Competition Report: CHC-COMP-21

Fedyukovich, Grigory; Rümmer, Philipp

doi:10.4204/eptcs.344.7

Cited by 14 publications

(16 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The number of labeled graph representations extracted from a collection of CHC-COMP benchmark [20]. For each SMT-LIB file, the graph representations for Task 1,2,3,4 are extracted together using the timeout of 3 hours, and for task 5 is extracted using 20 minutes timeout.…”

Section: Tablementioning

confidence: 99%

“…Table 7 shows the number of labelled graph representations from a collection of CHC-COMP benchmarks [20]. All graphs were constructed by first running the preprocessor of Eldarica [25] on the clauses, then building the graphs as described in Section 3, and computing training data.…”

Section: Benchmarks and Datasetmentioning

confidence: 99%

“…Our benchmark data is extracted from the 8705 linear and 8425 non-linear Linear Integer Arithmetic (LIA) problems in the CHC-COMP repository 1 (see Table 1 in the competition report [20]). The verification problems come from various sources (e.g., higher-order program verification benchmark 2 and benchmarks generated with JayHorn 3 ), therefore cover programs with different size and complexity.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Exploring Representation of Horn Clauses using GNNs (technique report)

Liang¹,

Rümmer²,

Brockschmidt³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

In recent years, the application of machine learning in program verification, and the embedding of programs to capture semantic information, has been recognised as an important tool by many research groups. Learning program semantics from raw source code is challenging due to the complexity of real-world programming language syntax and due to the difficulty of reconstructing long-distance relational information implicitly represented in programs using identifiers. Addressing the first point, we consider Constrained Horn Clauses (CHCs) as a standard representation of program verification problems, providing a simple and programming language-independent syntax. For the second challenge, we explore graph representations of CHCs, and propose a new Relational Hypergraph Neural Network (R-HyGNN) architecture to learn program features.We introduce two different graph representations of CHCs. One is called constraint graph (CG), and emphasizes syntactic information of CHCs by translating the symbols and their relations in CHCs as typed nodes and binary edges, respectively, and constructing the constraints as abstract syntax trees. The second one is called control-and data-flow hypergraph (CDHG), and emphasizes semantic information of CHCs by representing the control and data flow through ternary hyperedges.We then propose a new GNN architecture, R-HyGNN, extending Relational Graph Convolutional Networks, to handle hypergraphs. To evaluate the ability of R-HyGNN to extract semantic information from programs, we use R-HyGNNs to train models on the two graph representations, and on five proxy tasks with increasing difficulty, using benchmarks from CHC-COMP 2021 as training data. The most difficult proxy task requires the model to predict the occurrence of clauses in counter-examples, which subsumes satisfiability of CHCs. CDHG achieves 90.59% accuracy in this task. Furthermore, R-HyGNN has perfect predictions on one of the graphs consisting of more than 290 clauses. Overall, our experiments indicate that R-HyGNN can capture intricate program features for guiding verification problems.

show abstract

Section: Tablementioning

confidence: 99%

Section: Benchmarks and Datasetmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Exploring Representation of Horn Clauses using GNNs (technique report)

Liang¹,

Rümmer²,

Brockschmidt³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…The tables in Appendix A also report in column '#unique' the number of sat or unsat results produced by a solver for benchmarks for which all other solvers returned unknown. The 'job information' files also include data about the space and memory consumption, which we consider less relevant and therefore are not reported in the tables (see also the CHC-COMP 2021 report [7]).…”

Section: Evaluation Of the Competition Runsmentioning

confidence: 99%

CHC-COMP 2022: Competition Report

Angelis

2022

Electron. Proc. Theor. Comput. Sci.

View full text Add to dashboard Cite

CHC-COMP 2022 is the fifth edition of the competition of solvers for Constrained Horn Clauses. The competition was run in March 2022; the results were presented at the 9th Workshop on Horn Clauses for Verification and Synthesis held in Munich, Germany, on April 3, 2022. This edition featured six solvers, and eight tracks consisting of sets of linear and nonlinear clauses with constraints over linear integer arithmetic, linear real arithmetic, arrays, and algebraic data types. This report provides an overview of the organization behind the competition runs: it includes the technical details of the competition setup as well as presenting the results of the 2022 edition.

show abstract

“…First, it provides a separation of concerns, allowing users of our framework to focus only on aspects related to the tree data structure at hand, while giving CHC solver developers a clean framework that can be instantiated using various model checking algorithms and specialized decision procedures. Furthermore, by expressing CHCs in the standard SMT-LIB language, one can take advantage of different CHC engines, whose performance keeps improving year-over-year, as witnessed by the competition on constrained Horn clauses CHC-COMP [19].…”

Section: Introductionmentioning

confidence: 99%

Reasoning About Data Trees Using CHCs

Faella

Parlato

2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Reasoning about data structures requires powerful logics supporting the combination of structural and data properties. We define a new logic called Mso-D(Monadic Second-Order logic with Data) as an extension of standard Mso on trees with predicates of the desired data logic. We also define a new class of symbolic data tree automata (Sdtas) to deal with data trees using a simple machine. Mso-D and Sdtas are both Turing-powerful, and their high expressiveness is necessary to deal with interesting data structures. We cope with undecidability by encoding Sdta executions as a system of CHCs (Constrained Horn Clauses), and solving the resulting system using off-the-shelf solvers. We also identify a fragment of Mso-D whose satisfiability can be effectively reduced to the emptiness problem for Sdtas. This fragment is very expressive since it allows us to characterize a variety of data trees from the literature, solving certain infinite-state games, etc. We implement this reduction in a prototype tool that combines an Mso decision procedure over trees (Mona) with a CHC engine (Z3), and use this tool to conduct several experiments, demonstrating the effectiveness of our approach across different problem domains.

show abstract

Competition Report: CHC-COMP-21

Cited by 14 publications

References 21 publications

Exploring Representation of Horn Clauses using GNNs (technique report)

Exploring Representation of Horn Clauses using GNNs (technique report)

CHC-COMP 2022: Competition Report

Reasoning About Data Trees Using CHCs

Contact Info

Product

Resources

About