Supernova success and sound-horizon failure of Everpresent Λ: an independent test of stochastic dark energy

Abstract

Independent full-scale reproduction and observational follow-up of the Everpresent Λ Model 1 rare-supernova-tail result. Two independent implementations recover the same six Pantheon+SH0ES winners in 90,000 seeds under two redshift conventions. Under the stated calibrated background sound-horizon prescription, all six histories are strongly disfavored by the joint supernova, cosmic-chronometer, and DESI DR2 baryon-acoustic-oscillation likelihood. This is a conditional result for the six selected histories. It is not a universal Model 1 no-go theorem and does not replace a self-consistent recombination and perturbation calculation. This non-peer-reviewed preprint record includes the eight-page manuscript, isolated REVTeX source, a pinned reproducibility archive, and a SHA-256 manifest. The audited public code repository is https://github.com/romainsimon/everpresent-lambda-adjudication. Automated assistance and its limitations are disclosed in the manuscript. A plain-language illustrated explanation is available at Yuki Capital.