CoTenN: Constrained Optimization with Tensor Networks

This artifact provides a complete reproducible environment for the experiments presented in our paper on efficient tensor network algorithms using compressed orthogonal projections. It contains the following: Artifact Components: Pre-built Docker image (cotenn_artifact.tar) with Ubuntu 22.04, Python 3.12.2, Julia 1.12.1, and all dependencies for x86 architectures. Source code and experiment scripts in artifact/ Reference results for validation and comparison Dockerfile for building from source README.md with complete documentation Quick Start # Extract the artifact tar -xf artifact.tar # Load the pre-built Docker image docker load -i cotenn_artifact.tar # Run all experiments (~10 hours for complete results) docker run --rm -v "$(pwd)/results":/cotenn_artifact/results cotenn_artifact ./run_all.sh --all # Compare against reference results diff results/table2_half.csv reference_results/table2_half.csv System Requirements Minimum: Docker, 8 GB RAM, 4 CPU cores Recommended: 16 GB RAM, 10 cores Tested on: 20 logical CPUs (10 cores), 15.46 GiB memory The artifact supports both a quick validation mode (~2 hours) and a complete reproduction mode (~10 hours). Documentation See README.md for detailed instructions on: Running individual experiments Controlling parallelism Understanding output structure Troubleshooting common issues