Computational framework for optimal control of hybrid systems with a partitioned state space is presented. It is shown that necessary conditions for optimality for a discrete-time dynamic system can be solved concurrently for various boundary conditions, according to the recent development of discrete-time Hamilton-Jacobi theory. This unique property is utilized to construct computationally efficient numerical optimization of hybrid systems where discrete switching dynamics occurs at the boundary between partitions of the configuration space. A benchmark example shows that the proposed approach has substantial computational advantages compared with the existing ones. (C) 2014 Elsevier Ltd. All rights reserved.