NeST-BO: Fast Local Bayesian Optimization via Newton-Step Targeting of Gradient and Hessian Information

This repository contains the code to reproduce the NeST-BO and NeST-BO-sub algorithms proposed in the paper NeST-BO: Fast Local Bayesian Optimization via Newton-Step Targeting of Gradient and Hessian Information.

NeST-BO has been published as a conference paper in AISTATS 2026.

Installation

pip install -r requirements.txt

Running Experiments

Experiments can be run using the main_NeSTBO.py and main_NeSTBO_sub.py script. You must specify a benchmark to run the algorithms.

Basic Command

python main_NeSTBO.py benchmark=<benchmark_name>
python main_NeSTBO_sub.py benchmark=<benchmark_name>

• To see a list of available benchmarks, run python main_NeSTBO.py.
• Adding seed=<number> is recommended for reproducibility.

Configuration Overrides

All default settings are stored in configs/default.yaml. Since this project uses Hydra, you have the flexibility to modify these values on the fly via the command line without editing the file.

# Example: override the evaluation budget for the ackley benchmark
python main_NeSTBO.py benchmark=ackley seed=0 benchmark.n_tot=1000

Citation

If you use this code in your research, please cite the following paper:

@article{tang2025nest,
  title={NeST-BO: Fast Local Bayesian Optimization via Newton-Step Targeting of Gradient and Hessian Information},
  author={Tang, Wei-Ting and Kudva, Akshay and Paulson, Joel A},
  journal={arXiv preprint arXiv:2510.05516},
  year={2025}
}