/fys-stk-project3

Primary LanguageJupyter Notebook

fys-stk-project3

Repo covering project 3 of FYS-STK4155 where we compare neural network methods for solving differential equations with more traditional methods, such as finite differences.

  1. Heat Equation - Finite Differences Solution.ipynb contains the code to run and plot solutions to the heat equation with the explicit Euler method. The class object implementation of the heat equation and the explicit Euler method is implemented in the program HeatEquation.py.

  2. Heat Equation - Neural Network Model.ipynb contains the TensorFlow implementation of a neural net- work model, and the code to train and plot the solution. To run, I recommend opening in Colab and ensuring that GPU processing is enabled.

  3. Black-Scholes Equation - Deep Galerkin Method.ipynb contains the TensorFlow implementation of the layer and model described in the DGM paper as well as the code to perform the training method described in the paper. Using Colab with GPU is also recommended here.

See Project.pdf for more details.