Machine Learning As an Advanced Algorithm To Solve Optimization Problems in Physics

Abstract

It is argued that the standard procedures to solve problems in physics particularly in the field of electrodynamics have in a tacit manner the actions of Machine Learning, such as the criteria of Tom Mitchell, (i) task, (ii) performance, and (iii) experience. In this way, it is presented the case of electric interaction of two charged objects inside a finite cylindric. It is found that Machine Learning concepts is matching well to the requirements to limit the usage of space and energy. Beyond of using such principles as a methodology to solve problems, the concepts of Machine Learning can be projected in the theory of physics to improve and calibrate the mathematical structure of physics equations without touching their fundamental roles.