A parallel subgradient projections method for the convex feasibility problem (Q1091948): Difference between revisions

The author considers the convex feasibility problem in the following form. Let Q be a closed convex subset of \(R^ n\) and let \(g_ i:\) \(R^ n\to R\), \(i=1,...,m\), be convex functions. A solution \(x^*\) is to be found for the system: \(x\in Q\), \(g_ i(x)\geq 0\), \(i=1,...,m\). The main idea is based on a generalization of an acceleration procedure which is suitable for parallel processing. It is proved that the presented method converges for any starting point. Some comparative numerical experiences are also reported. The paper ends with some concluding remarks and hints for further investigations to choose those relaxation parameters of the algorithm properly to which it seems to be sensitive.