Efficient Algorithm for Sparse Fourier Transform of Generalized $q$-ary Functions

GFast is a an efficient algorithm to compute the sparse Fourier transform of generalized $q$-ary functions. This repository contains code from the article "Efficient Algorithm for Sparse Fourier Transform of Generalized $q$-ary Functions" by Darin Tsui*, Kunal Talreja*, and Amirali Aghazadeh. A link to the paper can be found here.

* Equal contributions

Quick start

For an overview of GFast, as well as how to run the algorithm for your own functions, see tutorial/tutorial.ipynb.

Description of folders

gfast/: GFast module for generalized $q$-ary functions.

tabular_exp/: Code needed to run NR-GFast for the multi-layer perceptron (MLP) trained on the heart disease dataset found here. More information has been left in the folder.

tabular_results/: Folder containing the results of the heart disease experiments.

gfp_exp/: Code needed to run NR-GFast for the multi-layer perceptron (MLP) trained on the avGFP protein found in this paper. More information has been left in the folder.

gfp_results/: Folder containing the results of the GFP experiments.

synt_exp/: Code needed to run GFast and NR-GFast on synthetic data.

synt_results/: Folder containing the results of the synthetic experiments.

tutorial/: Folder containing the tutorial for running GFast.

Figures

All figures in the paper and the code to generate them are found in figures.ipynb.