Connecting HPC@LR

For my more significant computational tasks, I use the Meso Center HPC@LR leaded by the University of Montpellier and hosted at CINES. I provide here practical informations relative to my activity on this cluster HPC (High Performance Computing). HPC@LR provides a new cluster called Muse (formerly τ-au). The cluster can be touched with :

# Or:
# with the following alias defined in my $HOME/.profile file:
alias sshmuse='ssh'

As I use a ssh key, providing a keyword is not my concern. If you got your account on use but don’t have a ssh key, you can do on your linux client:

ssh-keygen -t rsa # (If you don't have any ssh key)

You may use a passphrase. But if you set one, you’ll have to write it. Then in all cases:

ssh-copy-id -i ~/.ssh/

Now,you can connect directly with the same syntax as mine above.
I wrote a tutorial about how to play with Muse. It focuses on my scientific activity, but you may easily adapt it to your own.

HPC@LR manages users and projects.

Each project corresponds to a workspace with an amount of hours, some disk space and possibly several users. I have several projects hosted at HPC@LR which are:


Since years, with colleagues and former students, I use to analyse coastal / beach hydrodynamics with the SIROCCO environment eventually coupled with WW3. This coupling has been first used in the Gulf of Lions (so the name GOL for the project). But SIROCCO is now used for wider applications in coastal dynamics & paleo-oceanography. In particular, with my friends Mathieu Schuster, Jeff Ghienne and Alexis Nutz, we developed the concept of Wind-driven Water Body (WWB), which are (usually) closed water bodies of intermediate to shallow water depth. In this group, I am in charge of the quantification of (paleo-)hydrodynamics in WWBs and of the reconstruction by numerical means of the paleo-forcings at the origin of the wind-driven geomorphic features that occur within the sedimentary deposits. The systematic modelling of several tens of WWBs around the world require robust computation facilities like HPC@LR.


I use several numerical tools dedicated to the modelling of shoreline and nearshore morphodynamics. One type of models is based on non-linear diffusion equations. Another category relies on optimal design / optimization. At now, their usage in real case studies or engineering do not require large numerical resources. But I project the usage of such tools for more realistic applications in the future. So I tentatively deploy the tools in an HPC environment.


This project is special insofar as it is not devoted to any specific place around the world nor any specific tool. It is dedicated to the manipulation of a SINGLE set of global forcings for all the other coastal projects at HPC@LR: we store, we extract and we prepare the input files for hydro-morphodynamic simulations performed in other HPC projects. The numerical effort corresponds mainly to the pre-processing (and eventually post-processing) of meteo-marine and bathymetric data. Although it is pre-processing, it is performed on HPC nodes.