Crash course tutorial on SED modeling, specifically with Prospector.
Prospector (https://github.com/bd-j/prospector) is an SED modeling code that wraps FSPS stellar modeling and Dynesty dynamical nested Bayesian sampling. It enables flexible modeling of galaxy SEDs with several model choices and data handling functions. In this tutorial, I walk through the fundamentals of both SED modeling and Bayesian inference, with two example fits to show hands-on how to use Prospector and understand the output of Prospector.
A few prereqs (also mentioned in the tutorial notebook preamble):
-
sedpy : this manages the details of our observations and allows us to interface with the properties of the photometric filters used. You can find sedpy here: https://github.com/bd-j/sedpy
-
dynesty : this is the backbone of propsector that handles the actual fitting methods. It is a form of Bayesian modeling similar to MCMC codes like 'emcee' but a bit more sophisticated. You can find dynesty here: https://github.com/joshspeagle/dynesty
-
fsps : this handles the stellar modeling for prospector. by itself, it's a super useful tool for generating stellar spectra. The core of fsps is a Fortran code but these days, the python bindings for fsps now come with its own fsps source code, meaning we no longer have to compile the Fortran code first then install the python wrapper. All we need is to clone the fortran fsps
export SPS_HOME="/path/where/you/want/to/download/fsps" git clone https://github.com/cconroy20/fsps.git $SPS_HOME
And then pip install python-fsps
python -m pip install fsps
- Prospector : prospector is the tool that combines the above packages to model the SEDs of galaxies. You can find it here: https://github.com/bd-j/prospector It also has a pretty decent demo on how to use prospector: https://github.com/bd-j/prospector/blob/main/demo/InteractiveDemo.ipynb For visualization purposes, we'll also want to install 2 packages: corner, which helps us plot corner plots (https://github.com/dfm/corner.py) and arviz, which allows us to do 'advanced' things with corner (conda install arviz)
What's in this repo:
- Tutorial Notebook, allows you to get started with Prospector; explains the different model components, how to process Powderday SEDs, and basic analysis steps to understand results.
- Prospector runtime scripts that take the contents from the tutorial notebook and package them as a python script; two SFH models.
- Script for processing prospector output, calculates posterior distributions for stellar and dust mass, SFR, metallicity, and the model SEDs.
- In the "prospector_nonpara_SHFs" directory, scripts for the two other nonparametric models as well as their respective output processing scripts. These files are not tailor made for this tutorial but are useful jumping off points from which you can copy/paste the model setup for your own use (one useful thing is that these scripts were used to model z=7 simba galaxies so you can see how to set up z>0 runs since the tailor made scripts just use a z=0 galaxy).