Solving stochastic convex feasibility problems in hilbert spaces |
| |
Authors: | G Crombez |
| |
Institution: | Department of Applied Mathematics and Computer Science , University of Ghent , Gent, B–9000, Belgium |
| |
Abstract: | In a stochastic convex feasibility problem connected with a complete probability space (Ω,A,μ) and a family of closed convex sets (Cω)ωεΩ in a real Hilbert space H, one wants to find a point that belongs to Cω for μ almost all ω ε Ω. We present a projection based method where the variable relaxation parameter is defined by a geometrical condition, leading to an iteration sequence that is always weakly convergent to a μ almost common point. We then give a general condition assuring norm convergence of this equation to that μ almost common point |
| |
Keywords: | stochastic convex feasibility problems expected projection methods almost common point |
|
|