Hiroshi Unno

Satoshi Kura, Hiroshi Unno, Takeshi Tsukada

Many quantitative properties of probabilistic programs can be characterized as least fixed points, but verifying their lower bounds remains a challenging problem. We present a new approach to lower-bound verification that exploits and extends the connection between the uniqueness of fixed points and program termination. The core technical tool is a generalization of ranking supermartingales, which serves as witnesses of the uniqueness of fixed points. Our method provides a simple and unified reasoning principle applicable to a wide range of quantitative properties, including termination probability, the weakest preexpectation, expected runtime, higher moments of runtime, and conditional weakest preexpectation. We provide a template-based algorithm for automated verification of lower bounds and demonstrate the effectiveness of the proposed method via experiments.

Satoshi Kura, Hiroshi Unno

We propose new supermartingale-based certificates for verifying almost sure satisfaction of $ω$-regular properties: (1) generalised Streett supermartingales (GSSMs) and their lexicographic extension (LexGSSMs), (2) distribution-valued Streett supermartingales (DVSSMs), and (3) progress-measure supermartingales (PMSMs) and their lexicographic extension (LexPMSMs). GSSMs, LexGSSMs, and DVSSMs are derived from least-fixed point characterisations of positive recurrence and null recurrence of Markov chains with respect to given Streett conditions; and PMSMs and LexPMSMs are probabilistic extensions of parity progress measures. We study the hierarchy among these certificates and existing certificates, namely Streett supermartingales, by comparing the classes of problems that can be verified by each type of certificates. Notably, we show that our certificates are strictly more powerful than Streett supermartingales. We also prove completeness of GSSMs for positive recurrence and of DVSSMs for null recurrence: DVSSMs are, in theory, the most powerful certificates in the sense that for any Markov chain that almost surely satisfies a given $ω$-regular property, there exists a DVSSM certifying it. We provide a sound and relatively complete algorithm for synthesising LexPMSMs, the second most powerful certificates in the hierarchy. We have implemented a prototype tool based on this algorithm, and our experiments show that our tool can successfully synthesise certificates for various examples including those that cannot be certified by existing supermartingales.

Minchao Wu, Takeshi Tsukada, Hiroshi Unno et al.

Loop-invariant synthesis is the basis of program verification. Due to the undecidability of the problem in general, a tool for invariant synthesis necessarily uses heuristics. Despite the common belief that the design of heuristics is vital for the performance of a synthesizer, heuristics are often engineered by their developers based on experience and intuition, sometimes in an \emph{ad-hoc} manner. In this work, we propose an approach to systematically learning heuristics for template-based CounterExample-Guided Inductive Synthesis (CEGIS) with reinforcement learning. As a concrete example, we implement the approach on top of PCSat, which is an invariant synthesizer based on template-based CEGIS. Experiments show that PCSat guided by the heuristics learned by our framework not only outperforms existing state-of-the-art CEGIS-based solvers such as HoICE and the neural solver Code2Inv, but also has slight advantages over non-CEGIS-based solvers such as Eldarica and Spacer in linear Constrained Horn Clause (CHC) solving.

Naoki Kobayashi, Taro Sekiyama, Issei Sato et al.

We propose a novel framework of program and invariant synthesis called neural network-guided synthesis. We first show that, by suitably designing and training neural networks, we can extract logical formulas over integers from the weights and biases of the trained neural networks. Based on the idea, we have implemented a tool to synthesize formulas from positive/negative examples and implication constraints, and obtained promising experimental results. We also discuss two applications of our synthesis method. One is the use of our tool for qualifier discovery in the framework of ICE-learning-based CHC solving, which can in turn be applied to program verification and inductive invariant synthesis. Another application is to a new program development framework called oracle-based programming, which is a neural-network-guided variation of Solar-Lezama's program synthesis by sketching.