The project starts by picking any equation for force/acceleration that was learned in class, we used Hooke’s law as an example, but any could be used. The project then involves creating a differential equation for the position of the object and using a neural network to solve for the position. The neural networks results are then compared to numerical results from methods such as Runge-Kutta and analytical solutions if they exist. This same methodology could also be applied in most physics classes since differential equations are quite common in physics. See the project description on the class website here (released under a Creative Commons License): https://github.com/mhjensen/Physics321/blob/master/doc/Honorsprojects/NeuralNetworks.pdf
The project starts by picking any equation for force/acceleration that was learned in class, we used Hooke’s law as an example, but any could be used. The project then involves creating a differential equation for the position of the object and using a neural network to solve for the position. The neural networks results are then compared to numerical results from methods such as Runge-Kutta and analytical solutions if they exist. This same methodology could also be applied in most physics classes since differential equations are quite common in physics. See the project description on the class website here (released under a Creative Commons License): https://github.com/mhjensen/Physics321/blob/master/doc/Honorsprojects/NeuralNetworks.pdf