search

isds-neu / PhyCRNet

Physics-informed convolutional-recurrent neural networks for solving spatiotemporal PDEs
MIT License
118 stars 35 forks source link

PDE #7

Open mengjie123 opened 2 years ago

mengjie123 commented 2 years ago

Hi,

Very interesting work! I have one small question: taking the 2D Burger's equation as an example, if the viscosity value is changed, we need to retrain the neural network?

Best wishes

paulpuren commented 2 years ago

Hi - yes, we need retrain the network. If the viscosity value is changed, the governing equation will be different for the trained model. But here you may try retraining the network of a different viscosity model (e.g., 400) based on the current trained model (e.g., 200). It is also called curriculum learning. Hope that answer helps you.

mengjie123 commented 2 years ago

Make sense. Thanks for your quick reply!