NAUG comments
- imageprofilefitter save -- DS
- imfitter:=imageprofilefitter(infile='in.cm', infile2='in.mom0',
axis=4); click Save: in profilefitter GUI. You now have the option
to specify a text file to save spectra (vel vs. data) and fits.
Only available in daily aips++ right now.
11mar04
12mar04
- functionality: good, 2 enhancements requested:
- would not work if I did not have fits created, got an
error message: SEVERE: Data and model/fit not of same
length But I did not want to save fits, just data.
- I want the first column written out to always be the
channel number.
- usability: excellent
- documentation: This is a very new implementation in the
daily version of AIPS++, I did not see documentation
available on this in the imageprofilefitter - Tool web
site. So I cannot grade this part.
Additional comments -- JH 12Mar04
I'll concur with Debra's assessment of imageprofilefitter.
I've done some extensive testing of the results of the fits
vs. (1) AIPS SLFIT; (2) Harvy's DRAWSPEC; and (3) IDL GAUSSFIT.
I found excellent agreement with all of these, with the
exception that AIPS SLFIT gave errors which were about
a factor of 2 too low compare to the others (interestingly,
this is opposite what Walter found for errors reported by
imagefitter and JMFIT). Detailed results in resolved Defect
AOCso04745.
I was working with a velocity cube written out from AIPS that
had velocity on the 3rd axis. As we know, cubes are equal
width in frequency, not velocity, so aips++ recalculates
the frequency. Not exactly sure how it does this, but it
doesn't give you what you want, and you have to play
with the coordinate system to get the 3rd axis right
for external velocity cubes (wonder if this is only a problem
with velocity cubes from AIPS, or also from e.g. GMRT, ATCA,
IRAM, etc?). Note that images created in aips++ do not have
this problem. Comments on these are in Defect AOCso04744.
As for getting the fit parameters out in a useful format, I've
entered enhancment request AOCso04764 for the documentation to
either link to the chapter of glish records, or give a direct
example. FYI, its pretty easy to do:
click "copy" in imageprofilefitter
-results:=dcb.paste()
-print results.fit.elements
[*430=[type=GAUSSIAN, parameters=[0.010283893 860.92656 31.777157] ,
errors=[0.00045121694 0.68385197 1.6101082] ]]
The enhancement also request that someone tell me how to
access subfields in this record, e.g., something like
-print
results.fit.elements.parameters[0],'+/-',results.fit.elements.errors[0]
If one of you know, then please tell me.
John
this is one of Glish's more perverse features. When a record is created where
the fields are not explictly named, but just sequentially added, the field
names can be very odd as you saw. However, they can always be indexed with
sensible 1-based (glish is not zero based) indices r.fit.elements[1]
r.fit.elements[2]
etc. So I aded a little bit to the docs on this -- NK