[Lukas832]

Good evening,
I am using pyamg, with SU2, to obtain adaptive meshes. With the solvers in equilibrium there is no problem whatsoever and the desired result is achieved. However, as soon as I switch to NEMO (with any fluid model, either the su2 built-in or mutation++), then the pyamg output file amg.out complains that:

## ERROR: INTERPMET: INVALID METRIC: NULL OR NEGATIVE EIGEN-VAL

## FATAL ERROR: FEFLOA: INVALID METRIC FILE 

And so no adapted mesh is produced. Curiously, if I use the same mesh and edit the config file to bring me back to an equilibrium case (simple euler solver), everything works.

In detail, I compiled the brach feat_adap_dev. I have seen that AMR with pyamg and NEMO is used in a lot of various research, so I think it is possible to couple pyamg with su2NEMO.

I attach the .cfg and .su2 files of the case: for simplicity's sake the .cfg is essentially that of the NEMO testcase for the inviscid wedge (just add the adap parameters and disable MUSCL) and the mesh is a simple mesh with triangles. The mesh is very coarse to stay in the size limits of the forum. I tried to modify the mesh both in number of elements and by putting all the triangles of almost identical size but it did not solve the problem. I also tried to use the standard transonic NACA0012 mesh and config (with the addition of NEMO) but I get the same type of error.

Has anyone ever encountered this kind of problem? Any idea on the approach to follow?
Thanks

[Lukas832]

Update: could it be that the branch feature_adap_dev is only suitable for equilbrium Euler simulations? I tried to go back to the feature_adap_tne2: both non equilibrium and mesh adaptation seems to work, however TNE2 has grown and changed hugely (becoming NEMO) and it looks that the mesh adaptation implementation changed a bit between feature_adap_tne2 and feature_adap_dev (as 2 years passed by)

I think that there must be some newer branch to perform that, but I still have no clues

[giovanni.medici]

If I recall correcly feature_adapt branch was a 7.0.3 branch, well before NEMO was introduced into the master

[Lukas832]

Hi, giovanni.medici thank you for your answer.
You're right the feature_adapt branch is with the version 7.0.3, but the feature_adap_dev is with the 7.3.1 that comes with the NEMO solvers.