LLNL/pyranda

Possibility of running 'PARCOP' only.

Closed this issue · 6 comments

bgin commented

Hello

Is it possible to run parcop solver? only , i.e. without the python wrapping layer?
I'm mainly interested in such a usage as it will nicely fit my project requirements.

Best regards
Bernard

bgin commented

Seems nobody really cares to answer.

  • 😕1
flow-phys commented

@bgin , yes. This is possible and there is an example in the parcop directory and a makefile. The miniApp.f90 is the main program which might serve as an example of who to use library.

bgin commented

Hello @flow-phys
Thank you for your reply (sorry for not waiting enough time).

I did have a look at "miniApp.f90' main program, but I was slightly confused by not being able
to identify the solver entry point.
Now I'm seeing the calls to "div" subroutine:

Are these a calls to solver itself?
Here I'm seeing the calls to vector divergence subroutine

Put it simply -- in order to call the 'solver process' is there a need to call those "div" subroutines,
or those calls are made do vector or tensor divergence subroutines only?

Thank you
Best regards
Bernard Gingold

bgin commented

It seems that pasted by me links somehow disappeared.
I'm providing the links here:

https://github.com/LLNL/pyranda/blob/master/pyranda/parcop/miniApp.f90#L204 <-- is this a direct call to solver?
Secondly -- how I can pass a boundary condition data?

Regards
Bernard Gingold

flow-phys commented

To see if I understand your question, are you wanting a hydro-dynamics solver in pure fortran? The miniApp is not quite a full fluids solver but rather exercises the operators that a solver would use. Its not turn key and you would need to develop your own full/expanded version of miniApp.f90.

Does that make sense?

Calls to "div", "grad", "filter" are the solver components that parcop supplies but you'd need to write your own code to use these to make a hydro-solver. One thing that is lacking in the miniApp.f90 example is the time integration. Also, you'd need to define your own boundary conditions, etc natively in the fortran arrays.

Hopefully this helps clear things up.

-Britton

bgin commented

Hello @flow-phys

are you wanting a hydro-dynamics solver in pure fortran? The miniApp is not quite a full fluids solver but rather exercises the operators that a solver would use. Its not turn key and you would need to develop your own full/expanded version of miniApp.f90.

Yes this is what I wanted to know!
Thank you for clearly explaining those slightly confusing (to me) things.

Best regards

Bernard Gingold