Abstract

Constraint logic programming (CLP) languages allow natural and concise programming with constraints. In executing CLP languages the cost of constraint solving dominates the time and space considerations, so optimization of constraint solving is of paramount importance. CLP(R) is a constraint logic programming language over real numbers. Typically, in the execution of a CLP(R) program, many variables remain in a state where the constraint solver continually manipulates them, even if they will never be referenced in the remaining computation. This phenomenon is critical since continually manipulating these variables which are no longer of interest causes slower execution. Detecting such variables is a non-trivial problem. In this paper we describe a data flow analysis to detect redundant variables, and explain how the solver can be modified to take advantage of this information. Experiments indicate that removing such variables yields substantial space and time saving in the execution of...

Citations