ADMM_proj: ADMM algorithm

In celiaescribe/BDcocolasso: Implementation of CoCoLasso and Block Descent CoCoLasso

Description

Finds the nearest positive semi-definite matrix with respect to the max norm

Usage

ADMM_proj(
  mat,
  epsilon = 1e-04,
  mu = 10,
  it.max = 1000,
  etol = 1e-04,
  etol_distance = 1e-04
)

Arguments

mat	Matrix to be projected
epsilon	Approximation of the space of positive semi-definite matrix at epsilon
mu	Penalty parameter of ADMM algorithm
it.max	Number maximum of iterations
etol	Tolerance parameter for the convergence of primal and dual residual
etol_distance	Tolerance parameter for the convergence of the distance

Value

list containing

• mat Projected matrix

• df_ADMM dataframe containing parameters of the ADMM algorithm for each iteration of the algorithm

See Also

https://web.stanford.edu/~boyd/papers/pdf/admm_distr_stats.pdf

Examples

M = matrix(-1,20,20)
mat_proj <- BDcocolasso::ADMM_proj(mat=M)$mat

celiaescribe/BDcocolasso documentation built on Feb. 11, 2020, 11:41 p.m.