Recent preprints

Unreviewed method papers (Bayesian optimization is the contribution). Capped at 20, newest first. After peer review, open a PR to move an entry into the matching Papers section. Application papers are dropped on sight.

• GRAPE: Gradient Refinement and Progress-Aware Exploitation for Query-Efficient High-Dimensional Bayesian Optimization - 2026. High-dimensional BO with gradient refinement.
• BOCoDe: Engineering-Centered Benchmarking for Bayesian Optimization - 2026. Engineering BO benchmark.
• Adaptive KappaSharp: Condition-Number Shaping for Preferential Bayesian Optimization - 2026. Preferential BO via condition-number shaping.
• Constraint-Bound Agnostic Bayesian Optimization: One Model for All Thresholds - 2026. One constrained-BO model across thresholds.
• How Many Initial Points Does Bayesian Optimization Need? - 2026. Design of the initial design.