mr2.algorithms.optimizers

Optimizers.

Functions

adam(f, initial_parameters, *[, ...])	Adam for non-linear minimization problems.
admm_l2(g, op, b, a, initial_values, *, tau)	ADMM for \(\min_x \frac{1}{2}\|Op\,x-b\|_2^2 + g(Ax)\).
admm_linear(f, g, operator, initial_values, ...)	Linearized ADMM for \(\min_x f(x) + g(Ax)\).
bicg(operator, right_hand_side, *[, ...])	(Preconditioned) Bi-Conjugate Gradient Stabilized method for \(Hx=b\).
cg(-> tuple[~torch.Tensor])	(Preconditioned) Conjugate Gradient for solving \(Hx=b\).
lbfgs(f, initial_parameters[, ...])	LBFGS for (non-linear) minimization problems.
pdhg(f, g, operator, initial_values[, ...])	Primal-Dual Hybrid Gradient Algorithm (PDHG).
pgd(f, g, initial_value[, stepsize, ...])	Proximal gradient descent algorithm for solving problem \(min_x f(x) + g(x)\).

Classes

OptimizerStatus

Base class for OptimizerStatus.