The 'GPR' folder contains three subfolders: Code, Data and Figures
- The 'Code' folder contains a demo of the MATLAB code I used in the Soft Matter paper linked below:
- 'GPR_Predict.m' predicts the structure characteristics of a phase-separating (demixing) polymer blend from the corresponding scattering data
- 'Pred_vs_True_Plot_Best.m' uses the output of 'GPR_Predict.m' to plot a comparison of the best predicitons of each structure characteristic with the true values
- The remaining code belongs to the GPML MATLAB code package: http://gaussianprocess.org/gpml/code/matlab/doc/
- The 'Data' folder contains two subfolders:
- The 'Input' folder contains the data required to run 'GPR_Predict.m'
- The 'Output' folder contains the output of 'GPR_Predict.m' (need to run 'GPR_Predict.m' first)
- The 'Figures' folder contains the figures produced by 'Pred_vs_True_Plot_Best.m'
Soft Matter paper: https://pubs.rsc.org/en/content/articlelanding/2021/sm/d1sm00818h/ [Jones, Clarke; Soft Matter, 2021, 17, 9689]
For specific details regarding our implementation of Gaussian process regression, including an overview of the data we used, please refer to sections 3.1 and 3.2 of the paper
The 'Simulations' folder contains a demo of the Julia code I used to generate the data for my PhD research:
- 'SD_Scattering_Parallel_3D.jl' simulates spinodal decomposition in 3D, making use of parallel processing where possible, and calculates the corresponding scattering data at regular intervals
- 'Parameters.sh' and 'Run_Simulations.sh' are batch submission scripts used to run 'SD_Scattering_Parallel_3D.jl' on the University of Sheffield's HPC, Stanage (https://docs.hpc.shef.ac.uk/en/latest/stanage/index.html)
- 'Parameters' specifies the blend and simulation parameters
- 'Run_Simulations' specifies the number of repeat simulations, ensuring they are sent to different nodes
For details regarding the scientific model on which the simulations are based, please refer to [Glotzer, 'Computer Simulations of Spinodal Decomposition in Polymer Blends', Annual Reviews of Computational Physics II, 1995]