Skip to content

Data and licenses

What ships with the package

Data Contents API
ACTIVSg2000 regions 150 bus pairs, 35 batches of two or three buses, 8 location multisets, exact activsg2000(buses), bundled("activsg2000")
ACTIVSg2000 reference results max-min fair points, price of anarchy and its bound for 25 batches of three and five buses; max-min fair points of all 216 placements of the location game, all computed with DCOPFRegion reference("activsg2000", "poa_implicit"), reference("activsg2000", "location_w"), activsg2000_location()

Every region file records its provenance: - the buses and periods; - the method and tolerance; - the SHA-256 hash of the input scenario; - whether counterflow appears in the exact facets.

Built on your machine: IEEE 14-bus

The IEEE 14-bus regions are not shipped. On first use, ieee14(buses) does four things: 1. It downloads the public case file case14.m from MATPOWER's repository (release 8.1, pinned). That file is a conversion of the University of Washington archive's IEEE Common Data Format file. 2. It checks the file's SHA-256 hash. 3. It builds the scenario in a local cache (ICGYM_DATA_DIR, default ~/.cache/icgym), adding the package's own data. 4. It exports and caches the region of any set of buses, in a few seconds each.

The package's own data, in icgym/data/overlays/ieee14.json (MIT), is: - the thermal limits, which the case lacks; - linear generator costs; - the hourly load multipliers of the five representative days.

The case file supplies the topology, reactances, base loads, and generator sizes. The rebuilt scenario is fixed byte for byte, and a test checks its file hashes.

Without network access, download case14.m yourself and pass case_file=... to icgym.datasets.fetch_ieee14, or set ICGYM_IEEE14_CASE. icgym.datasets.fetch_ieee14() returns the scenario folder, for icgym-export --data and icgym-compare --data.

The scripts that rebuild the catalogs are scripts/build_catalog.py and scripts/analyze_catalog.py, run with the bus specifications in scripts/.

Where the data comes from

  • ACTIVSg2000:
  • Source: the synthetic 2,000-bus Texas grid of the Texas A&M Electric Grid Test Case Repository (https://electricgrids.engr.tamu.edu/electric-grid-test-cases/activsg2000/; mirror https://github.com/caseformat/ACTIVSg2000).
  • Scenario: the regions are computed from a scenario derived from that case, with five representative days of hourly demand (120 periods). The bus demands are the case's values times a seasonal, daily, and weekly multiplier. Generators and branch limits are the case's.
  • Not real data: the scenario is synthetic. It is not historical ERCOT data, an official model, or a forecast.
  • IEEE 14-bus:
  • Source: the IEEE 14-bus test case of the University of Washington Power Systems Test Case Archive.
  • Scenario: a representative-day scenario with the same five days and a scaled hourly load.

Licenses and attribution

  • Code: MIT (see LICENSE).
  • ACTIVSg2000-derived data:
  • License: the repository's mirror of the case is licensed under Creative Commons Attribution 3.0 Unported (CC BY 3.0). The Texas A&M repository describes the case as free for commercial and non-commercial use.
  • Adapted: the bundled regions are adaptations. The demand profile was changed as described above, and the regions are computed results, not the case files.
  • Citation requested: A. B. Birchfield, T. Xu, K. M. Gegner, K. S. Shetye, and T. J. Overbye, "Grid Structural Characteristics as Validation Criteria for Synthetic Networks," IEEE Transactions on Power Systems, vol. 32, no. 4, pp. 3258-3265, July 2017, doi:10.1109/TPWRS.2016.2616385.
  • IEEE 14-bus case: not redistributed. You download it from its source on first use. The archive states no license and describes its material as provided for research and education, not for commercial gain. Check the terms of the sources before commercial use. Our additions are MIT.

Verification

  • ACTIVSg2000: every bundled batch matches an independent computation by the accompanying paper. The max-min fair points agree to 3e-9 MW and the price of anarchy to 4e-12, with no violation of the price-of-anarchy bound.
  • Location game: the shares give the paper's 3 pure equilibria and 1,617 improving moves, all raising the sorted shares.
  • IEEE 14-bus: the regions reproduce the paper's max-min fair points and price of anarchy to 0.01 MW.