In the standard Constraint Programming (CP) framework, an integer variable represents a signed integer and its domain is bounded by some minimal and maximal integer type values. In existing CP tools, the integer type is used to represent domain values, and hence domain bounds are inherently limited by the minimal and maximal signed integer values representable on a given platform. However, this implementation of integer variable fails to satisfy use cases where modeled integers can be arbitrarily large. An example of such CP application is the functional test generation where integer variables are used to model large architectural fields like memory addresses or operand data. In addition, in such applications, the set of standard arithmetic operations on an integer variable provided by the traditional CP framework does not represent the whole range of operations required for modeling. In this paper, we define a new type of integer variables with arbitrarily large domain size and a modified operation set. We show how this variable type can be realized on top of a traditional CP framework by means of global constraints over standard integer variables. The ideas presented in this paper can also be used to implement a native variable of the introduced type in a CP tool. The paper provides experimental results to demonstrate the effectiveness of the proposed approach.
展开▼
