BACI-VI-Bench: A Dataset of Variational Inequality Benchmark Instances for Multi-Agent Trade-Network Equilibrium

BACI-VI-Bench is a processed benchmark dataset for variational inequality (VI) and multi-agent reinforcement learning (MARL) studies of multi-commodity trade-network equilibrium. The dataset is derived from CEPII-BACI HS17 V202601 bilateral trade records (years 2017–2024). The repository provides:- 8 year-level VI benchmark instances (dimension d=500: K=5 sectors, m=10 exporters, n=10 importers)- 5 sector-level instances for 2022 (dimension d=100)- Calibrated Nagurney-style trade-network operators and feasible sets- Observed and normalized trade-flow tensors- Projection residual diagnostics- Benchmark characterization data for four solvers (EG_spectral, EG_oracle, IEG, SAISE)- Python construction and validation scripts- Publication-quality figures This dataset complements a companion research article on self-adaptive inertial extragradient methods submitted to Computational and Applied Mathematics (2026). Source data: CEPII-BACI HS17 V202601. Raw BACI files are not included; see raw_source_manifest/baci_source_manifest.csv for download instructions from CEPII (www.cepii.fr).