1
24. 10. 2022.
Solving the electrostatic Laplace's equation with a parameterizable physics informed neural network
In this work, we present an approach for training parametrized physics informed neural networks to solve PDEs in a self supervised fashion, which means that no labeled input-output data is needed to train the neural network. The main contribution of this work is the achievement of a model with parameterizable boundary condition functions. This means that no retraining is needed to produce correct results for changing boundary conditions.