Combining Computer Algebra with SAT for Word-Level Reasoning
Combining Computer Algebra with SAT for Word-Level Reasoning
Disciplines
Computer Sciences (20%); Mathematics (80%)
Keywords
-
Formal Verification,
Computer Algebra,
SAT Solving,
Proof Logging,
Automated Reasoning,
Word-Level Reasoning
Formal verification aims to ensure the correctness of complex systems, hardware designs, and software. Unlike traditional testing methods, which rely on executing test cases and observing outputs, formal verification employs precise mathematical techniques and logical reasoning to thoroughly analyze and validate system properties, guaranteeing that they meet specified requirements and standards. Indeed, formal verification is essential in modern engineering practices as it enables the construction of robust and dependable systems that fulfill high quality, safety, and security criteria. The successful development of sophisticated automated reasoning tools such as solvers for the boolean satisfiability problem (SAT) and computer algebra algorithms opened up new perspectives and challenges for formal verification. Although both SAT and computer algebra have a long history, they have mostly been utilized for problem solving separately. Because of the absence of close integration, it is currently not possible to simultaneously harness the strengths of both worlds for real-world problem solving in a single method. The mission of this project is to alter the reasoning landscape in bit-precise formal verification by combining SAT and computer algebra to develop unique SAT-based algebraic methods for word- level reasoning over polynomials. Here, words describe vectors and sequences of bits, capturing for example, portions of computer memory. However, while discussing the broad area of computer algebra, we have to put our focus on selected polynomial rings. We concentrate on integrating SAT solving into algebraic reasoning over pseudo-boolean integer polynomials, which are for instance used to verify hardware circuits, as well as polynomials over finite domains, which can be used to model computer memory and cryptographic encodings. To validate the novel methods we additionally develop proof logging techniques to certify the verification results and hence are able to provide an additional layer of trust. Tightly linking algebraic reasoning with SAT solving will enable us to fully harness the potential of both techniques, and has the potential to significantly increase the capacity of state-of-the-art methods for reasoning over finite fields, bit-vectors, or pseudo-boolean integer polynomials. Advancing formal method techniques is indispensable and we believe that linking these orthogonal reasoning approaches is a key step in this direction. Success in this project will yield unique theoretical and practical solutions with practical applications in hardware verification, bit-vector reasoning, blockchain technology and post-quantum cryptography.
- Technische Universität Wien - 100%
- Jakob Nordström, University of Copenhagen - Denmark
- Armin Biere, Albert-Ludwigs-Universität Freiburg - Germany
- Christoph Scholl, Albert-Ludwigs-Universität Freiburg - Germany
- Mate Soos - Germany
- Toni Jussila - Germany
Research Output
- 2 Citations
- 4 Publications
- 1 Datasets & models
- 2 Software
- 5 Disseminations
-
2025
Title Extracting Linear Relations from Gröbner Bases for Formal Verification of And-Inverter Graphs DOI 10.1007/978-3-031-90643-5_19 Type Book Chapter Author Kaufmann D Publisher Springer Nature Pages 355-374 -
2025
Title PolySAT: Word-level Bit-vector Reasoning in Z3 DOI 10.1007/978-3-031-86695-1_4 Type Book Chapter Author Rath J Publisher Springer Nature Pages 47-69 -
2024
Title MCSat-Based Finite Field Reasoning in the Yices2 SMT Solver (Short Paper) DOI 10.1007/978-3-031-63498-7_23 Type Book Chapter Author Hader T Publisher Springer Nature Pages 386-395 Link Publication -
2025
Title Guess and Prove: A Hybrid Approach to Linear Polynomial Recovery in Circuit Verification Type Conference Proceeding Abstract Author Hofstadler C Conference International Conference on Principles and Practice of Constraint Programming
-
2025
Link
Title MultiLinG - Extracting Linear Relations from Gröbner Bases for Formal Verification of And-Inverter Graphs (Artifact) DOI 10.5281/zenodo.14609934 Type Database/Collection of data Public Access Link Link
-
2025
Link
Title MultiLing DOI 10.5281/zenodo.14610365 Link Link -
2025
Link
Title TalisMan - Trusted Algebraic LInearization of Sub-Circuits with Matrix-based Algorithms using Normalforms Link Link
-
2024
Link
Title Talk (Freiburg) Type A talk or presentation Link Link -
2024
Link
Title Workshop Primary School (Vienna) Type Participation in an activity, workshop or similar Link Link -
2025
Title Talk (Paris) Type A talk or presentation -
2025
Title Talk (Stanford) Type A talk or presentation -
2025
Link
Title Seminar Organization (Dagstuhl) Type Participation in an activity, workshop or similar Link Link