Abstract: A constraint satisfaction problem (CSP) is used to model a real-world problem in an application domain. The CSP includes a plurality of variables, a plurality of states, and one or more constraints determining which states each variable is permitted to be assigned to in a solution to the CSP. A binary tree is created using the CSP. The binary tree is iteratively searched to determine an efficient solution to the CSP. A new iteration of the search of the binary tree is started when a search depth is less than a restart depth. The restart depth based on a deepest depth previously searched minus a restart value. When the search is restarted, a NoGood marker is created indicating a portion of the binary tree that is not to be searched in the new iteration of the search. An efficient solution to the CSP is reported.

Abstract: A constraint satisfaction problem (CSP) is used to model a real-world problem in an application domain. The CSP includes a plurality of variables, a plurality of states, and one or more constraints determining which states each variable is permitted to be assigned to in a solution to the CSP. A binary tree is created using the CSP. The binary tree is iteratively searched to determine an efficient solution to the CSP. A new iteration of the search of the binary tree is started when a search depth is less than a restart depth. The restart depth based on a deepest depth previously searched minus a restart value. When the search is restarted, a NoGood marker is created indicating a portion of the binary tree that is not to be searched in the new iteration of the search. An efficient solution to the CSP is reported.