# Yifan Du             [email protected]
# Tamer A. Zaki        [email protected]
# Johns Hokpins University
# US patent submitted 03/08/2021


This main folder containts several realization of EDNN for different 
time dependent partial differential equations. In each folder there 
are four files: 

main.py : The main function for this problem. It contains the problem 
          definition and parameters, the high level initialization, 
          training and marching commands for the EDNN. 
ednn.py : The EDNN abstract class. It contains the detailed implementation
          of EDNN structure, training and marching. Because the 
          boundary conditions and constraints for different PDEs are 
          different, the ednn files are slightly different for these
          cases. 
rhs.py  : The evolutional PDE implementation. It containts The right
          -hand-side of different PDEs. These nonlinear differential 
          operators are implemented using tensorflow. 
marching_schemes : The time marching methods for the neural network 
                   parameters. The forward Euler and Runge-Kutta are 
                   implemented. 

To run one of these cases, first enter the corresponding directory. 

To train the network for initial condition, type: 
           
         python main.py 0

To march the trained network, type: 

         python main.py 1