Skip to content

Latest commit

 

History

History
58 lines (40 loc) · 2.42 KB

File metadata and controls

58 lines (40 loc) · 2.42 KB

IX-Fusion Proof of Concept

Release 0.1 tests the C6-derived seed against a matched five-field-period helical control using the repository's dimensionless reduced-order model and equal optimizer budget.

Verdict

FAIL_OR_INCONCLUSIVE

Scientific stage remains geometry_hypothesis.

Why the result is not simply "bad"

The C6 candidate produces a 28.65% lower mean radial-excursion proxy than the matched control in the internal field-line screen. That is the most interesting favorable signal in release 0.1.

However, the candidate also has:

  • a 4.87% worse composite screening objective;
  • a 0.17% worse bounce-action variation proxy;
  • a 10.47% higher engineering burden proxy, narrowly exceeding the predeclared 10% allowance;
  • the same zero escape fraction in the tested reduced field-line set.

Because the project uses multiple independent gates, the favorable radial-excursion result does not override the other failures.

Additional evidence

The six-source phased-actuator signal model shows that abstract feedback increases median target spatial-mode purity from approximately 99.21% to 99.93% under the committed error assumptions, corresponding to an approximately 11.05x reduction in median unwanted-mode power in that signal model.

This is not evidence of RF heating or plasma control. It only verifies the control-layer mathematics under the declared abstract assumptions.

The geometry-error Monte Carlo also shows a mixed result: the C6 candidate has slightly lower median relative degradation than the matched baseline under the committed parameter perturbation distribution, but the candidate begins from a worse nominal composite score.

Important model warning

The no-axis-helical-shaping ablation improves the internal composite objective. That is not a physical recommendation. It exposes a known limitation: rotational transform is parameterized in the reduced model instead of being derived self-consistently from 3-D coils/equilibrium. The result is retained as a falsification signal showing why a proper MHD/coil solver is mandatory before any stronger claim.

Next decision

Do not tune the internal model until C6 "wins." The next scientifically meaningful step is to integrate a solved equilibrium/coil pipeline and retest the same hypothesis against fair controls.

Exact evidence: results/poc/POC_RESULT.md, results/poc/verdict.json, and results/evidence/IXFUSION-POC-001.json.