The native Anderson Acceleration ability had a restriction in place that required that a solution step be taken at least once before the solver was reset. This normally makes sense, however, for multistate cases where only a single iteration will be taken on a state this restriction becomes a problem. This MR simply removes the restriction.
The native Anderson Acceleration ability had a restriction in place that required that a solution step be taken at least once before the solver was reset. This normally makes sense, however, for multistate cases where only a single iteration will be taken on a state this restriction becomes a problem. This MR simply removes the restriction.