Formal Hardware Verification

Overview

The field develops methods to prove that hardware designs meet their specifications, find design bugs, and establish equivalence between implementations. Core techniques include bounded model checking, IC3, SMT-based reasoning, abstraction and refinement, and checking SystemVerilog assertions. Current work also uses learned models and LLM-based agents to generate properties, guide searches, close coverage gaps, and repair RTL, with evaluation on suites such as HWMCC and RTL assertion benchmarks. A central challenge is ensuring that generated properties reflect design intent: syntax and tool acceptance alone do not guarantee meaningful proofs, broad coverage, or reliable bug detection.

Latest papers

CardsList

32 papers3 papers in the last four weeks, down 25% on the four weeks before. 0.0% of all new papers.

  1. MPBMC: Multi-Property Bounded Model Checking with GNN-guided Clustering

    Feb 26, 2026Soumik Guha Roy, Sumana Ghosh, Ansuman Banerjee +2Bounded Model CheckingFormal Verification

  2. EvolveGen: Algorithmic Level Hardware Model Checking Benchmark Generation through Reinforcement Learning

    Feb 26, 2026Guangyu Hu, Xiaofeng Zhou, Wei Zhang +1Synthetic Benchmark GenerationFormal Hardware Verification