Steps to MegaBEAST paper

[karllark]

As of today, the megabeast has successfully run!

Specifically the megabeast correctly determined the ensemble average A(V) for a simulation of 100 stars with A(V) values in a narrow range (e.g., 1.0 to 1.05) with a starting value of A(V)=3. The BEAST fitting was done with a physics model that had a flat prior from A(V) = 0 to 5. The simulation was based on a Scylla LMC field with 7 bands. In addition, testing was able to simultaneously recover A(V) and R(V) if only very bright stars (F475W mag < 20). R(V) was not recovered when using 100 stars with the full range of fluxes.

The next steps are:

1. Move the simulation code from the BEAST to the MegaBEAST to take advantage of being able to change the prior weights without regenerating the physics grid. See Move simulations to MegaBEAST #66
2. Update the MegaBEAST fitting to start with a chisqr minimizer (optimizer) and then got to the MCMC fitting to get the uncertainties. As part of this, save the MCMC chains.
3. Setup a set of simulations that vary all parameters. Start with testing recovery with only a single BEAST parameter varying (e.g., A(V), R(V), IMF, etc.). What parameters are recovered well? How many stars are needed? How many bands? Will not do an exhaustive search, but at least a number of examples.
4. Setup simulations that vary multiple parameters at once. How many is TBD.
5. Identify one Scylla/METAL field that has significant dust column variations. Run new fits to use for testing the MegaBEAST recovery of dust parameters (mainly A(V)).
6. Draft an outline of the MegaBEAST paper (similar to the BEAST 2016 paper).