Verification vs Validation

Verification asks whether we solve the equations right; validation asks whether we solve the right equations. Both are required, and they are different.

Two different questions

Verification and validation are often blurred together, and blurring them is where trust leaks out. Verification is a mathematics question: does the code correctly solve the equations we intended? Validation is a physics question: do those equations describe reality? A model can be perfectly verified and completely wrong, or physically sound but wrongly implemented.

Where each one lives today

Verification we can do fully now: exact test cases, method-of-manufactured-solutions, convergence studies, and unit tests all run without any hardware. Validation is partial by necessity. Some sub-models validate against existing experiments and published data; the integrated machine cannot be validated until it exists. For the breeder that gate is first-of-a-kind (FOAK) first tritium around 2030.

Why the distinction is a standard, not pedantry

Calling a verified-but-unvalidated result a measurement is the single most common way fusion claims mislead. We keep the words separate so a reader always knows which evidence a number rests on. The burner (Aegis / MetroVolt) is, honestly, largely pre-validation: key regimes sit far beyond any operated device, so we present them as verified simulation, not validated performance.

Both halves are mandatory. Verification without validation is untethered mathematics; validation without verification is measuring the wrong thing carefully. We therefore report the two separately for every headline number, so a reader can always see whether a figure rests on a solved equation, a measured reality, or -- as with much of the burner today -- a verified simulation still waiting for the reality that would validate it.