Open AccessA learning-enhanced projection method for solving convex feasibility problems
Author(s) -
Janosch Rieger
Publication year - 2022
Publication title -
discrete and continuous dynamical systems. series b
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.864
H-Index - 53
eISSN - 1553-524X
pISSN - 1531-3492
DOI - 10.3934/dcdsb.2021054
Subject(s) - generalization , regular polygon , projection (relational algebra) , convergence (economics) , orthographic projection , carry (investment) , basis (linear algebra) , computer science , algorithm , mathematics , projection method , mathematical optimization , artificial intelligence , dykstra's projection algorithm , geometry , mathematical analysis , finance , economics , economic growth
We propose a generalization of the method of cyclic projections, which uses the lengths of projection steps carried out in the past to learn about the geometry of the problem and decides on this basis which projections to carry out in the future. We prove the convergence of this algorithm and illustrate its behavior in a first numerical study.