Install Request: HOOMD-blue 2.9.6

[heatherkellyucl]

IN:05609617

CPU version at present, but we should probably add a GPU version if we can too.

I tried updating our existing buildscript to these modules but the test suite segfaults:

require gcc-libs/4.9.2
require compilers/gnu/4.9.2
require mpi/openmpi/3.1.1/gnu-4.9.2
require python3/3.9
require cmake/3.21.1

Compute Canada have an EasyConfig for 2.9.3 and 3.6.0 https://github.com/ComputeCanada/easybuild-easyconfigs/tree/computecanada-main/easybuild/easyconfigs/h/HOOMD-blue

The gompi-2020a toolchain used for HOOMD-blue-2.9.3-gompi-2020a.eb is gcc 9.3.0 and openmpi 4.0.3, and they use Python 3.8.2 and SciPy-Stack 2020a (and build it with tbb).

Spack only has 2.1.6 and 2.2.2 from the older location at https://bitbucket.org/glotzer/hoomd-blue.git at present. If we can get a version that works, we should write an Spack recipe for it.

[heatherkellyucl]

Andy Turner created an Ubuntu 20.04 Dockerfile for an ARCHER2 install of this:

requirements.txt:

# glotzer tools
freud-analysis==2.7.0
gsd==2.5.1
garnett==0.7.1
pythia-learn==0.3.0
rowan==1.3.0.post1
coxeter==0.6.1
plato-draw==1.12.0
signac==1.7.0
signac-flow==0.17.0

# commonly used packages
h5py==3.6.0
jupyter==1.0.0; platform_machine != 'ppc64le'
matplotlib==3.5.0; platform_machine != 'ppc64le'
notebook==6.4.6; platform_machine != 'ppc64le'
opencv-python==4.5.4.60
pandas==1.3.4; platform_machine != 'ppc64le'
Pillow==8.4.0
pyqt5==5.15.6; platform_machine != 'ppc64le'
pyyaml==6.0
scikit-learn==1.0.1
scikit-image==0.18.3; platform_machine != 'ppc64le'
scipy==1.7.3
statsmodels==0.13.1

# test requirements
hypothesis==6.29.0
pytest==6.2.5
miniball==1.1.0
sympy==1.9
tables==3.6.1

requirements-source.txt:

Cython==0.29.24
numpy==1.21.4
pkgconfig==1.5.5
pybind11==2.8.1
pythran==0.10.0
scikit-build==0.12.0
scipy==1.7.3

requirements-mpi.txt:

mpi4py==3.1.3