search

evanoconnor / NuLib

open-source neutrino interaction library
24 stars 16 forks source link

Unreasonable value for high Ye at low temperature and low density #38

Open suniverse opened 4 years ago

suniverse commented 4 years ago

Hi, I have made a NuLib data from the SFHoEOS file Hempel_SFHoEOS_rho222_temp180_ye60_version_1.1_20120817.h5 downloaded from the stellarcollapse.org. I produced the neutrino data with four energy bins and 3 species. The output h5 file has some unreasonable values at high Ye, low temperature and low density. I attach two plots of emissivities and absorption opacities at Ye=0.486 and temperature 0.01 MeV. You can see the low density values go wild. Only for the first species, the other two has almost zero values. abs emis

evanoconnor commented 4 years ago

Thanks for this. I already know the issue is because the temperature is too low, I never go below 0.05 for this reason.

The question is what do we need to do to fix it. I think the transition happening at 1e11g/cm^3 signals that it is something related to the final state nucleon blocking (see https://github.com/evanoconnor/NuLib/blob/master/src/absorption_crosssections.F90#L366 and similar lines)

Below 1e11 g/cm^3 we just take the nucleon densities from the EOS table. But these must be inconsistent with the chemical potentials used in the cross section calculation at these low temperatures and therefore there are some exponentials left unchecked.

suniverse commented 4 years ago

Hi Evan, is there any way I can fix the values below 1e11g/cc? I am studying the accretion disc after the binary merger, and the highest density is just a few times 1e11g/cc. I don't care if the data is wrong at low temperature, since the emissivities is already low there. But I need a smooth data. I have tried to set the values to 0 artificially below some threshold, but it seems the code is not stable. I have also tried with leakage, and the code is stable there.

evanoconnor commented 4 years ago

I would suggest choosing a Tmin that is not so low for the NuLib table itself. There shouldn't be any neutrino physics happening at T=0.01, that is really cold. I typically choose Tmin = 0.05, but even Tmin = 0.1 should be ok.

If that is not acceptable, you need to explore the nucleon blocking for low T to figure out if/how it can be extended. In theory, one should just get a blocking factor close to 1 (i.e. no blocking) for low densities. This was orignally in NuLib, but it gives some issues at low densities too (see. https://github.com/evanoconnor/NuLib/commit/679683b22377f021f0310728d1f5f074d34f29b5)