Most optimisation problems arising in industry and commerce involve different subproblems that are best addressed by different algorithms and constraint solvers. In ECLiPSe it is easy to use different constraint solvers in combination. The different solvers may share variables and even constraints.
We discuss reasons for combining the eplex and IC solver libraries and explore ways of doing this. The repair library plays a useful role in propagating solutions generated by a linear solver to other variables handled by the domain solver. We show how this works in a generic hybrid algorithm termed probing.