Parallel implementation of Monte Carlo method using mpi on the example of heat transfer equation

Monte Carlo is a valuable tool for solving problems where randomness and uncertainty are the major issues. It finds its applications in mathematics, physics, finance, and many other fields, thus using random sampling for numerical results. The strength of Monte Carlo lies in its effectiveness in complex systems where traditional methods may not work. Its capacity for processing uncertainty has made it indispensable at both research and practical levels. However, its computational complexity makes it time-consuming and, hence, not suitable for solving problems associated with large datasets. The present paper thus introduces an MPI-based parallel approach to Monte Carlo algorithm with concrete reference to the heat transfer problem – a topical and rather studied subject in physics and engineering. Therefore, by using parallelization, we expect to balance the computational processing and achieve better results. The findings show the appropriateness of distributing the load across the various processors, greatly increasing the rate of simulation. The experiments which we are going to depict now prove that our implementation is faster than the sequential version more prominently when the grid becomes large. When using 1,000,000 points, 2 processes render 31.1% faster while and with 4 processes the computation reduces to 47.2%. Yet, if 8 processes are used then there are decreasing performance advantages due to excessive inter process communication. Implementation of parallel processing reduces the computational time by to the half of the sequential method when the grid points in question are 100000, thus useful in large simulations. Thus, works utilizing parallelization, and especially, show that the approach is effective and can be recommended to deal with issues, such as heat transfer prediction in engineering applications.