search

jacobseiler / rsage

The Reionization using Semi-Analytic Galaxy Evolution model.
MIT License
1 stars 0 forks source link

Explicitly Track Gas Ejected by Supernovae/Quasar #9

Closed jacobseiler closed 6 years ago

jacobseiler commented 6 years ago

Currently the EjectedGas reservoir is ambivalent to how the gas was ejected. However it could be useful to explicitly track which processes ejected gas to determine the relative impact of these two processes.

jacobseiler commented 6 years ago

So there's an issue here regarding how to treat the gas when it's reincorporated back into the galaxy. Whilst we can track through which channel gas was ejected, it's more uncertain about when each channel of gas becomes reincorporated.

My current running idea is to have it so the reincorporated gas is drawn equally from both channels. Then, in the case that one channel does not have sufficient gas I simply just draw the remaining from the other.

jacobseiler commented 6 years ago

There was numerical issues regarding subtracting gas from the two channels. Within model_infall.c, if the halo has lost gas over a time step, the baryons are first subtracted from the EjectedMass reservoir.

However for galaxies with very low masses (the ejected reservoir had ~2e3Msun) the numerical precision would cause problems when I required that EjectedMassSN and EjectedMassQSO reservoirs to be >=0.0. This was solved by putting a numerical tolerance in and saying if the galaxy has low mass, just fudge the tracking.

jacobseiler commented 6 years ago

Issue marked as closed for now. The numerical issues couldn't be properly resolved for Kali so I've removed the XASSERT calls and the fescPrescription options that make use of the ejected channels.