Installation

Installing Python

Anaconda is the recommended method to install Python for scientific applications. It is supported on Linux, Windows and Mac OS X. Download Anaconda here. Note that NEXTorch runs on Python 3.7 and above.

Creating a new conda environment (optional)

It may not be possible for one Python installation to meet the requirements of every application. To avoid such conflicts, we recommend working inside a virtual environment dedicated to all PyTorch applications. If you have not done so, use conda to create a new environment:

conda create -n torch

Activate the new environment:

conda activate torch

Deactivate it after each use:

conda deactivate

Installing NEXTorch using pip

Using pip is the most straightforward way to install NEXTorch.

1. Activate the virtual environment for PyTorch.

2. Install NEXTorch by typing the following in the command prompt (the fresh installation takes ~1-2 minutes):

pip install nextorch

Installing NEXTorch from source

If you would prefer to install from source or you are interested in development, follow the instructions below.

pip install git+https://github.com/VlachosGroup/nextorch.git

Upgrading NEXTorch using pip

To upgrade to a newer release, use the –upgrade flag:

pip install --upgrade nextorch

Running unit tests

NEXTorch has a suite of unit tests built on the pytest framework. One should run the tests to ensure all code functions as expected. Run the following commands in a Python terminal (usually takes less than a minute):

pytest --pyargs nextorch

The expected output is shown below. The number of tests will not necessarily be the same.

PACKAGE_PATH\nextorch\test\test_1d_function.py ..                                     [ 15%]
PACKAGE_PATH\nextorch\test\test_EHVI.py ..                                            [ 30%]
PACKAGE_PATH\nextorch\test\test_io.py ....                                            [ 61%]
PACKAGE_PATH\nextorch\test\test_parameter.py ...                                      [ 84%]
PACKAGE_PATH\nextorch\test\test_plotting.py ..                                        [100%]

================================== 13 passed in 44.00s ======================================