spectroscopy: does identify weigh lines?
Gabor Furesz wrote on May 08, 2006
Hi,
Can one tell me if the identify/reidentify tasks weigh the lines or not? I have found not much about it in the help.
The IRAF ThAr line list does not have an "intensity" column, which would tell what lines have high counts so therefore more accurate positions. I don't see a sign that identify would mesure and use the line intensity in the fit. In case of a MOS with a very fast camera, it seems to be a problem. Some of the lines I have have positions accurate to 0.01 pixel, but others have less intensity and their measured positions vary by +/- 0.05-0.1 pixel. It would be very important that the dispersion solution rely more on the accurate lines, and use th eother lines with less weight.
Thanks for any help,
gabor
Can one tell me if the identify/reidentify tasks weigh the lines or not? I have found not much about it in the help.
The IRAF ThAr line list does not have an "intensity" column, which would tell what lines have high counts so therefore more accurate positions. I don't see a sign that identify would mesure and use the line intensity in the fit. In case of a MOS with a very fast camera, it seems to be a problem. Some of the lines I have have positions accurate to 0.01 pixel, but others have less intensity and their measured positions vary by +/- 0.05-0.1 pixel. It would be very important that the dispersion solution rely more on the accurate lines, and use th eother lines with less weight.
Thanks for any help,
gabor
Mike Fitzpatrick wrote on May 08, 2006
Frank will correct me if I'm wrong, but my understanding is that all points are initially uniformly weighted. The 'v' keystroke can be used to assign a weight to a particular feature.
-Mike
-Mike
Gabor Furesz wrote on May 08, 2006
Thanks, Mike. I figured out the "v" keystroke, but my problem is I need to add weight in a script which runs reduction on dozens of images, each having 240 spectra... So I need an automated way of setting weight. Anybody knows anything about it?
thanks,
gabor
thanks,
gabor
Gabor Furesz wrote on May 08, 2006
Hi,
Here is my answer to my quiestion, thanks to Maureen Conroy, who looked at the source code of reidentify and told me the following:
the database id file looks like this, for a given aperture:
# Wed 20:26:52 07-Jun-2006
begin identify COMP_f.ms - Ap 1
id COMP_f.ms
task identify
image COMP_f.ms - Ap 1
aperture 1
aplow 29.99
aphigh 35.99
units Angstroms
features 44
4.82 6646.54164 6646.5407 4.0 1 1
30.35 6644.66133 6644.6635 4.0 1 1
43.41 6643.69688 6643.6976 4.0 1 1
96.70 6639.74356 6639.7403 4.0 1 1
107.84 6638.91354 6638.9119 4.0 1 1
117.14 6638.21926 6638.2207 4.0 1 1
and so on......
where the last column is the actual weight, what is 1 by default. If you have a reference id file you use in reidentify (under the current database directory), than you can go into this text file and with a script --based on the wavelength as identification-- one can edit the last column and set a given weight for a given line, in all aperures if multispectra file. The weight can be a float, doesn't have to be an integer. After this file is preapred, and reidentify is run, it will read the user set weights and apply those in the fit.
So, that's tha way to set weight for lines automatically in a data reduction pipeline: prepare a 'master id reference' file and call it as a reference.
Thanks for Maureen Conroy (Harvard-Smithsonian C fA) again!
gabor
Here is my answer to my quiestion, thanks to Maureen Conroy, who looked at the source code of reidentify and told me the following:
the database id file looks like this, for a given aperture:
# Wed 20:26:52 07-Jun-2006
begin identify COMP_f.ms - Ap 1
id COMP_f.ms
task identify
image COMP_f.ms - Ap 1
aperture 1
aplow 29.99
aphigh 35.99
units Angstroms
features 44
4.82 6646.54164 6646.5407 4.0 1 1
30.35 6644.66133 6644.6635 4.0 1 1
43.41 6643.69688 6643.6976 4.0 1 1
96.70 6639.74356 6639.7403 4.0 1 1
107.84 6638.91354 6638.9119 4.0 1 1
117.14 6638.21926 6638.2207 4.0 1 1
and so on......
where the last column is the actual weight, what is 1 by default. If you have a reference id file you use in reidentify (under the current database directory), than you can go into this text file and with a script --based on the wavelength as identification-- one can edit the last column and set a given weight for a given line, in all aperures if multispectra file. The weight can be a float, doesn't have to be an integer. After this file is preapred, and reidentify is run, it will read the user set weights and apply those in the fit.
So, that's tha way to set weight for lines automatically in a data reduction pipeline: prepare a 'master id reference' file and call it as a reference.
Thanks for Maureen Conroy (Harvard-Smithsonian C fA) again!
gabor
mz79 wrote on May 08, 2006
Hi,
With this workaround the task identify allows for a weighted dispersion solution from a line list. It works fine.
Can one also create a new weighted line list from a wavelength calibrated emission spectrum?
With
one can export the position, but not the intensities (as weights) of the new features.
Is there a (other) way to do this?
Cheers
With this workaround the task identify allows for a weighted dispersion solution from a line list. It works fine.
Can one also create a new weighted line list from a wavelength calibrated emission spectrum?
With
:maxfeature <value>
y
:features
one can export the position, but not the intensities (as weights) of the new features.
Is there a (other) way to do this?
Cheers
Last post on May 08, 2006