Contact
Email is the best way to reach me. I read everything, but between school, sports, and research, replies can take a while.
Get in touch
Questions about my projects, corrections, pointers to things I should read, or anything else on this site.
FAQ
The solar code is mine and it is MIT licensed, so you can read it, run it and build on it: quantum-solar. The circuit repository is different. ecdsa-circuit-optimization is my fork of the challenge, so most of the code in it belongs to other people and I added no license of my own, because you can only grant permission for work you own. Its NOTICE file says who owns what.
Yes, on the solar repository: bug fixes, test cases, other algorithms, better figures. For the circuit work, send anything to the competition repository upstream, not to my fork, which is where my own three findings went.
Run python -m quantum_solar. It prints the figure from data committed to the repository, with no key and no network, and because the classical solvers need only numpy, nothing quantum has to be installed. The demo notebook rebuilds the same result three independent ways, and the arithmetic behind it fits in one paragraph on the project page.
Generated from the raw result files, hardware-jobs.md lists every IBM job ID with its date, shot count and the device calibration at submission. The measurement counts themselves are committed in docs/results, with the job IDs inside. Every run's plan was committed before the run, in docs/plans, and PREDICTIONS.md scores each registered prediction against the write-up that resolved it.
The score comes from ./benchmark.sh in the fork. While results.tsv logs every benchmark run with the commit it was measured on, the failure-rate figures come from the raw per-trial data in docs/data. The screening tool those measurements used was checked end to end against the official checker, a validation that its README shows.
Sign up for IBM Quantum's free tier, the Open Plan, at quantum.cloud.ibm.com. It gives about 10 minutes of real machine time a month, which is more than this whole project used. Run it yourself, free walks through the account, the catches, and sending a first circuit to a real machine and reading what comes back.
Three public datasets, all from U.S. Department of Energy sources: NREL PVWatts for hourly sunlight, the OpenEI Utility Rate Database for Xcel Energy's time-of-use prices, and NREL ResStock for household electricity use. The first two are fetched live with a free NREL key, while the ResStock profiles are kept in the repository as four small CSV files. The data behind the numbers says what each one is, why it is used, and how to get it yourself.
Yes. The solar repository is archived on Zenodo under one concept DOI, 10.5281/zenodo.22105805, which always resolves to the newest release. CITATION.cff at the repository root holds the metadata.Amissah, A. (2026). quantum-solar: an exactly-solved battery-scheduling instance for QAOA method work, with a slack-free encoding. Zenodo. https://doi.org/10.5281/zenodo.22105805
Email me. What was wrong stays published and marked wrong rather than deleted. The corrections page shows what that looks like for the solar project, and the ECDSA conclusion that turned out wrong is still in the repository with each of its claims marked.
If you are working on something related, or you spot a mistake in my work, I would like to know.