Seven DESI DR2 distances reject expansion with no past acceleration at p = 0.0142 in a Gaussian FLRW model with one fixed ruler and ΩK ≥ −25/9

We test expansion with no past acceleration using seven public DESI DR2 BAO distances. The test assumes FLRW geometry, a fixed common comoving ruler, and the released Gaussian point-distance model. It allows spatial curvature ΩK ≥ −25/9. A closed model then has curvature radius at least 0.6 c/H0; open curvature is unrestricted. We apply Sutherland's known BAO inequality as six linear contrasts. Each contrast uses the isotropic distance at redshift 0.295 and one higher radial distance. We retain their shared measurement error and compute the sharp worst-case Gaussian tail for the largest violation. The maximum is 2.8091 standard errors at redshift 0.934. Correction for all six radial bins gives p = 0.014211. A Bonferroni upper bound is 0.014906. The same construction gives a simultaneous 95% upper confidence limit of -0.01264 for the minimum deceleration parameter over 0 ≤ z ≤ 0.934. A separate 50-digit calculation and one million full-source Gaussian simulations support the calibration. The evidence is modest. Omitting the redshift 0.934 radial bin gives p = 0.647. Increasing all standard errors by about 19% removes rejection at the 5% level. The curvature bound is imposed, not measured. The exploratory result does not locate acceleration or measure it today. The contribution is a DR2 application and probability calibration, not a new physical inequality or the first BAO evidence for acceleration. The package preserves exact source data, provenance, code, proof, search records, results, figures, and reproduction instructions.

DESI DR2 · baryon acoustic oscillations · cosmic acceleration · spatial curvature · cosmic speed-trap · Gaussian contrasts · correlated maximum · reproducible research