Geodesic-informed neural networks for testing spacetimes with S2
Source check: Source checked. Not independent scientific confirmation.
Finding: Across five tested alternatives, the fit improvement over Schwarzschild did not offset the evidence penalty. The reported log Bayes factors span approximately -3.4 to -0.8 for the stated priors and dataset. Verified against an exact numerical solver and checked posterior approximation in appendices. The authors used existing observations without the newer GRAVITY astrometry; prior choices and data precision limit conclusions.
Technology: A trained geodesic-informed differentiable network represented each orbit to fit published S2 astrometry and line-of-sight velocities and obtain likelihood derivatives for model comparison.
Publication: Preprint · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- A trained geodesic-informed differentiable network represented each orbit to fit published S2 astrometry and line-of-sight velocities and obtain likelihood derivatives for model comparison.
- Test and result
- Across five tested alternatives, the fit improvement over Schwarzschild did not offset the evidence penalty. The reported log Bayes factors span approximately -3.4 to -0.8 for the stated priors and dataset.
- Limitation / still unknown
- Verified against an exact numerical solver and checked posterior approximation in appendices. The authors used existing observations without the newer GRAVITY astrometry; prior choices and data precision limit conclusions.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://arxiv.org/html/2611.24370v1
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 22 Nov 2026. Review scope: Primary source full text Methods and Results; version/date checked against bibliographic record..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 3/5 · Validation 2/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.