Skip to content

INSTALL.md

Latest commit

 

History

History
63 lines (39 loc) · 2.03 KB

INSTALL.md

File metadata and controls

63 lines (39 loc) · 2.03 KB
Raw

Installation notes

There are three ways to obtain lunapi:

pip

We are aiming to support binary wheels for Linux and macOS (arm64 and amd64). If the following is able to find a matching binary wheel for your platform and Python version (3.8-3.12 currently), you can use that.

pip install lunapi

It is possible the support for Windows will be added downstream, but that is not likely to happen soon: i.e. Windows users should look to the Docker image (below).

Docker

See notes at https://github.com/remnrem/luna-api-notebooks for installing the Docker version of lunapi (based on a JupyterLab image). This is currently the only option for Windows users.

Manual build

We cannot currently provide support for manual builds - but if you wish to follow these rough steps and have a working knowledge of compiling software, this should work on most Linux/macOS platforms with typical Python installations, etc.

  • follow the instructions here for compiling Luna locally

  • this will involve obtaining the following two dependencies: FFTW3 and LightGBM libraries

  • download this repository; edit the CMakeLists.txt file to point to the location of Luna, FFTW and LightGBM headers and libraries; you may also need to make sure libomp/libgomp is available on your system

  • run pip install .

  • test in Python with import lunapi as lp

Resources

The Docker image contains all resources (e.g. tutorial data, POPS models, PREDICT models). If you are building manually, or if you use pip above, then you can obtain these resources from these locations;

  • notebook/tutorials : https://github.com/remnrem/luna-api-notebooks

  • POPS models : git clone https://gitlab.partners.org/zzz-public/nsrr.git

  • PREDICT models : git clone https://github.com/remnrem/moonlight/tree/main/models

  • NSRR tutorial EDFs : wget http://zzz.bwh.harvard.edu/dist/luna/tutorial.zip