Learn R Programming

splitFeas (version 0.1.0)

Multi-Set Split Feasibility

Description

An implementation of the majorization-minimization (MM) algorithm introduced by Xu, Chi, Yang, and Lange (2017) for solving multi-set split feasibility problems. In the multi-set split feasibility problem, we seek to find a point x in the intersection of multiple closed sets and whose image under a mapping also must fall in the intersection of several closed sets.

Copy Link

Version

Install

install.packages('splitFeas')

Monthly Downloads

21

Version

0.1.0

License

MIT + file LICENSE

Maintainer

Eric Chi

Last Published

April 11th, 2018

Functions in splitFeas (0.1.0)

Projection onto a ball

Project onto a cube

Compute the gradient of the majorization.

Self-adaptive projection-type method algorithm for nonlinear multiple-sets split feasibility problem

Backtracking Line Search

MM-quasi-Newton step

Compute the inverse approximate Hessian of the majorization using the Woodbury inversion formula. wood_inv_solve computes the inverse of the Hessian term of the majorization of the proximity function using the Woodbury formula. The function mmqn_step invokes wood_inv_solve instead of ddg if the argument woodbury=TRUE. This should be used when p << n.

nmsfp_sap_one_step

One step of self-adaptive projection-type method for the NMSFP

MM algorithm (accelerated) for nonlinear multiple-sets split feasibility problem

MM algorithm for nonlinear multiple-sets split feasibility problem

split_feasibility

split_feasibility

Compute the approximate Hessian of the majorization.

Quasi-Newton acceleration of MM algorithm

Proximity function

Compute soft-max

project_halfspace

Projection onto a halfspace

Project onto a square