def __init__(self, opt=None, method='cns'):
"""Initialise ConvBPDN dictionary learning algorithm options.
Valid values for parameter ``method`` are documented in function
:func:`.ConvCnstrMOD`.
"""
self.defaults.update({
'CCMOD':
copy.deepcopy(
ccmod.ConvCnstrMODOptions(method=method).defaults)
})
dictlrn.DictLearn.Options.__init__(
self, {
'CBPDN':
cbpdn.ConvBPDN.Options({
'MaxMainIter': 1,
'AutoRho': {
'Period': 10,
'AutoScaling': False,
'RsdlRatio': 10.0,
'Scaling': 2.0,
'RsdlTarget': 1.0
}
}),
'CCMOD':
ccmod.ConvCnstrMODOptions(
{
'MaxMainIter': 1,
'AutoRho': {
'Period': 10,
'AutoScaling': False,
'RsdlRatio': 10.0,
'Scaling': 2.0,
'RsdlTarget': 1.0
}
},
method=method)
})
if opt is None:
opt = {}
self.update(opt)
def test_08(self):
N = 16
M = 4
Nd = 8
X = np.random.randn(N, N, 1, 1, M)
S = np.random.randn(N, N, 1)
dt = np.float64
opt = ccmod.ConvCnstrMODOptions(
{'Verbose': False, 'MaxMainIter': 20,
'AutoRho': {'Enabled': True},
'DataType': dt})
c = ccmod.ConvCnstrMOD(X, S, (Nd, Nd, M), opt=opt)
c.solve()
assert(c.X.dtype == dt)
assert(c.Y.dtype == dt)
assert(c.U.dtype == dt)
def test_06(self):
N = 16
M = 4
Nc = 3
Nd = 8
X = np.random.randn(N, N, Nc, 1, M)
S = np.random.randn(N, N, Nc)
try:
opt = ccmod.ConvCnstrMODOptions({'Verbose': False,
'MaxMainIter': 20, 'LinSolveCheck': True})
c = ccmod.ConvCnstrMOD(X, S, (Nd, Nd, 1, M), opt=opt, dimK=0)
c.solve()
except Exception as e:
print(e)
assert(0)
assert(np.array(c.getitstat().XSlvRelRes).max() < 1e-5)
cri = cnvrep.CDU_ConvRepIndexing(D0.shape, sh)
"""
Define X and D update options.
"""
lmbda = 0.2
mu = 0.1
optx = cbpdn.ConvBPDNJoint.Options({'Verbose': False, 'MaxMainIter': 1,
'rho': 50.0*lmbda + 0.5, 'AutoRho': {'Period': 10,
'AutoScaling': False, 'RsdlRatio': 10.0, 'Scaling': 2.0,
'RsdlTarget': 1.0}})
optd = ccmod.ConvCnstrMODOptions({'Verbose': False, 'MaxMainIter': 1,
'rho': 10.0*cri.K, 'AutoRho': {'Period': 10, 'AutoScaling': False,
'RsdlRatio': 10.0, 'Scaling': 2.0, 'RsdlTarget': 1.0}},
method='ism')
"""
Normalise dictionary according to dictionary Y update options.
"""
D0n = cnvrep.Pcn(D0, D0.shape, cri.Nv, dimN=2, dimC=0, crp=True,
zm=optd['ZeroMean'])
"""
Update D update options to include initial values for Y and U.
"""
D0 = util.convdicts()['G:12x12x36']
# Compute sparse representation on current dictionary
lmbda = 0.01
opt = cbpdn.ConvBPDN.Options({
'Verbose': True,
'MaxMainIter': 200,
'RelStopTol': 5e-3
})
b = cbpdn.ConvBPDN(D0, sh, lmbda, opt)
b.solve()
# Update dictionary for training set sh
opt = ccmod.ConvCnstrMODOptions({
'Verbose': True,
'MaxMainIter': 100,
'rho': 5.0
})
c = ccmod.ConvCnstrMOD(b.Y, sh, D0.shape, opt)
c.solve()
print("ConvCnstrMOD solve time: %.2fs" % c.timer.elapsed('solve'))
D1 = c.getdict().squeeze()
# Display dictionaries
fig1 = plot.figure(1, figsize=(14, 7))
plot.subplot(1, 2, 1)
plot.imview(util.tiledict(D0), fgrf=fig1, title='D0')
plot.subplot(1, 2, 2)
plot.imview(util.tiledict(D1), fgrf=fig1, title='D1')
fig1.show()