HPC module loading

[edg-13]

Hello,
I'm trying to run a Firedrake simulation using the Imperial HPC facility but I am unsure how to load the required modules in the job script, or create a personal conda environment with Firedrake installed (since I am not sure whether I have permission to manually install Firedrake).

Any help is appreciated.

[connorjward]

Please check out our wiki entry on installing to HPC. If you still have problems @JDBetteridge should be able to advise.

Also note that Firedrake is not compatible with Anaconda at present.

[JDBetteridge]

You need to convert the docker images following this guide here (steps 1 & 2): https://github.com/firedrakeproject/firedrake/wiki/singularity

[edg-13]

Thank you, would the example job script be similar, e.g same global variables?

On another note, I have noticed firedrake is a loadable module on the HPC system. Using just module load firedrake in the job script gives me this error.

If there's an easy fix for this, then I think this may have a cleaner job script.
But if using the singularity image is less fiddly, I'm happy to go ahead with that.

[JDBetteridge]

That screenshot looks like a Firedrake install from 2016, it almost certainly won't work.

The environment variables (=? global variables) all need to change to the ones on the HPC system, without access I can't tell you what these are, you will have to check the HPC documentation.

[edg-13]

I think in that case the spack method may be the easiest. I am running into this error when running spack install

What am I missing? Here are my .spack/packages.yaml and /spack.yaml files as txt files for github.
spack.txt
packages.txt

[JDBetteridge]

The Spack recipe is in serious need of work and has not been updated for a long time. Fixing the Spack installation is on my very long to do list