Performance of MATLAB and Python for Computational Electromagnetic Problems

摘要

MATLAB and Python are two common programming languages commonly used in computational electromagnetics. Both provide simple syntax and debugging tools to make even complicated tasks relatively simple. This paper studies how these programming languages compare in throughput for a variety of tasks when utilizing complex numbers which are common in electromagnetics applications. The compared tasks include basic operations like addition, subtraction, multiplication, and division, along with more complex operations like exponentiation, summation, Fourier transforms, and matrix solving. Each of these tests is performed for both single and double precision on the CPU. A 2D finite difference frequency domain problem and a planar array beamforming problem are also presented for comparison of throughput for realistic simulations.