Recent developments in HNCO
Arnaud Berny · Proceedings of the Genetic and Evolutionary Computation Conference Companion · 2022
We present the recent developments in HNCO which started as a C++ framework for the optimization of black box functions defined on bit vectors. Beyond changes in the library, we emphasize three new features: representations for data types other than booleans (integers, real and complex floating-point numbers, categorical values, and permutations), multi-objective optimization, and Python bindings. With the help of a command-line application, compiled once, it is now possible to define at runtime and optimize a wide range of functions. Those functions can be written as mathematical expressions or plain Python code. We also present the experiments available in HNCO which are built upon the same command-line application. Most experiments can be run in parallel, locally or remotely.