This repository contains the code used to run the experiments in the ICLR 2025 paper CONGO: Compressive Online Gradient Optimization. Each top-level folder in the repository corresponds to a different group of experiments, and each has different software requirements. For instructions on how to set up and run the experiments in each group, see the ReadMe file in the corresponding folder.
In order to implement CONGO-Z, which is built upon the ZORO algorithm, we have borrowed some code from the repo for that paper (found here). Furthermore, the Jackson network simulations make use of the Python package queueing-tool. Finally, we use the social network application from the DeathStarBench suite to test the performance of CONGO algorithms applied to autoscaling.