What problem does this paper attempt to address?

The main problem that this paper attempts to solve is **verifying the correctness and security of parallel programs**, especially for parallel programs written using MPI (Message Passing Interface). Specifically, the author proposes a new mathematical model to describe and verify MPI programs, aiming to overcome the limitation that existing methods are only applicable to a limited category of parallel programs. ### Core issues of the paper 1. **Complexity of parallel programs**: Parallel programs are designed for multi - processor computing systems (MPCS), and their development is more difficult. Ensuring their correctness and security is a complex mathematical problem. 2. **Limitations of existing methods**: Existing verification methods can only be applied to a limited category of parallel programs and cannot handle all types of MPI programs, especially those that can generate an arbitrary number of processes. ### Solutions To solve the above problems, the author proposes the following solutions: - **Introducing a new mathematical model**: This model can describe and verify MPI programs on a specific subset, and is especially suitable for MPI programs that can generate an arbitrary number of processes. - **Verification example of matrix - multiplication MPI program**: The effectiveness of the new model is verified through a specific matrix - multiplication MPI program, showing how to apply this model for actual parallel - program verification. ### Main contributions 1. **Enhanced universality**: The new model is not only applicable to a fixed number of processes, but can also handle the situation of dynamically generating an arbitrary number of processes. 2. **Practicality verification**: The effectiveness and feasibility of the model are verified through an actual matrix - multiplication MPI program, proving its potential in practical applications. ### Conclusion The main objective of this paper is to improve the verification methods of parallel programs by introducing a new mathematical model, especially in terms of MPI - program verification, thereby enhancing the correctness and security of parallel programs. This will contribute to promoting the development of the parallel - computing field, especially in high - performance computing and distributed systems.

Verification of MPI programs

Towards deductive verification of MPI programs against session types

Matrix Multiplication Verification Using Coding Theory

Methods to Model-Check Parallel Systems Software

Verifying Parallel Low-Level Programs for Multi-core Processor

Distributed Bounded Model Checking

Parallel Software Model Checking

Parallel approach of three-dimensional phase-field model of binary alloy in MPI+OpenMP environment

A New Method of Verification of Functional Programs

Verifying Programs Using Abstraction and Theorem Proving.

To parallelize or not to parallelize, bugs issue

A graph model of message passing processes

GPU-Accelerated Verification of Machine Learning Models for Power Systems

A Novel Approach for Bounded Model Checking Through Full Parallelism

Verification of Concrete Programs with Respect to Abstract Programs

MPIrigen: MPI Code Generation through Domain-Specific Language Models

To parallelize or not to parallelize, control and data flow issue

A Formal Approach for Modeling and Verification of Distributed Systems.

Verification of deep probabilistic models

Code Generation Verification – Assessing Numerical Equivalence between Simulink Models and Generated Code

Certifying MIP-based Presolve Reductions for 0-1 Integer Linear Programs