L-BFGS algorithm for Physics-informed neural networks

Question

0 个投票

Hi,

I am trying to solving ODEs using neural networks. I am using adamupdate function to train the network. I would like to try L-BFGS alogorithm. However, trainbfg function availble with Statistics and Machine Learning Tool box is taking only network, input data and target data as input parameters. I couldn't find a way to plug-in the loss function associated with the ODE and boundary conditions into the training algorithm.

I would be grateful if someone can help me with this.

Thanks in advance.

2 个评论
显示无隐藏无

Conor Daly 2022-3-24

Hi there!

Thanks for your question. Have you had a look at this example? It walks through how to fit a neural network for the Burgers equation using the LBFGS optimizer:

https://www.mathworks.com/help/deeplearning/ug/solve-partial-differential-equations-with-lbfgs-method-and-deep-learning.html