Mining global constraints for improving bounded sequential equivalence checking

Abstract

In this paper, we propose a novel technique on mining relationships in a sequential circuit to discover global constraints. In contrast to the traditional learning methods, our mining algorithm can find important relationships among several nodes efficiently. The nodes involved may often span several timeframes, thus improving the deductibility of the problem instance. Experimental results demonstrate that the application of these global constraints to SAT-based bounded sequential equivalence checking can achieve one to two orders of magnitude speedup. In addition, because it is orthogonal to the underlying SAT solver, it can help to enhance the efficacy of typical SAT based verification flows.

Constraint Satisfaction Approach on Product Configuration with Cost Estimation

“…provided a generic framework for automatically learning implied global constraints with parameters according to the actual domains of variables, not just from structural or syntactical properties, and applied the framework to both discrete and continuous properties. Similar work is offered by Wu and Hsiao[211]. Simonis[169] studied detailedly the modeling and solving of global constraints in three different applications.…”

“…A set of zero and more items is termed as an itemset, and a k -itemset refers to an itemset including k items. An association rule can be viewed as a probabilistic statement to infer the co-occurrence of certain items in data sets, taking the form of X→Y where X and Y are disjoint itemsets; i.e., X ∩ Y = ∅ [185,211]. For instance,…”

Mining Complex Boolean Expressions for Sequential Equivalence Checking

“…A data mining approach has recently been proposed [15] using Apriori [16], a popular methodology for association rule mining to compute potential three node invariants among signals. To further reduce the mining cost, domain knowledge of circuit is used to improve the quality of the discovered rules [17].…”

“…To further reduce the mining cost, domain knowledge of circuit is used to improve the quality of the discovered rules [17]. These static and dynamic multi node implications can significantly enhance bounded SEC.…”

Simulation-Directed Invariant Mining for Software Verification

“…In terms of mining, a machine learning approach was used in [17] to mine formal specifications by observing the program executions, assuming the program is generally correct. For hardware, in [5], data mining was applied for global constraints mining to improve the performance of SAT-based bounded equivalence checking of sequential circuits.…”