def prox(self, x, tau):
# define current number of iterations
if isinstance(self.niter, int):
niter = self.niter
else:
niter = self.niter(self.count)
# solve proximal optimization
if self.Op is not None and self.b is not None:
y = x + tau * self.OpTb
if self.q is not None:
y -= tau * self.alpha * self.q
if self.Op.explicit:
Op1 = MatrixMult(
np.eye(self.Op.shape[1]) +
tau * self.sigma * np.conj(self.Op.A.T) @ self.Op.A)
x = Op1.div(y)
else:
Op1 = Identity(self.Op.shape[1], dtype=self.Op.dtype) + \
tau * self.sigma * self.Op.H * self.Op
x = lsqr(Op1, y, iter_lim=niter, x0=self.x0)[0]
if self.warm:
self.x0 = x
elif self.b is not None:
num = x + tau * self.sigma * self.b
if self.q is not None:
num -= tau * self.alpha * self.q
x = num / (1. + tau * self.sigma)
else:
num = x
if self.q is not None:
num -= tau * self.alpha * self.q
x = num / (1. + tau * self.sigma)
return x
def prox(self, x, tau):
if self.Op is not None and self.b is not None:
y = x - tau * self.b
if self.Op.explicit:
Op1 = MatrixMult(np.eye(self.Op.shape[0]) + tau * self.Op.A)
x = Op1.div(y)
else:
Op1 = Identity(self.Op.shape[0], dtype=self.Op.dtype) + \
tau * self.Op.A
x = lsqr(Op1, y, iter_lim=self.niter, x0=self.x0)[0]
if self.warm:
self.x0 = x
elif self.b is not None:
x = x - tau * self.b
return x