How much should you trust these numbers?
Two answers. The validation matrix shows how each model is checked: against published results, against an independent tool, or against a known answer. The uncertainty lab propagates honest parameter ranges through the headline results, live in your browser.
1 · Validation matrix
Bars show the deviation from the reference as a fraction of the allowed tolerance; the white line is the tolerance. Markdown version: docs/VALIDATION.md.
2 · Uncertainty lab
Tornado: each input alone at its low (gold) and high (teal) value
Scientists don't know every number exactly. So we roll the dice thousands of times, each time picking values inside the range we believe, and see how the answer spreads out. The tornado shows which unknown number matters most.
Latin-hypercube sampling covers each parameter's range evenly with few samples. P10, P50, and P90 are the values below which 10, 50, and 90% of outcomes fall. A tornado diagram varies one input at a time; the widest bar is the one to measure first.
Uniform or log-uniform priors over the stated ranges, independent inputs, and the same models as adamas.uncertainty; the Python Monte Carlo gives P50 ≈ 3.8× (power) and ≈ 3.6 years (quantum). Ranges are this repository's judgment from the cited spreads; correlated inputs (critical field and mobility both improve with crystal quality) would widen the tails.
Sources: [isberg2002] [wort2008] [donato2020] [kimoto2014] [hopper2018] [maurer2012] [dolde2013] [litinski2019]