape,dtype=self.dtype)
self.sum_duplicates()
ptr,ind,dat = self.indptr,self.indices,self.data
rows, data = lil.rows, lil.data
for n in xrange(self.shape[0]):
start = ptr[n]
end = ptr[n+1]
rows[n] = ind[start:end].tolist()
data[n] = dat[start:end].tolist()
return lil
tolil.__doc__ = spmatrix.tolil.__doc__
def tocsr(self, copy=False):
if copy:
return self.copy()
else:
return self
tocsr.__doc__ = spmatrix.tocsr.__doc__
def tocsc(self, copy=False):
idx_dtype = get_index_dtype((self.indptr, self.indices),
maxval=max(self.nnz, self.shape[0]))
indptr = np.empty(self.shape[1] + 1, dtype=idx_dtype)
indices = np.empty(self.nnz, dtype=idx_dtype)
data = np.empty(self.nnz, dtype=upcast(self.dtype))
csr_tocsc(self.shape[0], self.shape[1],
self.indptr.astype(idx_dtype),
self.indices.astype(idx_dtype),
self.data,
indptr,
indices,
data)
from .csc import csc_matrix
A = csc_matrix((data, indices, indptr), shape=self.shape)
A.has_sorted_indices = True
return A
tocsr.__doc__ = spmatrix.tocsr.__doc__
def tobsr(self, blocksize=None, copy=True):
from .bsr import bsr_matrix
if blocksize is None:
from .spfuncs import estimate_blocksize
return self.tobsr(blocksize=estimate_blocksize(self))
elif blocksize == (1,1):
arg1 = (self.data.reshape(-1,1,1),self.indices,self.indptr)
return bsr_matrix(arg1, shape=self.shape, copy=copy)
else:
R,C = blocksize
M,N = self.shape
if R < 1 or C < 1 or M % R != 0 or N % C != 0:
raise ValueError('invalid blocksize %s' % blocksize)
blks = csr_count_blocks(M,N,R,C,self.indptr,self.indices)
idx_dtype = get_index_dtype((self.indptr, self.indices),
maxval=max(N//C, blks))
indptr = np.empty(M//R+1, dtype=idx_dtype)
indices = np.empty(blks, dtype=idx_dtype)
data = np.zeros((blks,R,C), dtype=self.dtype)
csr_tobsr(M, N, R, C,
self.indptr.astype(idx_dtype),
self.indices.astype(idx_dtype),
self.data,
indptr, indices, data.ravel())
return bsr_matrix((data,indices,indptr), shape=self.shape)
tobsr.__doc__ = spmatrix.tobsr.__doc__
# these functions are used by the parent class (_cs_matrix)
# to remove redudancy between csc_matrix and csr_matrix
def _swap(self,x):
"""swap the members of x if this is a column-oriented matrix
"""
return (x[0],x[1])
def __getitem__(self, key):
def asindices(x):
try:
x = np.asarray(x)
# Check index contents, to avoid creating 64-bit arrays needlessly
idx_dtype = get_index_dtype((x,), check_contents=True)
if idx_dtype != x.dtype:
x = x.astype(idx_dtype)
except:
raise IndexError('invalid index')
else:
return x
def check_bounds(indices, N):
if indices.size == 0:
return (0, 0)
max_indx = indices.max()
if max_indx >= N:
raise IndexError('index (%d) out of range' % max_indx)
min_indx = indices.min()
if min_indx < -N:
raise IndexError('index (%d) out of range' % (N + min_indx))
return (min_indx,max_indx)
def extractor(indices,N):
"""Return a sparse matrix P so that P*self implements
slicing of the form self[[1,2,3],:]
"""
indices = asindices(indices)
(min_indx,max_indx) = check_bounds(indices,N)
if min_indx < 0:
indices = indices.copy()
indices[indices < 0] += N
indptr = np.arange(len(indices)+1, dtype=indices.dtype)
data = np.ones(len(indices), dtype=self.dtype)
shape = (len(indices),N)
return csr_matrix((data,indices,indptr), shape=shape)
row, col = self._unpack_index(key)
# First attempt to use original row optimized methods
# [1, ?]
if isintlike(row):
# [i, j]
if isintlike(col):
return self._get_single_element(row, col)
# [i, 1:2]
elif isinstance(col, slice):
return self._get_row_slice(row, col)
# [i, [1, 2]]
elif issequence(col):
P = extractor(col,self.shape[1]).T
return self[row, :] * P
elif isinstance(row, slice):
# [1:2,??]
if ((isintlike(col) and row.step in (1, None)) or
(isinstance(col, slice) and
col.step in (1, None) and
row.step in (1, None))):
# col is int or slice with step 1, row is slice with step 1.
return self._get_submatrix(row, col)
elif issequence(col):
# row is slice, col is sequence.
P = extractor(col,self.shape[1]).T # [1:2,[1,2]]
sliced = self
if row != slice(None, None, None):
sliced = sliced[row,:]
return sliced * P
elif issequence(row):
# [[1,2],??]
if isintlike(col) or isinstance(col,slice):
P = extractor(row, self.shape[0]) # [[1,2],j] or [[1,2],1:2]
extracted = P * self
if col == slice(None, None, None):
return extracted
else:
return extracted[:,col]
elif ismatrix(row) and issequence(col):
if len(row[0]) == 1 and isintlike(row[0][0]):
# [[[1],[2]], [1,2]], outer indexing
row = asindices(row)
P_row = extractor(row[:,0], self.shape[0])
P_col = extractor(col, self.shape[1]).T
return P_row * self * P_col
if not (issequence(col) and issequence(row)):
# Sample elementwise
row, col = self._index_to_arrays(row, col)
row = asindices(row)
col = asindices(col)
if row.shape != col.shape:
raise IndexError('number of row and column indices differ')
assert row.ndim <= 2
num_samples = np.size(row)
if num_samples == 0:
return csr_matrix(np.atleast_2d(row).shape, dtype=self.dtype)
check_bounds(row, self.shape[0])
check_bounds(col, self.shape[1])
val = np.empty(num_samples, dtype=self.dtype)
csr_sample_values(self.shape[0], self.shape[1],
self.indptr, self.indices, self.data,
num_samples, row.ravel(), col.ravel(), val)
if row.ndim == 1:
# row and col are 1d
return np.asmatrix(val)
return self.__class__(val.reshape(row.shape))
def getrow(self, i):
"""Returns a copy of row i of the matrix, as a (1 x n)
CSR matrix (row vector).
"""
return self._get_submatrix(i, slice(None))
def getcol(self, i):
"""Returns a copy of column i of the matrix, as a (m x 1)
CSR matrix (column vector).
"""
return self._get_submatrix(slice(None), i)
def _get_row_slice(self, i, cslice):
"""Returns a copy of row self[i, cslice]
"""
if i < 0:
i += self.shape[0]
if i < 0 or i >= self.shape[0]:
raise IndexError('index (%d) out of range' % i)
start, stop, stride = cslice.indices(self.shape[1])
if stride == 1:
# for stride == 1, _get_submatrix is ~30% faster than below
row_slice = self._get_submatrix(i, cslice)
else:
# other strides need new code
row_indices = self.indices[self.indptr[i]:self.indptr[i + 1]]
row_data = self.data[self.indptr[i]:self.indptr[i + 1]]
if stride > 0:
ind = (row_indices >= start) & (row_indices < stop)
elif stride < 0:
ind = (row_indices <= start) & (row_indices > stop)
if abs(stride) > 1:
ind = ind & ((row_indices - start) % stride == 0)
row_indices = (row_indices[ind] - start) // stride
row_data = row_data[ind]
row_indptr = np.array([0, len(row_indices)])
if stride < 0:
row_data = row_data[::-1]
row_indices = abs(row_indices[::-1])
shape = (1, int(np.ceil(float(stop - start) / stride)))
row_slice = csr_matrix((row_data, row_indices, row_indptr),
shape=shape)
return row_slice
def _get_submatrix(self, row_slice, col_slice):
"""Return a submatrix of this matrix (new matrix is created)."""
M,N = self.shape
def process_slice(sl, num):
if isinstance(sl, slice):
if sl.step not in (1, None):
raise ValueError('slicing with step != 1 not supported')
i0, i1 = sl.start, sl.stop
if i0 is None:
i0 = 0
elif i0 < 0:
i0 = num + i0
if i1 is None:
i1 = num
elif i1 < 0:
i1 = num + i1
return i0, i1
elif isintlike(sl):
if sl < 0:
sl += num
return sl, sl + 1
else:
raise TypeError('expected slice or scalar')
def check_bounds(i0, i1, num):
if not (0 <= i0 <= num) or not (0 <= i1 <= num) or not (i0 <= i1):
raise IndexError(
"index out of bounds: 0 <= %d <= %d, 0 <= %d <= %d,"
" %d <= %d" % (i0, num, i1, num, i0, i1))
i0, i1 = process_slice(row_slice, M)
j0, j1 = process_slice(col_slice, N)
check_bounds(i0, i1, M)
check_bounds(j0, j1, N)
indptr, indices, data = get_csr_submatrix(M, N,
self.indptr, self.indices, self.data,
int(i0), int(i1), int(j0), int(j1))
shape = (i1 - i0, j1 - j0)
return self.__class__((data,indices,indptr), shape=shape)
def isspmatrix_csr(x):
return isinstance(x, csr_matrix)