MCMC Fitting¶

Warning

These functions are under construction.

Setup¶

Copy the config.yaml and run.py from the main directory to an external folder.
Edit your configuration script config.yaml, which should look something like below.
In your new directory run python run.py in terminal.

Configuration¶

# Instrument specifications
data:
  instrument: "APOGEE"
  data_path: "default" # defaults to $APOGEE_DATA path (see setup documentation), unless otherwise specified
  ID: "2M01195227+8409327"
  orders: [[15200,15800],[15860,16425],[16475,16935]] # wave ranges, and orders
  dtype: "ap1d"
  visit: 1
  sigma_clip: [.3,.05]
  pixel_buffer: [0,2]

# Make sure this config.yaml and run.py files are placed in your input directory
# I recommend copying config.yaml and run.py to a path external to apogee_tools
workdir: 
  input: "/home/jess/Desktop/Research/FAST/fit_models"
  output: "/home/jess/Desktop/Research/FAST/fit_models/output"
  
out:
  mcmc_sampler: False
  corner: False
  walkers: False
  print_report: True

# Specify which parameters will be sampled by MCMC
# otherwise parameters will be fixed at 'init' values
model:
  grid_name: "PHOENIX" #directory: phoenix/apogee/order
  theta: ['teff', 'logg', 'fe_h', 'rv', 'vsini', 'alpha']

fix_param: # specify fixed parameters (not sampled by MCMC)
  airmass: 1.0  # airmass of telluric model, either 1.0 or 1.5
  cont_deg: 5   # continuum polynomial degree
  interp_method: "splat" # or "cannon"
  resample_method: "fast" # or "splat"

# MCMC tuning
mcmc:
  nwalkers: 12
  nsteps: 3
  
# Initial parameters for MCMC
init:
  teff: 3500
  logg: 4.50
  fe_h: 0.0
  rv: -4.77
  vsini: 5.79
  alpha: 1.0

# Step parameters for MCMC
step:
  teff: 1
  logg: .01
  fe_h: .01
  rv: .1
  vsini: .1
  alpha: .01

# Prior ranges for MCMC (for flat prior)
prior:
  teff: [2500, 5500]
  logg: [0.0, 5.5]
  fe_h: [-1.0, 1.0]
  rv: [-200, 200]
  vsini: [0, 200]
  alpha: [0, 5]

Pre-MCMC Testing¶

To test to make sure all of the modeling modules are working, run the following command in terminal:

python run.py make_model

which should return something like:

[25.732014894485474s] MCMC initialization step complete.

##################################################
Making model: teff=3500 logg=4.5 fe_h=0.0 rv=-4.77 vsini=5.79 alpha=1.0

[0.07615256309509277s] Interpolated model
[0.0025053024291992188s] Shifted radial velocity
[0.0032796859741210938s] Applied vsini broadening
[0.05470013618469238s] Convolved telluric model
[0.08379793167114258s] Applied LSF broadening

To test by eye, that your initial MCMC parameters are some close to the data:

python run.py test_fit

Running the MCMC¶

Run the MCMC:

python run.py mcmc

Plot the outputs:

python run.py walkers
python run.py corner