def find(self):
"""
Return the sparse matrix in triple format (val,irow,jcol). If the
matrix data type is `None`, i.e., only the matrix sparsity pattern
is available, this method returns (irow,jcol).
"""
if self.dtype is not None:
return (self.values, self.irow, self.jcol)
return (self.irow, self.jcol)
if __name__ == '__main__':
import sys
fname = sys.argv[1]
A = MatrixMarketMatrix(fname)
print 'type: ', A.dtype
print 'symmetric, Hermitian, skew: ', A.symmetric, A.Hermitian, A.skewsym
print 'comments:'
print A.comments
from hsl.tools.spy import fast_spy
import matplotlib.pyplot as plt
fig = plt.figure()
fast_spy(A.nrow, A.ncol, A.irow, A.jcol,
sym=(A.symmetric or A.Hermitian or A.skewsym),
# val=A.values,
ax=fig.gca(),
)
plt.show()
if len(sys.argv) < 2:
sys.stderr.write('Supply input matrix as argument\n')
sys.exit(1)
fname = sys.argv[1]
# M = HarwellBoeingMatrix(fname, patternOnly=True, readRhs=False)
M = RutherfordBoeingData(fname, patternOnly=True, readRhs=False)
if M.nrow != M.ncol:
sys.stderr.write('Input matrix must be square\n')
sys.exit(1)
perm, nzdiag = nonzerodiag(M.nrow, M.ind, M.ip)
print 'Number of nonzeros on diagonal after reordering: ', nzdiag
try:
from hsl.tools.spy import fast_spy
import pylab
(_, irow, jcol) = M.find()
left = pylab.subplot(121)
fast_spy(M.nrow, M.ncol, irow, jcol, sym=M.issym, ax=left.get_axes())
right = pylab.subplot(122)
fast_spy(M.nrow, M.ncol, perm[irow], jcol, sym=M.issym, ax=right.get_axes())
pylab.show()
# pylab.savefig('mc21.pdf', bbox_inches='tight')
except:
pass
if len(sys.argv) < 2:
sys.stderr.write('Data file name must be supplied\n')
sys.exit(1)
fname = sys.argv[1]
#M = HarwellBoeingMatrix(fname, patternOnly=True, readRhs=False)
M = RutherfordBoeingData(fname, patternOnly=True, readRhs=False)
if M.nrow != M.ncol or not M.issym:
sys.stderr.write('Input matrix must be square and symmetric\n')
sys.exit(1)
# Compute reverse Cuthill-McKee ordering
perm, rinfo = rcmk(M.nrow, M.ind, M.ip)
# Or: Compute Sloan's ordering
#perm, rinfo = sloan(M.nrow, M.ind, M.ip)
# Plot original matrix
(_, irow, jcol) = M.find()
left = pylab.subplot(121)
fast_spy(M.nrow, M.ncol, irow, jcol, sym=M.issym,
ax=left.get_axes(), title='Original')
# Apply permutation and plot reordered matrix
right = pylab.subplot(122)
fast_spy(M.nrow, M.ncol, perm[irow], perm[jcol], sym=M.issym,
ax=right.get_axes(), title='Reordered')
pylab.show()
import matplotlib.pyplot as plt
A = LLSparseMatrix(mm_filename='bcsstk01.mtx',
itype=types.INT32_T, dtype=types.FLOAT64_T)
m, n = A.shape
irow, jcol, val = A.find()
A_csc = A.to_csc()
colptr, rowind, values = A_csc.get_numpy_arrays()
perm1, rinfo1 = rcmk(n, rowind, colptr) # Reverse Cuthill-McKee
perm2, rinfo2 = sloan(n, rowind, colptr) # Sloan's method
left = plt.subplot(131)
fast_spy(n, n, irow, jcol, sym=True,
ax=left.get_axes(), title='Original')
# Apply permutation 1 and plot reordered matrix
middle = plt.subplot(132)
fast_spy(n, n, perm1[irow], perm1[jcol], sym=True,
ax=middle.get_axes(), title='Rev. Cuthill-McKee (semibandwidth=%d)' % rinfo1[2])
# Apply permutation 2 and plot reordered matrix
right = plt.subplot(133)
fast_spy(n, n, perm2[irow], perm2[jcol], sym=True,
ax=right.get_axes(), title='Sloan (semibandwidth=%d)' % rinfo2[2])
plt.savefig('mc60_rcmk_sloan.png', bbox_inches='tight')
# plt.show()