New pip Functionalities

HyperMapper v2.2 adds HyperMapper to pip. Users can now install HyperMapper by running:

pip install HyperMapper

Users no longer need to clone the GitHub repository to install HyperMapper. Current users can keep using HyperMapper as before, however, they must make sure that the HYPERMAPPER_HOME environment variable is correctly set:

HYPERMAPPER_HOME=/path/to/hypermapper

Additionally, users no longer need to (and should not) add HyperMapper to their PYTHONPATH.

New pip functionality

After installing via pip, users can keep using HyperMapper as usual by using HyperMapper's optimize method:

from hypermapper import optimizer
optimizer.optimize(configuration_file, black_box_function)

See this file for an example.

Additionally, we provide a number of useful scripts via a new command-line interface for HyperMapper. These command-line options are only available if users install HyperMapper via pip. Also, the scripts are still available in the HyperMapper scripts folder, though they will be removed in a future release.

Quick-start Example

Users can run a simple Branin example to test the installation of HyperMapper using:

hm-quickstart

Client-Server mode

Users can now access HyperMapper's Client-Server mode via:

hypermapper /path/to/configuration_file

Plot optimization results

Users can now plot the optimization results for mono-objective applications using:

hm-plot-optimization-results /path/to/configuration_file

Pareto front

Users can now compute the Pareto front for applications with two objectives using:

hm-compute-pareto /path/to/configuration_file

and plot the Pareto front using:

hm-plot-pareto /path/to/configuration_file

Plot Hypervolume Indicator

Users can now plot the Hypervolume Indicator (HVI) for different explorations of HyperMapper using:

hm-plot-hvi /path/to/configuration_file