What load flexibility, financial cover and price forecasts are worth to an AI data centre. An MQFS working paper by Nicolò Angileri.
Each page carries real fitted objects from the paper rather than pictures of them, recomputes something in your browser, and reports how far its own result lands from the reference computed offline. Each is a single self-contained file.
Simulated price years grow in three dimensions while the distribution of the annual bill builds beside them. Move the forward cover, the cap and the flexible share while it runs; the schedule is a JavaScript port of the paper's receding-horizon rule.
Open the pageThe weights of the frozen quantile network and test days it never saw. Redraw the last week of prices by hand and watch the channels and the predictive fan respond, with the forward pass checked against PyTorch.
Open the pageThe quantum state that encodes the cost distribution, amplified one Grover step at a time, with the iterative estimator running live against the exact amplitude and plain Monte Carlo.
Open the page