[FLASH-USERS] gamc divergence at interface for laser/matter interaction

[Klaus Weide]

On Fri, 15 May 2020, Galtier, Eric Christophe wrote:

> I'm running a case closely related to the laser slab example of the 
> manual but modified to the material and laser irradiation conditions I'm 
> interested in. I use ionmix EOS/opacity tables for the target 
> material, and an ideal gas gamma law for the vacuum that is in front of 
> the target. I try to set the density of the chamber material to a small 
> value to mimic vacuum (say 1e-8 cm^-3 while the target is solid density 
> copper). The code runs properly, until it computes a negative gamc which 
> force exit the run. When I plot the gamc value in the grid, I can see 
> the value oscillating with time and increasing in amplitude along the 
> entire interface between the target and the 'vacuum' material reaching 
> up to 11 at the interface inside the target. Because of the oscillation 
> behavior, it actually reaches almost zero at the interface and on the 
> vacuum side, where the laser heat the target. And at some point in time 
> (not even at the maximum of the laser peak intensity) this value becomes 
> negative and force exit the code.

Hi Eric,

1. I am not familiar with this behavior.

2. It would be interesting to know whether something this also happens for 
simpler materials, maybe aluminum.

3. It is possible that the gamc behavior comes from the metal material, 
rather than the "vacuum" EOS, even if the location appears to be outside 
of the solid, because of mixed-material cells.

4. It would be interesting to know where in the code the exit is forced;
in particular, whether this is perhaps an effect of reconstruction within 
the Hydro unit rather than of the EOS itself.

5. I just sent a message to flash-users yesterday on a similar topic, 
where I referred to this code:

           eosData(gamc+i) = eosData(tempToUse+i)/eosData(dens+i) * &
                             eosData(dpt+i)**2 /  eosData(det+i) &
                           + eosData(dens+i) * eosData(dpd+i)

I wonder whether you can trace back the extreme gamc values you are 
getting to this calculation, and eventually to a specific location in your 
Eos table.


Klaus