A Bird's eye view of Matrix Distributed Processing

TL;DR

This paper introduces Matrix Distributed Processing (MDP), a C++ library that simplifies the development of parallel algorithms for differential equations and mesh problems by automating communication in MPI-based environments.

Contribution

The paper presents MDP, a novel C++ library that enables automatic parallelization of algorithms without explicit communication calls, streamlining development for MPI-based parallel computing.

Findings

MDP enables automatic parallelization of algorithms.

MDP simplifies development of parallel solvers for differential equations.

MDP is efficient for mesh-like problems.

Abstract

We present Matrix Distributed Processing, a C++ library for fast development of efficient parallel algorithms. MDP is based on MPI and consists of a collection of C++ classes and functions such as lattice, site and field. Once an algorithm is written using these components the algorithm is automatically parallel and no explicit call to communication functions is required. MDP is particularly suitable for implementing parallel solvers for multi-dimensional differential equations and mesh-like problems.