Discover AmeyaJagtap/Conservative_PINNs Open Source project

Stars
55
Rank 537,381 (Top 11 %)
Language
Python
License
MIT License
Created over 3 years ago
Updated almost 2 years ago

AmeyaJagtap

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

We propose a conservative physics-informed neural network (cPINN) on decompose domains for nonlinear conservation laws. The conservation property of cPINN is obtained by enforcing the flux continuity in the strong form along the sub-domain interfaces.

XPINNs

Extended Physics-Informed Neural Networks (XPINNs): A Generalized Space-Time Domain Decomposition Based Deep Learning Framework for Nonlinear Partial Differential Equations

146

Locally-Adaptive-Activation-Functions-Neural-Networks-

Python codes for Locally Adaptive Activation Function (LAAF) used in deep neural networks. Please cite this work as "A D Jagtap, K Kawaguchi, G E Karniadakis, Locally adaptive activation functions with slope recovery for deep and physics-informed neural networks, Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 20200334, 2020. (http://dx.doi.org/10.1098/rspa.2020.0334)".

Python

XPINNs_TensorFlow-2

XPINN code written in TensorFlow 2

Rowdy_Activation_Functions

We propose Deep Kronecker Neural Network, which is a general framework for neural networks with adaptive activation functions. In particular we proposed Rowdy activation functions that inject sinusoidal fluctuations thereby allows the optimizer to exploit more and train the network faster. Various test cases ranging from function approximation, inferring the PDE solution, and the standard deep learning benchmarks like MNIST, CIFAR-10, CIFAR-100, SVHN etc are solved to show the efficacy of the proposed activation functions.

Python

Adaptive_Activation_Functions