Code

ADEF
Implementation and experiment code for Accelerated Distributed Optimization with Compression and Error Feedback, including JAX synthetic experiments and PyTorch CIFAR experiments for ADEF.

EControl-DA
PyTorch implementation of the EControl-DA algorithms, with FashionMNIST logistic-regression and synthetic softmax experiments for Composite Optimization with Error Feedback: the Dual Averaging Approach.

MoTEF
Python experiment code for Towards Faster Decentralized Stochastic Optimization with Communication Compression, including synthetic experiments, logistic-regression, and MLP experiments. It also includes symbolic verification codes for some of the theoretical analysis in the paper.

Laplacians.jl
A Julia package maintained by Daniel A. Spielman with contributions from several contributors. I implemented the AC(k) solver variants studied in AC(k): Robust Solution of Laplacian Equations by Randomized Approximate Cholesky Factorization.

SDDM2023
Reproducibility resources for the Laplacian/SDDM solver experiments in the AC(k) paper.