DEV: Halotools

[Lucia-Fonseca]

Description

Halotools is a specialized python package for building and testing models of the galaxy-halo connection, and analyzing catalogs of dark matter halos.

The core functionality of the package includes:

• Fast generation of synthetic galaxy populations using HODs, abundance matching, and related methods

• Efficient algorithms for calculating galaxy clustering, lensing, z-space distortions, and other astronomical statistics

• A modular, object-oriented framework for designing your own galaxy evolution model

• End-to-end support for reducing halo catalogs and caching them as fast-loading hdf5 files

References

• https://halotools.readthedocs.io/en/latest/

Question

Do we feel we need this?

[bnord]

@Lucia-Fonseca would this perform simulation functions as the Vale+Ostriker implementation?

[gfavole]

@Lucia-Fonseca besides basic SHAM/HOD should we explore "decorated" models in which we match also secondary properties?

[bnord]

@gfavole what are some example secondary properties?

[Lucia-Fonseca]

That's a good question. It depends on whether that's within the scope of v0.2

[gfavole]

@bnord usually SHAM is performed using luminosity or stellar mass as main proxy for galaxies and maximum circular velocity for haloes. However, besides these primary properties, the best would be to match galaxies with haloes also in secondary props such as colours, age, halo assembly bias, etc.. An example of decorated SHAM is https://arxiv.org/pdf/1310.6747.pdf

[bnord]

Are these properties likely to be formatted in a table/list just like the primary properties? Or, put another away, do you see the data format requiring something to change for the secondary properties?

[gfavole]

yes, the secondary properties are formatted in a table exactly like the primary ones

[Lucia-Fonseca]

Then, I do not see why these secondary properties should not be included

[bnord]

Yeah, if they're easy to generate now and then they are modular so that they can be changed later, I think we can go for it?

[bnord]

@Lucia-Fonseca Is it correct that halotools does NOT generate initial halo information, like mass distributions?

[gfavole]

I think yes, it doesn't generate those primary properties

[Lucia-Fonseca]

@Lucia-Fonseca Is it correct that halotools does NOT generate initial halo information, like mass distributions?

Not an expert on Halotools. I think @gfavole knows more than me

[rrjbca]

Although it's good to think about secondary properties and plan ahead, I think we should prioritise getting a simple pipeline working for v0.2

[rrjbca]

Discussing with @aamara we should use halotools where possible as this is a requirement for astropy affiliation. It has HodModelFactory and SubhaloModelFactory base classes that we can use to build our own HOD or SHAM style galaxy-halo connection models within their framework.