The Reproducible Analysis Record

Every physics claim about the breeder (Hyperion) traces to one of 81 versioned analyses that regenerate from source, so a reader can reproduce the number, not just read it.

One record, 81 analyses

The breeder (Hyperion) is a compact D-T spherical tokamak studied in simulation. Its design point is not a slide deck of round numbers; it is a set of 81 analyses that each take fixed inputs, run defined code, and emit a defined output. The headline figures — scientific gain Q_sci of 3.076, fusion power of 85.0 MW, plasma current of 9.66 MA, peak field of 16.84 T with 8 T on-axis — are outputs of those analyses, not assertions layered on top of them.

What 'reproducible' means here

An analysis is reproducible when a third party, given the inputs and the code, obtains the same output within a stated tolerance. Each of the 81 carries its inputs, the exact code revision, a random seed where stochastic, and an expected output. The record is regenerated end-to-end rather than hand-edited, so a changed input propagates to every figure that depends on it and cannot silently disagree with the text.

Why it is structured this way

A fusion design claim is only as good as the ability of someone else to check it. Publishing prose without the underlying computation asks the reader to trust the author. Publishing the computation asks the reader to trust arithmetic they can rerun. Kronos chose the second posture across the whole breeder program, and the same discipline governs the open honest gates: the tritium-breeding-ratio reconciliation and the center-post lifetime are recorded as open items, not smoothed over.

This page describes a design-and-simulation study, not a built machine. Construction of the breeder (Hyperion) begins Q2 2027; first-of-a-kind first tritium is targeted near 2030. No hardware net-gain claim is made before FOAK.