FWHMs with FXCOR
m_alp wrote on Aug 13, 2009
Hello everyone,
I've been trying to use the FXCOR task to obtain some velocity dispersions for some galaxies. I am correlating them against a couple of template stars.
Anyway, I've been getting some FWHM values from the correlation plots, but I was wondering how exactly these relate to the actual velocity dispersions of the galaxies in question? Is it a direct relation, or is there something more intricate involved?
Thanks in advance,
Mehmet
I've been trying to use the FXCOR task to obtain some velocity dispersions for some galaxies. I am correlating them against a couple of template stars.
Anyway, I've been getting some FWHM values from the correlation plots, but I was wondering how exactly these relate to the actual velocity dispersions of the galaxies in question? Is it a direct relation, or is there something more intricate involved?
Thanks in advance,
Mehmet
Mike Fitzpatrick wrote on Aug 13, 2009
The FWHM of the correlation function is related to the average width of the spectral lines, and since you may have blended features or large lines like
H-beta this could throw off the values greatly. The velocity dispersion is a function of both the true dispersion and the instrumental width, by correlating stars with stars you can get an idea of the latter. Accuracy also depends a bit on the line shapes, i.e. the convolution of a Gaussian with a Gaussian is a Gaussian, but if your spectral features have broad wings you may be better off fitting e.g. a Lorentzian functon rather than a Gaussian to get a more realistic width value.
In any case, since the data are rebinned to a log-linear dispersion before correlation, the width can be multiplied directly by the velocity-per-pixel dispersion to get a value in km/s, but it's up to you to do a proper analysis to make sure the derived width is reasonable. Hope this helps.
-Mike
H-beta this could throw off the values greatly. The velocity dispersion is a function of both the true dispersion and the instrumental width, by correlating stars with stars you can get an idea of the latter. Accuracy also depends a bit on the line shapes, i.e. the convolution of a Gaussian with a Gaussian is a Gaussian, but if your spectral features have broad wings you may be better off fitting e.g. a Lorentzian functon rather than a Gaussian to get a more realistic width value.
In any case, since the data are rebinned to a log-linear dispersion before correlation, the width can be multiplied directly by the velocity-per-pixel dispersion to get a value in km/s, but it's up to you to do a proper analysis to make sure the derived width is reasonable. Hope this helps.
-Mike
m_alp wrote on Aug 13, 2009
Thanks for the information, Mike. I've been correlating mostly with Na and Mg lines, so I hope my results are relatively accurate.
My main source of confusion is whether or not I can take the FWHM given by FXCOR and use it as the velocity dispersion of the galaxy I'm looking at, or is there more analysis involved?
My main source of confusion is whether or not I can take the FWHM given by FXCOR and use it as the velocity dispersion of the galaxy I'm looking at, or is there more analysis involved?
Last post on Aug 13, 2009