nmiller95 / teb

Temperatures of Eclipsing Binary stars

Hi there, I'm Nikki 👋

Roles

• 🔭 Postdoctoral researcher at Uppsala University, Sweden
• ⭐ Astronomer interested in precision stellar astrophysics, eclipsing binary stars, M-dwarf characterisation, exoplanets, and interstellar reddening
• ✨ Contributor to PLATO mission, previous member of the Araucaria project
• 🌡️ Developer and maintainer of teb, a code for measuring accurate fundamental effective temperatures for binary stars

Read more

• 🕸️ https://nikkimiller.space/
• 🎓 Curriculum Vitae
• 🆔 ORCID
• 📝 List of publications on NASA ADS

teb's Issues

NIR flux ratio priors overhaul

• Support for non-FG type stars (new catalogue integration)
• Print info on number of stars used in fit to check it's using a reasonable amount
• Diagnostic plots for more verbose output

Better way to use radius ratio(?)

Hi Nikki,
I think we might be using the radius ratio information incorrectly. By calculating the log-likelihood using theta1, theta2 and k=r2/r1, we are effectively using the information from the light curve fit twice because r1 and r2 are both used to calculate theta1 and theta2.

Perhaps better to use k to directly calculate theta2 and drop the extra term in the calculation of the log-likelihood for k, i.e.,

In functions.angular_diameters,

    if config_dict['apply_k_prior']:
        print('Using k instead of r_2')
        r2 = list_to_ufloat(config_dict['k']) * r1
    else:
        r2 = list_to_ufloat(config_dict['r2'])

Then delete the following in functions.lnprob

    if config_dict['apply_k_prior']:
        k_prior = list_to_ufloat(config_dict['k'])
        lnprior += -0.5*((theta2/theta1 - k_prior.n)/k_prior.s) ** 2

This did increase the error on Teff1 from 39.4 K to 48.0 K for an EBLM system I an analysing.

Breaks if catalogue magnitude has value but no error

Noticed for a star with nan error in W4 band

Corner plot file overwritten by distortion plot

correlation parameter for systems with partial eclipses

reload=False not working in flint.py

flux_ratio_priors does not exclude missing values from masked table in fitcol

Add the following to the fitcol function

    y = V0 - m0
    try:
        x = x.filled(np.nan)
    except:
        pass
    try:
        y = y.filled(np.nan)
    except:
        pass
    i = np.isfinite(x) & np.isfinite(y)

Generalised NUV/FUX zeropoint

Currently is hardcoded in Flux2mag class; changing to function to automatically calculate correction given raw observed magnitudes

Output plot of distortion polynomials not working

Better treatment of parallax

Option for Gaia or other (e.g. LBI) parallax, along with options for manual ZPs or to automatically apply the ZP corrections proposed by Lindegren, Flinn, etc.. Currently parallax ZPs are hard-coded for a particular star.

Update TESS response function

New TESS response function available from
https://heasarc.gsfc.nasa.gov/docs/tess/data/tess-response-function-v2.0.csv

More choice for input SEDs

Will try to include the following SED catalogues in a future update:

• Coelho high resolution models (limited wavelength coverage; might work with some wrangling)
• Kurucz ODFNEW models (good parameter and wavelength coverage; less dense in wavelength space but may be ok)
• Kesseli models (same problem as Coelho)

Might also be a fun challenge to give the option to restrict the BT-Settl wavelength range to see how the distortion polynomials handle it, since the other models also require a different wavelength range to that currently in use.

Include Kepler band as standard

For flux ratio routines. Should be similar to TESS

Create "output" directory if it does not exist ...

... otherwise you get an error on this line ...
File "/Users/pflm/J0113+31/teb/functions.py", line 527, in convergence_plot
plt.savefig(fname)

Quality-of-life/automation improvements

• Option to run code based on tag used in -m argument (i.e. teb.py -m Vega; followed by teb.py -d Vega instead of teb.py -c Vega.yaml -p Vega_photometry_data.yaml etc.). Should still keep option to specify different files
• Routine to retrieve STILISM E(B-V) estimate if none provided
• Routine to retrieve Gaia parallax if none provided
• Option to automatically generate NIR FRP values based on starting values in main config file; i.e. just specify +/- range for each parameter and main code will deal with the rest. Still keep manual option
• Rosseland correction - at minimum, calculate % error on given radii and print warning for user to check they have the correct radii. Ideally, an automatic calculation, but it's easier said than done.
• Automatically estimate systematic error from the flux scale zero point (Bohlin) on finish
• Guidance (or suggestion) for alpha fraction could be useful

machine-friendly output of fit parameters

option to load MCMC chain and continue

Colours functionality broken

When I try to use colours (Stromgren) the crashes with:

c_2,2 = 0
c_2,3 = 0
Traceback (most recent call last):
 File "/home/bs/teb-master/teb.py", line 170, in <module>
    lnlike = lnprob(params, f2m, flux_ratios, 
  File "/home/bs/teb-master/functions.py", line 337, in lnprob
    chisq, lnlike_m, lnlike_c = flux2mag(wave, flux, sigma_ext, sigma_col)
   ValueError: not enough values to unpack (expected 3, got 2)          

All you need to do to recreate the error is, in config.yaml, set

apply_colors: True            # Include any colors data in photometry_data.yaml?

In photometry_data.yaml

# Photometric colors
colors_data: 
    - tag: b-y
      type: by  
        color: 
        - 0.0    
        - 0.1    

Catch case where no GALEX fluxes available

I had to comment out the GALEX query block in flux2mag.py because none exists for the system I am looking at. The error was confusing, "Star name not resolved by SIMBAD" rather than the except "Unable to find results... GALEX ...".