[daip] JMFIT

[Jim Ulvestad]

Hi all,

I think I just found a sneaky little JMFIT error, which
probably could drive someone crazy trying to find.
I am fitting a single component, and here's the output I 
get:

coyote> JMFIT1: Task JMFIT  (release of 31DEC03) begins
coyote> JMFIT1:           CHANNEL   1.
coyote> JMFIT1: Fit to histogram gives RMS = 1.001E-04
coyote> JMFIT1: Model fit to NAC4151-VLBA.ICL001.   1
coyote> JMFIT1: Source= NGC4151     Units are JY/BEAM
coyote> JMFIT1: X-ref pix=  78.7  RA 12 10 32.58220   pix sep (Asec)
-0.0002
coyote> JMFIT1: Y-ref pix= 513.0  DEC 39 24 21.0590   pix sep (Asec) 
0.0002
coyote> JMFIT1: Axes 3-7 pixels =      1      1      1      1      1
coyote> JMFIT1: Axis 3 : IPOL
coyote> JMFIT1: Axis 4 : 4987.474 MHZ
coyote> JMFIT1:  - - - - - - - - - - - - - - - - - - - - - - - - - - - -
- -
coyote> JMFIT1: ******** Input Model
****************************************
coyote> JMFIT1: Component=  1  Gaussian
coyote> JMFIT1:       Peak= 4.0000E-03    JY/BEAM
coyote> JMFIT1:       Xpos=  572.00      Ypos=  521.00     pixels
coyote> JMFIT1:       Size=  21.57    x   6.36    Pixels at pa -36.59   
deg
coyote> JMFIT1:  - - - - - - - - - - - - - - - - - - - - - - - - - - - -
- -
coyote> JMFIT1:       Beam=  21.57    X   6.36    pixels: in PA 
-36.6    deg
coyote> JMFIT1: Initial guess RMS  4.8892E-04 in  899 usable pixels
coyote> JMFIT1: Fitting to  899 data points for  6 parameters
coyote> JMFIT1: Initial Chi-squared, R.M.S., and gradient norm:
coyote> JMFIT1:               2.14896E-04    3.69517E-07    4.20776E-01
coyote> JMFIT1: ***  Convergence achieved.
coyote> JMFIT1: Solution RMS  2.0668E-04 in  899 usable pixels
coyote> JMFIT1: Post-fit Chi-squared, R.M.S., and gradient norm:
coyote> JMFIT1:               3.84011E-05    2.07370E-04    7.25477E-06
coyote> JMFIT1: Writing MF file version   1 attached to input image
coyote> JMFIT1: ********* Solution from JMFIT
*********************************
coyote> JMFIT1: Component   1-Gaussian
coyote> JMFIT1:   Peak intensity    = 3.4488E-03 +/-  9.63E-05 JY/BEAM
coyote> JMFIT1:   Integral intensity= 5.7932E-03 +/-  2.39E-04 JANSKYS
coyote> JMFIT1:   X-position        =    570.537 +/-   0.1996 pixels
coyote> JMFIT1:   Y-position        =    519.695 +/-   0.2150 pixels
coyote> JMFIT1:                       RA 12 10 32.57583    +/-  0.00000
coyote> JMFIT1:                       DEC 39 24 21.0600    +/-  0.00003
coyote> JMFIT1:   Major axis        =  22.539 +/-  0.6291 pixels
coyote> JMFIT1:   Minor axis        =  10.219 +/-  0.2852 pixels
coyote> JMFIT1:   Position angle    = 138.232 +/-  1.291 degrees
coyote> JMFIT1:   Major axis        =  0.00338 +/-0.00009 asec
coyote> JMFIT1:   Minor axis        =  0.00153 +/-0.00004 asec
coyote> JMFIT1:   Position angle    = 138.232 +/-  1.291 degrees
coyote> JMFIT1:   - - - - - - - - - - - - - - - - - - - - - - - - - - -
- - -
coyote> JMFIT1:                   Deconvolution of component in pixels
coyote> JMFIT1:                        Nominal     minimum     maximum
coyote> JMFIT1:     Major ax             9.61        0.00       10.89
coyote> JMFIT1:     Minor ax             3.81        1.27        9.74
coyote> JMFIT1:     Pos ang             87.91       73.01      143.41
coyote> JMFIT1:                   Deconvolution of component in asec
coyote> JMFIT1:                        Nominal     minimum     maximum
coyote> JMFIT1:     Major ax          0.00144     0.00000     0.00163
coyote> JMFIT1:     Minor ax          0.00057     1.26941     0.00146
coyote> JMFIT1:     Pos ang          87.91333    73.01111   143.41319
coyote> JMFIT1:   - - - - - - - - - - - - - - - - - - - - - - - - - - -
- - -
coyote> JMFIT1: Appears to have ended successfully
coyote> JMFIT1: coyote       31DEC03 TST: Cpu=       0.2  Real=       1

Note the last lines.  In the pixel deconvolution, the minor axis is
given as 3.81 pixels, with min/max = 1.27/9.74.  All seems okay.
Then the deconvolution in arcseconds is 0.00057.  Okay, since the
pixel size is .00015.  And the max. is okay.  But for the arcsecs,
it just gives me back the number of pixels in the min (with more digits)
rather than converting to arcseconds.  I've never seen JMFIT do
this before, and it's not doing it on other fits I've been doing
this morning.  If I had to guess, based on other fits (and on doing
this one after changing the window slightly), I'd say that if the
major axis has a minimum size of 0.00, the program is assuming that
the minor axis has a minimum of 0.00, so it doesn't have to go and
convert the number of pixels to arcseconds (since 0 times anything
is 0).

ju