Year: 2018
Type: conference-paper
Abstract: Satisfiability Modulo Theories, SMT, solvers are used in many applications. These applications benefit from the power of tuned and scalable theorem proving technologies for supported logics and specialized theory solvers. SMT solvers are primarily used to determine whether formulas are satisfiable. Furthermore, when formulas are satisfiable, many applications need models that assign values to free variables. Yet, in many cases arbitrary assignments are insufficient, and what is really needed is an optimal assignment with respect to objective functions. So far, users of Z3, an SMT solver from Microsoft Research, build custom loops to achieve objective values. This is no longer necessary with νZ (new-Z, or max-Z), an extension within Z3 that lets users formulate objective functions directly with Z3. Under the hood there is a portfolio of approaches for solving linear optimization problems over SMT formulas, MaxSMT, and their combinations. Objective functions are combined as either Pareto fronts, lexicographically, or each objective is optimized independently.
Source: EPiC series in computing
DOI: 10.29007/jmxj
Authors: Nikolaj Bjorner, Anh-Dung Phan
Institutions: Microsoft (United States), Technical University of Denmark
Language: English
Subfield: Computational Theory and Mathematics
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W2295061292