BlockProx v1.0.0

Reference Julia implementation of RandomEdge (the graph-guided specialization of BlockProx) accompanying the paper "Decentralized Optimization with Topology-Independent Communication" (Lin, Kuang, Alacaoglu, Friedlander),
arXiv:2509.14488.

This release reproduces every figure in the paper.

Contents

• src.jl — RandomEdge / BlockProx implementation, documented inline
• runcode_LS.jl — synthetic least-squares benchmarks (§6.1)
• runcode_strongly_convex.jl — linear convergence to a neighborhood under a Tikhonov regularizer (§6.1)
• runcode_m_scaling.jl — empirical neighborhood radius vs. edge count m = |E| (§6.1)
• runcode_housing.jl — Sacramento housing dataset application (§6.2)
• Manifest.toml — exact package versions (Julia 1.12.3) for reproducibility

Reproducing

Requires Julia ≥ 1.11. Pkg.activate(".") then Pkg.instantiate(), then run any runcode_*.jl. See README.md.