SunBURST : Deterministic GPU-Accelerated Bayesian Evidence via Mode-Centric Laplace Integration
Ira Wolfson, [1ex] Department of Electronics, Electrical EngineeringBraude College of Engineering, Israel · 2026
Read SunBURST : Deterministic GPU-Accelerated Bayesian Evidence via Mode-Centric Laplace Integration by Ira Wolfson, [1ex] Department of Electronics, Electrical EngineeringBraude College of Engineering, Israel (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-19957.txt.
Description
Ira Wolfson introduces a deterministic algorithm for a core quantity in Bayesian statistics. The Bayesian evidence measures how well a model explains data, averaged over its parameters. In high dimensions, computing it by sampling becomes far too slow. Sampling also runs step by step, so parallel hardware helps little.
SunBURST replaces global exploration with mode-centric integration. It locates the modes, the peaks of the posterior, and integrates around them with a Laplace approximation. The algorithm is built to run natively on graphics processors, which supply the parallelism sampling lacks. Deterministic output means the same answer every run, easing verification. Faster evidence would sharpen model comparison in fields from astronomy to medicine. The gain is speed with no loss of trust. Each run gives the same score. That is rare for this kind of task. The peaks are found first, then summed. No dice are rolled at any step.
Catalog record
| CREATOR | Ira Wolfson, [1ex] Department of Electronics, Electrical EngineeringBraude College of Engineering, Israel |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 100 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-19957