Release new version of hybridLFPy

[torbjone]

Aspect	Detail
Summary	hybridLFPy is the software used for co-simulation between NEST and LFPy (though NEURON), but official version relies on outdated versions of NEST, NEURON and LFPy, and is only available for python 2.7.
Task Area	Version upgrade
Assignee	Torbjørn V Ness and Espen Hagen
Information	torbjorn.ness@nmbu.no
Prerequisites
Dependencies

Summary

The official release of hybridLFPy (https://github.com/INM-6/hybridLFPy) is badly outdated, and needs to be updated to python 3. Calculation of EEG should also be made easily available in official version. We should migrate our simulation work from local supercomputers to EBRAINs resources if possible?

Tasks

• [x] Upgrade and test software with up-to-date dependencies (NEST, NEURON, LFPy, python, etc)
• [ ] Make software run through EBRAINs HPC resources

[w-klijn]

"We should migrate our simulation work from local supercomputers to EBRAINs resources if possible?" That would be a good idea. I would suggest packaging your work in a container solution. Singularity would be the go to solution for HPC workloads. This is available in JSC and CSCS (the current FENIX sites) and should also be deployed on the other locations.

A first step would be to apply for resources : fenix-ri.eu/access as this might take some time, it might be an idea to send it out early. If you need help with the forms let me know. I can have someone in the Simlab help you fill it out.

[torbjone]

Thanks, I will look into this!

[espenhgn]

Hi @torbjone and @w-klijn. Can you please add me to this team/organization and assign me to this issue? Can you also edit my personal email above? Got enough spam already....

Anyway, I've ported the hybridLFPy module and examples to Python3, LFPy v2.* and NEST v3 (current master branch) at https://github.com/INM-6/hybridLFPy/tree/nest3. I plan to merge and create the next hybridLFPy release when NEST 3 is finally released.

I aim to create an EBRAIN collabs demonstrating use of LFPy and hybridLFPy, but what you're saying is that I still need to apply for resources via fenix even if I already have regular access to Jureca?

[torbjone]

Hi Espen, I removed your personal email, and assigned you to this issue. I think we will apply for computing resources through fenix for our group at NMBU, but I don't think you need to think about this now.

[espenhgn]

Tx! Looks like we'll have to define some project for the application form at https://wiki.ebrains.eu/bin/view/Collabs/fenix-icei/Drive, so I'm not sure if it makes sense to apply as a group. Perhaps we be better discuss this offline w. Gaute and/or HEP. The Collab is useless without compute access so we should get that in place asap.

[espenhgn]

@torbjone; this issue can be closed. HybridLFPy is now released in version 0.2 across the usual channels (https://github.com/INM-6/hybridLFPy; https://pypi.org/project/hybridLFPy/; https://doi.org/10.5281/zenodo.4976625).

The EBRAINS showcase has been up for a while (https://wiki.ebrains.eu/bin/view/Collabs/hybridlfpy-showcase/; https://github.com/LFPy/hybridLFPy_EBRAINS.git), but probably now needs some updates. There are now also a developer Docker image that may be used, so I need to make sure things works on both (not that I expect problems as most of the work is done using a custom image on the HPC backend).