sparse_ho.algo.ImplicitForward

class sparse_ho.algo.ImplicitForward(tol_jac=0.001, max_iter=100, n_iter_jac=100, use_stop_crit=True, verbose=False)

Algorithm to compute the hypergradient using implicit forward differentiation.

First the algorithm computes the regression coefficients. Then the iterations of the forward differentiation are applied to compute the Jacobian.

Parameters
tol_jac: float

Tolerance for the Jacobian computation.

max_iter: int

Maximum number of iterations for the inner solver.

n_iter_jac: int

Maximum number of iterations for the Jacobian computation.

use_stop_crit: bool, optional (default=True)

Use stopping criterion in hypergradient computation. If False, run to maximum number of iterations.

verbose: bool, optional (default=False)

Verbosity of the algorithm.

__init__(tol_jac=0.001, max_iter=100, n_iter_jac=100, use_stop_crit=True, verbose=False)

Initialize self. See help(type(self)) for accurate signature.

Methods

__init__([tol_jac, max_iter, n_iter_jac, …])

Initialize self.

compute_beta_grad(X, y, log_alpha, model, …)

get_beta_jac(X, y, log_alpha, model, …[, …])

Compute beta and hypergradient using implicit forward differentiation.