search

MingjieJian / pymoog

Python wrapper for MOOG
MIT License
11 stars 4 forks source link

Interpolation of the stellar atmosphere model #9

Closed MingjieJian closed 4 years ago

MingjieJian commented 4 years ago

We can try to interpolate the model to cover the non-grid point location in the stellar parameter place. I don't have a good idea on the format but we can start from linear. We should also try to find better models - the initial Kurucz one may be outdated. Can start from http://adsabs.harvard.edu/abs/2012AJ....144..120M, also MARCS.

MingjieJian commented 4 years ago

I compared the models available in PyAstronomy and in Kurucz. There are model sets with vmicro not equal to 2 for soalr metallicity, but PyAstronomy does not include them. Also it is not accurate to judge the grid point from PyValError:

PyAValError:

A PyA error occurred:

Type of error: PyA Value Error What happened? No model available in grid for parameters: teff = 5.000e+04, logg = 5.000e+00, met = -2.000e-01 What are possible solutions?

Some grid points are not available (e.g., Teff>40000, logg=4.5) though they are listed in here.

For low temperature model (Teff <= 8750), there are two extra columns in the model; don't know what it is.

Anyway, the content downloaded from PyA and from Kurucz website is the same, so we can trust it.

DanielAndreasen commented 4 years ago

Hi there,

This looks like an interesting project. I have used MOOG extensively during my PhD myself, and also made a lot of wrappers. For interpolation this might help you: https://github.com/DanielAndreasen/FASMA/blob/master/interpolation.py That is how we (some colleagues and I) made it.

That repository also does minimization using the abfind driver, in order to obtain stellar parameters. We have a similar project for the synth driver here: https://github.com/MariaTsantaki/FASMA-synthesis

Lastly, I would like to point to two smaller projects I have made, which you might also benefit from: https://github.com/DanielAndreasen/MOOGly and https://github.com/DanielAndreasen/synhtesis-plotter.

If you need help with anything, then I'm available (not affiliated anymore, so just an amateur astronomer looking for cool projects). And I hope you can use some of the above links as inspiration.

MingjieJian commented 4 years ago

Hi there,

This looks like an interesting project. I have used MOOG extensively during my PhD myself, and also made a lot of wrappers. For interpolation this might help you: https://github.com/DanielAndreasen/FASMA/blob/master/interpolation.py That is how we (some colleagues and I) made it.

That repository also does minimization using the abfind driver, in order to obtain stellar parameters. We have a similar project for the synth driver here: https://github.com/MariaTsantaki/FASMA-synthesis

Lastly, I would like to point to two smaller projects I have made, which you might also benefit from: https://github.com/DanielAndreasen/MOOGly and https://github.com/DanielAndreasen/synhtesis-plotter.

If you need help with anything, then I'm available (not affiliated anymore, so just an amateur astronomer looking for cool projects). And I hope you can use some of the above links as inspiration.

Hi Daniel,

Thank you for pointing me to the repos! I am also using MOOG frequently now and decided to wrap up my functions to help others using MOOG easily. I will definitely look into your repos and use it to imporve this one.

MingjieJian commented 4 years ago

Caveat 1: The usual model title is some thing like:

TEFF 8750. GRAVITY 2.00000 LTE TITLE SDSC GRID [+0.0] VTURB 2.0 KM/S L/H 1.25

pymoog use [+0.0] to set the metallicity value.

But for Teff, logg, [Fe/H] = 8750, 1.5, 0.0, the title is:

TEFF 8750. GRAVITY 1.50000 LTE TITLE TEST RUN WITH SDSC OPACITY VTURB 2 KM/S L/H 1.25

Haven't check if other title have similar behaviour. Solution: localize the Kurucz models and modify the header of this one.

Caveat 2: The Kurucz model file for metallicity -5.0 (http://kurucz.harvard.edu/grids/gridm50/am50k2.dat) and -4.5 have a different format than others. The first character of each line is gone, making "TEFF" -> 'EFF, etc. Solution: use .dat files. They still fit the Kurucz format and can be converted to MOOG format.

Caveat 3: The model with Teff=5000, logg=1.0 and [Fe/H]=-3.0 is missing in Kurucz's data base (.dat) file, but appear in .datcd file. Since the model around have 72 lines in .dat but only have 64 lines in .datcd file, we filled this grid point by interpolation. Solution: interpolated this grid from those with logg=0.5 and 1.5.

Caveat 4: When Teff >= 9000 the model length change from 72 to 64 for models in .dat files (those in .datcd files are all 64). Teff=7750, logg=1.5, [Fe/H]=-5 also have a length of 64. Solution: cut the first 8 lines of long table and perform usual interpolation.