About the work

I am an entrepreneur and AI-native builder, born in the Czech Republic, without formal physics training. I direct the work and am responsible for its claims, corrections and limitations. “Independent researcher” is the affiliation used in the papers, not a claim to an academic qualification.

This research and manuscript are AI-generated under my direction. Language models performed algebraic construction, numerical implementation, literature retrieval, drafting and adversarial checking. Independently implemented arithmetic checks separate a calculation from its interpretation. They test supplied RG polynomials rather than independently deriving all diagrammatic counterterms. Agreement between models is not evidence of a physical law.

The October methods appendix distinguishes fixed inputs, calculation, documentary checks and physical premises. Registration dates are internal provenance, not automatically public priority. Retrospective comparisons stay labelled. Nearly all of the work was checking, but checking cannot fill a missing physical mechanism.

This is the first result of the larger system I’m building — the first one out, not the last. The scientific goal here is a physically normalized source and falsifiable observations—not more expressions fitted to the same number. The unresolved bridges are published beside the calculations.

arXiv declined the May paper in May 2026. Journal of Physics Communications desk-rejected it without review in July 2026. Neither paper is peer-reviewed; a Zenodo DOI is a stable record, not validation.