grid-diamondcutter

What this is

grid-diamondcutter is an open-source Python toolkit for measuring how efficiently the control layer of a power grid is routing electricity. It ships with a small toy grid you can run locally, and a contract you can use to plug a real grid model in behind it.

It is not a replacement for MATPOWER, PYPOWER, OpenDSS, GridLAB-D, or PSS®E. It sits on top of them and asks a different question: under realistic operating cycles, what does this grid’s behavior look like, and how confident can we be in any one number we report?

Why look at it

The unusual thing about this project is how it’s written down, not what it computes.

Every measurement claim in the repository was committed to a public file before the measurement was actually run. The prediction was written. The condition that would prove the prediction wrong was written. Then the code was run. Then the result was attached.

That’s the same discipline clinical trials use to keep researchers honest: you can’t change your prediction after seeing the answer, because everyone has the original.

There are currently 58 such measurements on the record. About 40% of them produced the predicted result. About 60% did not. Both kinds are kept. A failed prediction is the project working as intended, not a defect — the same way a clinical trial that ends with “the drug didn’t work” is still a useful trial.

Who this is for

Grid engineers

You work with real grid models (PYPOWER, pandapower, OpenDSS, MATPOWER). You want to see if the cycle-walk pattern in here applies to your topology before integrating. Five-minute look →

Researchers

You want to evaluate the methodology, not just the code. Every prediction and every kill condition is on the record. So is every failed prediction. Thirty-minute walkthrough →

Security and audit

You want to verify the claims. Every number on the site is computed from a file in the repository, and every file has a verifiable hash. Audit walkthrough →

Open-source contributors

You want to add a new measurement. There is a template, a test, and a short authoring guide. Contributor walkthrough →

What’s on this site

This site is a read-only window onto the repository. Every page on it is rebuilt from the repository’s contents whenever someone pushes a change to it.

• The measurements — every single measurement on the record, the prediction it made, whether it passed or failed, and a link to the file that produced it.
• What the project does not claim — the boundaries the project has committed to in writing, so a reader knows what it is not saying.
• Three historical events — the real-world events the methodology was tested against (Texas 2021, EU 2022, Japan 2011, plus three more under the monetary-substrate extension). Each has a thirty-second audio rendering of the model trajectory; the audio is not a recording of the real event, just the model’s behavior.
• Methodology background — the longer-form theory for readers who want to know how the pieces fit together.

What this is not

This site does not give investment advice. It does not predict what any specific real power grid will do. It does not replace a production grid simulator. The full list of things the project has committed in writing not to claim is on the boundaries page.

The repository is Apache 2.0. The source of truth is always github.com/eir-inc/grid-diamondcutter.