Spectrum Fitting (Payne NLTE)

  • Home
  • Contact
  • Login/Register
  • References
  • Quick Start

    1. Choose a Payne model:
      • 4most-hr — R≈20k. 3 windows: 3926-4355, 5160-5730, 6100-6790 Å. Δλ = 0.05 Å.
      • 4most-lr — R≈7k. 1 window: 3700-9450 Å. Δλ = 0.25 Å.
      • 80k — R≈80k; 3 windows: 3925-6800, 7700-7800 (for O 777 line), 9050-9130 (for some C I lines) Å. Δλ = 0.05 Å. When selected, a Resolution (R) field appears—enter your spectrum's resolving power to degrade Payne to the requested resolution (slower fitting though) or leave blank.
    2. Provide the observed spectrum:
      • Either upload your own file, or choose a preset.
    3. Set fit parameters:
      • Good guess speeds convergence; tick Fit to include a parameter.
      • Min/Max default to the network's valid bounds—don't widen them (extrapolation is unsafe).
      • vmac/vsini/resolution convolutions slow down the fitting the most. Only turn on the ones you need or improve the guess/reduce fitting range.
      • RV cannot be fitted well, unless you set a larger Δλ. Small RV shifts will be fitted without any changes.
    4. Line masks:
      • 📝 opens the mask editor. One line per region: first column = λcenter, second = Δλ. Δλ automatically increases with vsini for indidual element fitting.
      • 📐 activates plot selection. Drag over a region and it's appended to the mask.
      • Stellar parameters (teff, logg, [Fe/H], vmic, broadening) share the same masks; element abundances are fitted with their own masks (individually).
      • Masks are not persistent, so please save them locally if you need to use them later.
    5. Run & iterate:
      • Click Run fitting. Results appear in Fitted with ±σ. σ at the moment is often underestimated. Pure spectroscopic teff and logg might have systematic uncertainty on the order of up to ~100-150K and ~0.25 dex
      • ALL stellar parameters (teff, logg, [Fe/H], vmic, broadening) are fit together, by default using Fe, Mg I b triplet and Ca I triplet lines (default for 4MOST-HR, add more if needed).
      • Use the ← arrow to copy a fitted value back into Guess.
      • Inspect the plot; you can add masks via 📐 and re-run if needed.
      • NaN after the fit means either that the line failed to fit (e.g. no linemask within the wavelength range). Or the line is too weak (EW above the minimum bound is around 1-2 on average for all lines).
    6. Paper citation:
      • Storm et al. in prep. This is a beta version, please report any bugs, problems etc. Please wait until publishing of the paper to use this tool for scientific results.

    Observed spectrum

    Fit parameters

    Neural Network Model library

    PLEASE READ README IN FULL BEFORE USING THIS.

    Fit the following parameters.

    To reset all parameters to their default values, click the button below.

    Parameter Fit? Guess Min Max Fitted ±σ Mask
    RVNaNNaN
    TeffNaNNaN
    loggNaNNaN
    [Fe/H]NaNNaN
    vmicNaNNaN
    vsiniNaNNaN
    vmacNaNNaN
    A(Li)NaNNaN
    [C/Fe]NaNNaN
    [O/Fe]NaNNaN
    [Na/Fe]NaNNaN
    [Mg/Fe]NaNNaN
    [Al/Fe]NaNNaN
    [Si/Fe]NaNNaN
    [Ca/Fe]NaNNaN
    [Ti/Fe]NaNNaN
    [Cr/Fe]NaNNaN
    [Mn/Fe]NaNNaN
    [Co/Fe]NaNNaN
    [Ni/Fe]NaNNaN
    [Sr/Fe]NaNNaN
    [Y/Fe]NaNNaN
    [Ba/Fe]NaNNaN
    [Eu/Fe]NaNNaN
    Running fit - please wait…

    Mask


webmaster: This service is developed by MPIA