Revisit Rectification Settings
Today's goal: experiment with giving twinsyn a wider slice of the spectrum to see if rectification and results are improved.
I created a wider (5115 - 5195 Å) synthetic binary spectrum from the solar template. I then ran twinsyn once with these input files and again with these input files. The first set of files only includes the slice of the spectrum we have been using until this point. The second set includes the wider band. I only expect results to change because I used the following rectification string during the second run:
rect_str = '5116.5-5119,5134.5-5135.5,5156.5-5160.5,5175-5180,5186-5187,5189.5-5190',
I found that the same (incorrect) output spectrum had the lowest chi-square:
0.80 1.20 6000. 0. 30. 1.40 1.20 6000. -53. 10. 0.23767622E+00
The spectra with the lowest chi-square for both the original rectification string (5175-5180,5186-5187; blue) and the extended string (pink) are plotted against the input file rectified with the second string is shown below.
The spectra (shown below) that were produced with the parameters that match the sun almost had the lowest chi-square values, e.g., 0.26434779E+00 for the short rectification string spectrum.
A few thoughts:
It's possible that the Katelyn's solar template has a sight blueshift. Nick's searches through velocity parameter space suggest that the original spectrum was shifted by about -4 km/s, which could explain why the above spectra don't line up perfectly. Tomorrow I'll test with some of 9020's spectra, which have known radial velocities. When Katelyn originally found those velocities she might have used this red template as a reference. I'll have to ask her about it.
It seems that we'll be able to us the rotational velocity parameter to account for resolution issues. Hopefully we can use iraf to find the typical FWHM for our lines in 9020 spectra so we can set v_rot accordingly.
I didn't create the synthetic binary spectrum correctly. I took the sun's flux array that was evenly spaced in wavelength space and combined the flux from two (mostly-overlapping) sections of that array. The doppler shift should have been wavelength dependent. Instead, all of wavelengths for the secondary were shifted by 0.89740653 angstroms. If we calculate the velocity of the secondary based on the shift of the part of the spectrum at 5187.9394 Å, it would be -51.857488 km/s. If we calculate the velocity of the secondary based on the shift of the part of the spectrum at 5164.3504 Å, it would be -52.094355 km/s. The question is, has this mistake caused problems when we tested twinsyn? I think the answer is no. If I were working with a much wider spectrum, it would be an issue.