Marketing JuliaFEM to find more contributors

[TeroFrondelius]

Hi all, This is a discussion opener to find more contributors. The whole idea is to make an open source FEM around julia language thus the name JuliaFEM. We have also purchased the domain www.juliafem.org for next ten years. Main idea is to collect all FEM contributors under the same organisation. We have also planned to publish all packages documentations under www.juliafem.org.

Our project is in planning phase, thus this is a perfect time to join the project, if you are interested about software architecture and the big fundamental decisions. Our field is solid mechanics, but the idea is to make multi physical "commercial" grade FEM solvers.

/CC @JuliaFEM/owners, @mauro3, @billmclean, @PetrKryslUCSD, @KristofferC.

[ahojukka5]

Some information about project current status:

Basically we are already able to solve 3d nonlinear elasticity problems using single core. Naturally this can easily be extended for example diffusion problems.

We are still thinking about how to access and store data model. Probably we will end up using hdf and xml because they are well supported and easy to read using other programming languages too. For visualisation we are thinking to use Xdmf and Paraview.

One of the biggest currently unresolved issue is how to efficiently parallelize the solution process. Any contributions (material, code, good ideas, ...) are very welcome. Of course we have a lot of other currently unresolved issues but this parallel computing is thing which needs to be solved asap in order to proceed to another subjects.

Like @TeroFrondelius mentioned, while our field is mainly solid mechanics the plan is to develop multi physical framework. We haven't mentioned anything about for example CFD in our issues but of course it fits naturally to JuliaFEM. I find fluid-structure-interaction problems to be very interesting subject.

One of our goal is to make program structure simple and accessible in a way that this package could be used for example in educational purposes in universities. To achieve this goal we are putting effort for comprehensive documentation and tutorials so potential contributors don't have to read thousands lines of code to understand how it works. Hopefully in some day we have a fast framework where researchers can easily test their ideas and study different phenomenons using notebooks. Running FEM models in notebooks using efficient high level scripting language for rapid prototyping, is there anything cooler?

[mauro3]

Hi, thanks for reaching out! I was snooping around bit on the internet but couldn't find any information on who you are. So, who are you?

I am interested in (kinda) fluid dynamics problems. see my web-site for details. I don't have any FEM code in Julia yet. Maybe of interest, I'm working on a large-ish (serial) scale implicit time stepper https://github.com/JuliaLang/ODE.jl/pull/72. I also have a mesh library https://bitbucket.org/maurow/lmesh.jl. I'll have a look at your codebase over the next wee while.

Also, there are some efforts in Julia to unify geometry and visualisation, maybe of interest to you: https://github.com/JuliaGeometry and https://github.com/JuliaGL/

[KristofferC]

Good initiative!

I'm a Ph.D student at Chalmers, Sweden doing constitutive modeling in solid mechanics (more specifically modeling crystal viscoplasticity).

For my project I have some FEM code which is currently mostly for my research: https://github.com/KristofferC/FEM.jl. Here is an example input file to generate something that looks like this in Paraview.

I plan to get it in a better state when I get time.

I looked though your code and it seems you will have some performance bottle necks (passing function arguments for example) However, right now, you are probably more focused on getting stuff up and running than performance.

[ahojukka5]

I'm mechanical engineering student at University of Oulu. Currently I'm working with my thesis, subject is mortar methods in contact mechanics. Probably I'm implementing something related to that in some schedule. We haven't done any serious performance measurements yet. But yes, I'm aware that code is quite slow at the moment to what we are expecting in future.

[TeroFrondelius]

Here is a link to my LinkedIn profile: https://fi.linkedin.com/pub/tero-frondelius/71/a06/63b

Currently this JuliaFEM is an hobby for me.

[ovainola]

I'm also a mechanical engineering student at University of Oulu, working with my thesis (micromechanics & fatigue).

[TeroFrondelius]

Related to #11

[marcosmarxm]

I'm a electrical engineering student from Brazil. Maybe I could help with data structure initially. I have developed so far a FEM software in Julia using gmsh to do mesh and pos.

[TeroFrondelius]

Maybe one can use FEM for fusion plasma physics applications. @tlycken asked to ping him any related topics here: https://groups.google.com/forum/m/#!topic/julia-users/4b997lOKjc4

@tlycken do you think we have a potential? Would you be interested in trying to solve your equations with our concept? See the notebooks, especially the one that uses ForwardDiff.

[tomasaschan]

Unfortunately, I'm not working actively with the plasma physics stuff anymore, so don't count on me as an active member of the JuliaFEM community (just yet, anyway...). I do spend my fair share of spare time in Julia though (mostly working on Interpolations.jl and some other related stuff; I've also spent some time on ODE.jl in the past) so if there's anything specific that you need help with, and that piques my interest, I might join the party temporarily :)