Physics-Informed Neural Networks Application To Mass-Spring System Solution

In this paper, the application of the Physics-Informed Neural Network (PINN) tool to solve the mass-spring system with free oscillations described by an ordinary differential equation is studied and the results are compared to the analytic solution. PINNs offer the dual benefit of being data-driven for model learning and ensuring consistency with physics. The paper also focuses on the computation complexity of the selected methods, the efficiency and the accuracy. The results emphasise the benefits and limitations of PINNs and of the analytic solution.