Notice: At present, bot7 requires use of updated Torch7 features not yet merged into torch/torch7. For the time being please use: https://github.com/j-wilson/torch7.

# bot7 # __bot7__ is a framework for Bayesian Optimization implemented in Torch7. For it's companion package on Gaussian Processes, please refer to [gpTorch7](https://github.com/j-wilson/gpTorch7). This package is currently still in an alpha-build stage; so, please feel free to pass along suggestions and/or feedback. ## Package Content ##

Directory	Content
scientists	Experiment runners
models	Target function models
scores	Acquisition functions (incl. EI, UCB)
grids	Candidate grids
samplers	Sampling methods (e.g. slice sampling)
examples	Demos / Benchmark Functions

Installation and Usage

You can install bot7 via luarocks provided by the torch distribution.

git clone https://github.com/j-wilson/bot7
cd bot7
luarocks make

After installing bot7, you can load the package inside Torch via:

require('bot7');
th> bot7.
bot7.grids.       bot7.models.      bot7.samplers.    
bot7.scientists.  bot7.scores.      bot7.utils.

Developers' Notes

Please be sure to grab an updated copy of Torch7 prior to using bot7, as the package relies on newer functions (e.g. torch.potrf/s).

Further documentation coming soon.