핵심 개념
Probabilistic programs pose challenges for program analyses, NPA-PMA framework offers efficient solutions.
초록
The article introduces the NPA-PMA framework for analyzing probabilistic programs efficiently. It addresses challenges in designing compositional and efficient program analyses for probabilistic programs with unstructured control-flow, nondeterminism, and general recursion. The framework is based on Newtonian Program Analysis (NPA) and introduces 𝜔-continuous pre-Markov algebras (𝜔PMAs) to factor out common parts of different analyses. It allows for non-iterative strategies to solve linearized equations, outperforming Kleene iteration and providing generality for designing program analyses. The article discusses the theoretical foundation, implementation, and experimental evaluation of NPA-PMA.
Overview:
Introduction to Probabilistic Programs
Challenges in Program Analyses
Introduction of NPA-PMA Framework
Key Features of NPA-PMA
Experimental Evaluation
Contributions:
Development of NPA-PMA Framework
Adoption of Regular Infinite-Tree Expressions
Prototype Implementation and Instantiation with Abstract Domains
통계
NPA-PMA holds considerable promise for outperforming Kleene iteration.
NPA-PMA provides great generality for designing program analyses.
인용구
"NPA-PMA allows analyses to supply a non-iterative strategy to solve linearized equations."
"Our experimental evaluation demonstrates that NPA-PMA holds considerable promise for outperforming Kleene iteration."