Swift-EAGLE IMF variations

[christopherlovell]

In progress work towards supporting arbitrary stellar initial mass function parametrisations in the Swift-EAGLE model.

In order to support an arbitrary IMF the code must be updated so that the feedback, enrichment and star formation models use the same self consistent IMF assumptions.

Feedback and enrichment changes:

• Implements custom IMFs in feedback/EAGLE/imf.h. These include Chabrier+03, Salpeter+55, Kroupa+05, and a custom broken power law. Defaults to Chabrier+03 to ensure backwards compatibility.
• Chosen IMF updated throughout the EAGLE feedback scheme
• Example parameter file provided in examples/EAGLE_low_z/EAGLE_6/imf_example_6.yml with the new (optional) IMF configuration options.

Star formation changes:

• A new python script has been added to tools/, imf_ks_normalisation.py, which uses FSPS to find the new normalisation of the Kennicut-Schmidt relation used in the EAGLE star formation model.

Questions / TODOs:

• How should we update EAGLEStarFormation:KS_normalisation when users choose a new IMF? Should we tabulate some common values using the FSPS script, as well as a table for different broken power law choices that can be interpolated? Or should this be left to the user to update manually?
• Are the changes to the IMF correctly propagated to the enrichment schemes?

[MatthieuSchaller]

Hi Chris,

Thanks for contributing this. We should probably do this via the gitlab server since the github here is just a mirror updated every night.

From a quick look at the changes, it seems everything is sensible.

I would recommend running the example examples/SubgridTests/StellarEvolution. It consists of a single star in a box of gas running for 15Gyr. It then plots the mass, Z, etc. produced. That should let us see whether the differences expected from the different IMFs are indeed working. I don't know if we have a quantitative answer for all these for the different IMFs but I think we know qualitatively what to expect.

[MatthieuSchaller]

Are you still looking at using these changes?

[christopherlovell]

At some point yes. Should I move to gitlab now?

[MatthieuSchaller]

That would be easier, yes.

[MatthieuSchaller]

@christopherlovell could you move this to the gitlab if you want it integrated?

[christopherlovell]

I tried to get access to the Gitlab a while ago with Alastair's help, but no success. I'll look into it again

[MatthieuSchaller]

You are a registered developer of the project there already.