Compatibility of EMBED + AXI + ADAPT

[Edoardo Cipriano]

Dear all,

Inspired by the (relatively) recent patch that extended EMBED's capabilities including AXI metrics, I tried to setup a simulation of single phase flows with EMBED and AXI.

The simulation works fine only with multigrid (or using tree but without calling adapt_wavelet()).

Since I haven't found any example of AXI + EMBED + adapt_wavelet(), my question is: is the AXI and EMBED feature supposed to work with AMR or not?

As previously reported in other threads, the simulation crashes with the error:

Assertion `fm.x[]' failed.

In the timestep.h file. Therefore, it looks like the refinement/coarsening functions are setting fm.x to zero where it is supposed to be non-null. Please find attached the simulation file, which is very simple. You can try to comment (or uncomment) the first line to verify the aforementioned behavior.

Is there anybody who encountered this problem and managed to solve it?

Best,

Edoardo

[Edoardo Cipriano]

Here is the correct simulation file.

[jmlopez...@gmail.com]

Hi Cipriano,

EMBED + AXI + ADAPT works, but implementing a C file right now is not straightforward. A proper adaptation requires adjusting cs and fs first, followed by cm and fm, before proceeding with the remaining variables. For now, this process must be done manually, which is quite inconvenient. If I find some time, I will try to create a patch to simplify the task.

   

Cheers

[Edoardo Cipriano]

Hi Lopez,

Thank you very much for your time and help.

I will play around with the version you attached.

Best,

Edoardo