Skip to content

This repository contains the source code for my MSc Project on "Scalable Inference for Generative Models using Quasi-Monte Carlo" at the Department of Statistical Science, UCL.

Notifications You must be signed in to change notification settings

johannnamr/MSc-Project-Code

Repository files navigation

MSc-Project-Code

This repository contains the source code for my MSc Project on "Scalable Inference for Generative Models using Quasi-Monte Carlo". It is optimised for the use with Google Colab and mounts the Google drive to load required other notebooks located in the default folder for Colab notebooks. For local use, file paths have to be adjusted where indicated.

Content of sub-folders:

Illustrative examples

This folder contains a Python notebook generating all other illustrations for MC, QMC, and RQMC.

Helper functions

This folder comprises two Python notebooks with the necessary functions to obtain results for optimisation/convergence ("utils.ipynb") and plots ("Plot_fcts.ipynb") for all analysed models. Therefore, both notebooks are loaded in all other notebooks listed below so that their file paths might need to be adjusted where indicated.

Gaussian location model

In this folder, all results for the Gaussian location model can be found:

  • Gaussian_check.ipynb: test of kernel, generator and all partial derivatives
  • Gaussian_optim.ipynb: optimisation procedure using MC, QMC and RQMC
  • Gaussian_conv.ipynb: convergence of the squared MMD using MC, QMC and RQMC
  • Gaussian_conv_W.ipynb: convergence of Wasserstein distance using MC, QMC and RQMC
  • Gaussian_conv_sink.ipynb: convergence of Sinkhorn loss using MC, QMC and RQMC

Beta distribution

This folder provides the results for the beta distribution:

  • beta_check.ipynb: test of kernel, generator and all partial derivatives
  • beta_conv.ipynb: convergence of the squared MMD using MC, QMC and RQMC with options Halton, Sobol or lattice point sets

G-and-k distribution

This folder contains all results for the g-and-k distribution:

  • gandk_check.ipynb: test of kernel, generator and all partial derivatives
  • gandk_optim.ipynb: optimisation procedure using MC, QMC and RQMC
  • gandk_conv.ipynb: convergence of the squared MMD using MC, QMC and RQMC

Stochastic volatility model

This folder comprises the results for the stochastic volatility model. The results for this generative model were not used in the final report:

  • sv_check.ipynb: test of kernel, generator and all partial derivatives
  • sv_optim.ipynb: optimisation procedure using MC, QMC and RQMC

About

This repository contains the source code for my MSc Project on "Scalable Inference for Generative Models using Quasi-Monte Carlo" at the Department of Statistical Science, UCL.

Topics

Resources

Stars

Watchers

Forks