Getting started with MLGLUE

On this page you will find all the information to get started with MLGLUE. It is assumed that you are somewhat familiar with programming in Python and that you have a working Python distribution installed.

Installing MLGLUE

The easiest way to install MLGLUE is to use the Python Package Index (PyPI). To get the latest version of MLGLUE, open the Anaconda Prompt, a Windows Command Prompt (also called command window) or a Mac/Linux terminal and type:

pip install mlglue

Updating MLGLUE

If you have already installed MLGLUE, it is possible to update MLGLUE easily. To update, open a Anaconda/Windows command prompt or a Mac terminal and type:

pip install mlglue --upgrade

Dependencies

MLGLUE depends on a number of Python packages. The necessary packages are all automatically installed when using pip. The following packages are necessary for a minimal function installation of MLGLUE:

numpy
Ray

Strictly speaking, Ray is not necessary but installing it is very strongly encouraged (which is why it is listed as a necessary package). Ray enables the parallel execution of models, often resulting in substantial computational savings. And due to Monte Carlo-type algorithms (and therefore MLGLUE) being embarassingly parallel, using Ray almost always results in substantial speed-up. However, there is Python-native support included, such that Ray is installed but optionally not used.