MGA Compass

Near-optimal solutions cost a small percentage more than the optimum. While slightly expensive, they can be preferable for political feasibility, social acceptance, or environmental impact.

This project contains two parts:

MGA Compass

An interactive platform to explore the near-optimal solutions of an energy system model in real time.

MGA Compass
MGA Surveyor

The implementation of the G-ORACLE method for PyPSA: computes the near-optimal space of a PyPSA model, with a guaranteed bound on how much of it is still unexplored.

MGA Surveyor

MGA Surveyor finds the near-optimal solutions; MGA Compass lets you explore them. The two libraries are independent, and the platform can serve points from any source.

New here? Getting started takes you through both in about ten minutes.

Live demo: enormous-raphaela-eins-tu-darmstadt-215a3282.koyeb.app

Presentations

Trondheim

EEM 2026

presentations/Trondheim/
Freiburg

openmod Workshop 2026

presentations/Freiburg/

Citation

If you use MGA Compass, please cite:

S. Hajikazemi and F. Steinke, “Interactive Exploration of Near-Optimal Solutions of Energy Planning Models,” in 2026 22nd International Conference on the European Energy Market (EEM), 2026, pp. 1–6. doi:10.1109/EEM68581.2026.11589682

@inproceedings{hajikazemi2026compass,
  title     = {Interactive Exploration of Near-Optimal Solutions of Energy Planning Models},
  author    = {Hajikazemi, Sina and Steinke, Florian},
  booktitle = {2026 22nd International Conference on the European Energy Market (EEM)},
  year      = {2026},
  pages     = {1--6},
  doi       = {10.1109/EEM68581.2026.11589682}
}

If you use MGA Surveyor or G-ORACLE, please cite:

S. Hajikazemi and F. Steinke, “G-ORACLE: A Fast Geometric Method To Explore All Near-Optimal Designs for Energy Systems,” arXiv:2610.32759, 2026. arxiv.org/abs/2610.32759

@misc{hajikazemi2026goracle,
  title         = {G-ORACLE: A Fast Geometric Method To Explore All Near-Optimal Designs for Energy Systems},
  author        = {Hajikazemi, Sina and Steinke, Florian},
  year          = {2026},
  eprint        = {2610.32759},
  archivePrefix = {arXiv},
  primaryClass  = {math.OC},
  url           = {https://arxiv.org/abs/2610.32759}
}