The code repository for Python programmes used for the Final Year Project (FYP).

1. Data_Generating_NoteBooks folder: Python notebooks to generate 2D datasets for DeepMod

   • Diffusion_noisy.ipynb: Notebook to produce noisy 2D data for Diffusion

   • Diffusion_clean.ipynb: Notebook to produce clean 2D data for Diffusion

   • Diffusion_Decay_defaultBC.ipynb: Notebook to produce clean/noisy 2D data for Diffusion-Decay with default boundary condition as stated in the FYP thesis (sections 5.2.1 to 5.2.2 of thesis).

   • Diffusion_Decay_bicoidSurge.ipynb: Notebook to produce clean/noisy 2D data for Diffusion-Decay with alternative boundary condition 1 as stated in the FYP thesis (section 5.2.3 of thesis).

   • Diffusion_Decay_flux.ipynb: Notebook to produce clean/noisy 2D data for Diffusion-Decay with alternative boundary condition 2 as stated in the FYP thesis (section 5.2.3 of thesis).

2. DeepMod_Training folder: Python programmes to train Deepmod algorithm

   • deepmod_original.py: Software pipeline to load data to train DeepMod.

   • gpu_util.py: Programme to search for GPU resources within MBI's computing infrastructure for Deep Learning.

   • parameters.py: Programme containing DeepMod's neural network architecture and Deep/Machine Learning hyper-parameters.

   • pde_util.py: Programme containing library of PDE terms for DeepMod and other relevant library parameters.

   • result_util.py: Programme to store results after training DeepMod.

3. Algorithm1.py: Python programme which contained an enhanced procedure of DeepMod's thresholding (section 5.3.1 of thesis).

4. Figure31.png: Image of Figure 31 for section 6.1 of the FYP thesis.

5. LR_expt folder: LR means Lasso Regression. Python notebook which discusses DeepMod's normalisation issues (section 5.3.2 of the thesis).

   • LR_expts.ipynb: Demo for section 5.3.2 for the thesis.

   • sparse_pattern_deepmod.npy: Bit mask generated by DeepMod.

   • theta.npy: Theta matrix.

   • time_deriv.npy: du_dt vector.

1. Python programming language version: 3.7.4.
2. Tensorflow version: 1.14.0.
3. Numpy version: 1.17.5.
4. Scikit-learn version: 0.22.1.
5. Pandas version: 1.0.0.
6. DeepMod (written in Tensorflow, Python): https://github.com/PhIMaL/DeePyMoD.

Ho Wei Chin (National University of Singapore)

Tricia Loo, for the Forward-Time-Central-Space Python programme.

Gert-Jan Both et al, for their DeepMod algorithm.