1: \begin{abstract}

2: We present convergence results in expectation for stochastic subspace correction schemes and their accelerated versions to solve symmetric positive-definite variational problems, and discuss their potential  for achieving fault tolerance in an unreliable compute network. We employ the standard overlapping domain decomposition algorithm for PDE discretizations to discuss the latter aspect.

3: 

4: \keywords{subspace correction \and Schwarz iterative methods \and randomization \and convergence rates   \and fault tolerance}

5: \subclass{65F10 \and 65N22 \and 65N55 \and 65Y05 \and 68W20}

6: \end{abstract}

7: