The Frostyverse: Discrete-Grid Candidate for Gravitational Photon Deflection, Shapiro Delay, and Redshift

Version 1.0.1 software correction — 29 September 2026 Version 1.0.1 corrects a finite-path endpoint/wall-count bookkeeping defect in the reviewer-accessible photon fly-by software. When a requested finite-path endpoint fell inside a FrostCell, rounded endpoint indices could disagree with the number of completed wall-to-wall transits, causing otherwise valid higher-resolution runs to abort. The correction makes the Frostyverse and GR comparison terminate at the same requested finite endpoints. No photon steering law, calibrated normalization K, FrostField mechanics, clock mechanics, or other physical model parameter was changed or retuned. The Version 1.0 paper PDF is unchanged. After the software correction, three independently selected photon fly-by test families were extended through R40. In all three, the finite-path deflection residual relative to GR decreased monotonically with increasing resolution. At R40 the residuals were approximately 0.4824%, 0.2597%, and 0.6294%, respectively. Across the tested resolution ladders, the residuals are approximately consistent with a first-order finite-grid effect, roughly proportional to 1/R. This is an observed numerical trend, not a claim that exact convergence to GR has been proven. These follow-up bend-angle tests are distinct from the Version 1.0 paper’s separate R80–R160 maximum physical path-separation audit; the latter remains reported in the original paper exactly as published. _______________________________________________________________ The Frostyverse is a speculative discrete mechanical model in which propagation, deformation, and local interaction occur through a three-dimensional elastic grid rather than a continuous spacetime metric. This preprint tests a frozen Frostyverse gravity-family candidate against three related weak-field observables: gravitational photon deflection, Shapiro time delay, and static gravitational redshift. The photon candidate uses a current-cell steering response proportional to K R² sqrt(current-cell convergence) sin³(theta), with one historically calibrated absolute normalization K that is explicitly disclosed and frozen before later holdouts. The timing sector uses a three-axis FrostCell clock coupling derived from principal logarithmic strains and an independently derived flat-grid clock exponent alpha = 0.5. Frostyverse outputs are generated and sealed before General Relativity is opened as an external comparison ruler in the stated tests. At R160, the worst reported Shapiro difference is 0.1217% and the worst static-redshift difference is 0.2723%. The photon-path separation remains much smaller than one FrostCell but increases over the R80-R160 ladder; the paper therefore preserves that photon residual as unresolved rather than claiming demonstrated convergence. The claim is computational and falsification-oriented. The work does not experimentally validate the Frostyverse, does not claim equivalence with General Relativity, and does not present the remaining local photon law as a complete first-principles derivation. A supplementary reproducibility package contains the final source, a direct reviewer-accessible Python source tree, preserved research archives, canonical outputs, assumptions/firewalls, negative controls, claim-to-file map, hashes, run instructions, and selected provenance.