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

[Galtier, Eric Christophe]

Hi Klaus,

Sorry for taking this long to answer your email. I have identified and fix the problems I had on my simulation by following your advice and look more carefully on the EOS.

The weird gamc oscillation I had came from a too low-resolution EOS grid for the target material. Once I increased the density point by an order of magnitude, the oscillations were replaced by smooth evolution of the index as a function of pressure and density and the simulation could proceed. Also, the initial conditions of the material needed to be adapted to the EOS to prevent the negative pressures of cold and solid density from affecting the initialization phase. The ambient temperature conditions were set to the minimum value that was producing a non-negative pressure at the initial solid density, so about 4000K in my case instead of 293K at ambient, but since the plasma evolves up to some few keV temperature, it is not a problem or a limitation in my running conditions. I believe these negative pressures in cold solid matter are not uncommon with this type of plasma EOS and maybe there could be a setting in the configuration file, like a temperature threshold, at which hydro starts and allows for expansion. Alternatively, maybe the minimum temperature at which the pressure is non-negative could be set automatically since the code has access to the complete EOS table.

Despite these fixes, the 'negative sound speed' problem was still preventing completion of the code execution. Always at the same location and independently of the resolution. It happened to also come from the EOS but from the vacuum material this time. To mimic the vacuum, I was using a low density He atmosphere. The low density was not the problem, but rather the opacity constants I was using for it. I had to put them at a value of 1e-5 (absorption, emissivity and transport opacities) for the simulation to converge. I could run as low as 10^-10 g/cc without sign of unhappiness on the results by using these 'vacuum' values.

Now, the simulation runs without problems up to the end of the simulation time and I can shift focus on the analysis of its output.

Thanks a lot for your help!

Eric 

---
Dr. Eric Galtier
Matter in Extreme Conditions
SLAC-LCLS
2575, Sand Hill Road
Menlo Park, 94025, CA, USA
Tel : (650) 926-6227

-----Message d'origine-----
De : Klaus Weide <klaus at flash.uchicago.edu>
Date : mardi 26 mai 2020 à 2:52 PM
À : "Galtier, Eric Christophe" <egaltier at slac.stanford.edu>
Cc : flash-users <flash-users at flash.uchicago.edu>
Objet : Re: [FLASH-USERS] gamc divergence at interface for laser/matter interaction

    On Sat, 23 May 2020, Galtier, Eric Christophe wrote:

    > ... What this plot shows is that the large change in gamc is not 
    > localized to a cell but extends to the entire surface mapped by the 
    > material/vacuum interface. Having this large value of gamc at the 
    > interface might not seem completely weird since it is a function of the 
    > derivative of the pressure vs density and the interface is essentially a 
    > gradient. Does it sound reasonable? 

    I don't think we see strange oscillations (or extreme values) in gamc 
    like this when LaserSlab is run with other materials. So the fact alone 
    that there is a strong local gradient of pressure (or of anything else) 
    does not quite explain it.

    I suspect it is most likely that the problem lies primarily with the 
    tabulated EOS. Bu I am just guessing.

    > It also seems that gamc start to be 
    > negative very 'abruptly' and goes from a somewhat normal value to a 
    > negative value without much intermediate. The laser has its maximum at 
    > 500 ps and the tails of the gaussian pulse shape extends all the way 
    > down to t=0ps. At 350 ps, the laser intensity is about 1/10 of the peak 
    > intensity so there is already some expansion that happened and there 
    > should be no specific discontinuities there. I'll keep trying to link it 
    > to a pressure point in the EOS. To answer your point 2, the simulation 
    > runs ok with this same material and EOS but when I reduce the resolution 
    > going from 2.5 mic per cell to 10 mic. So maybe it is more tight to the 
    > hydro scheme I'm using rather than the material? 

    Of course there are Hydro parameters that you could vary, starting with 
    the CFL factor.

    I also wonder whether you see other quantities going to unphysical values 
    when (or just before) gamc goes crazy. For example, very low or high 
    pressures or temperatures (either Tele or Tion).

    Klaus