PNe8 - New box-selection methods!

Plan:
Select random boxes. For the 1+ PNe, make sure random boxes have at least 1 PNe in them, otherwise select a new box.
                                For 0 PNe, make sure randoms have 0 PNe in them, or throw them out.
                                For Randoms, just pick random boxes.

Big difference in this method: PNe no longer have to be centered in the boxes for 1+ PNe detected. In previous methods, our 1 PNe selections were centered on each PNe. Now, they just contain 1+ PNe, anywhere in the box.

In this iteration, we used the same number of boxes for each field as there are (detected) PNe in it (from the PNe catalog from Feldmeier & Okamura). For the fields, the # of PNe (and therefore also the number of boxes selected) are:
Field 3 : 75 PNe
Field 4 : 11 PNe
Field 7 : 81 PNe
Field Sub : 34 PNe
Field Core: 108 PNe
Field FCJ: 38 PNe
Field LPC: 14 PNe (only 5 are in unmasked areas)

There are options in this code: If PNe.param file is set properly, there CAN be mandated spacing of boxes. In this iteration, we allow boxes to overlap and have no restrictions
There is also a limitation so that if random boxes canNOT be found with 1+ or 0 PNe in them, the program will only make a predefined number of attempts to find them. This limit of attempts is set to be 10,000 times the number of PNe in the field and has never been met in any iterations yet.

ERROR BARS on these plots come from the distributions of whatever it is being plotted. That means for Deltas, the error bars are from the sigma and quartiles found across Delta values in various trials of the program, not some quadrature or otherwise addition of error.

(skip to results)

Code --> PNe.sm - PNe   -   initialises some variables, sets up loops, figures out how many boxes to use [# of PNe in field], calls other programs to run
                               PNew -  preps fits data, reads in PNe catalog, preps vectors to store pixels
                               PNe1  -  looks for boxes until it's found enough to meet the requirement:
                                                 pick a random location / convert x,y to RA,DEC
                                                 find distance to all PNe / see if any are close enough
                                                 make sure box isn't overlapping any other boxes, if applicable
                                                 save coordinates / concatenate pixels
                                                 make sure we're not stuck in a loop without getting results
                                                 write out coordinates of selected boxes to [ex] 8.1.Im3.fits.R10.wfound.dat, where 1 is iteration #
                               PNe0  -  basically same ideas as PNe1, just set to find boxes with NO PNe, instead of 1+ [thus it's not linked to as heavily, peruse at will]
                                             finds random boxes with 0 PNe in them
                                             collects unmasked pixels from those boxes
                                             saves coordinates to 8.1.Im3.fits.R10.0found.dat
                               PNer  -   just picks random boxes
                                             collects unmasked pixels from those boxes
                                             saves coordinates to 8.1.Im3.fits.R10.Rfound.dat
                               PNeg  -  graphs histograms of found pixels
                                             saves median values of each method as 8.1.Im3.fits.R10.med
                                             postscript file: 8.1.Im3.fits.R10.ps of the histograms graph

PNeError.sm - PNeError -  makes loops and runs for selected images/ranges, runs other programs
                        PNeErrorc - reads in median values from *.med files and calculates statistics on them
                                             writes statistics out to 8.0.Im3.fits.R10.stats files (0 just means it's for all iterations)
                        PNeErrorg - does graphing
                                             uses vectors set by PNeErrorc to graph average median values for each selection method as well as sigma and upper/lower quartile error bars
                                             saves 8.0.med3.fits.we.ps file
                        qc           -     converts .ps to .gif
                                             copies .gif files to website and Virgo/Web/ folder


Results:                                           
                           Perhaps these (1+ PNe) should be converging with the previous method (link to PNe7) at small box sizes?  Maybe MC the error bars on this method by sub-sampling the PNe catalog/selections?   Look into this at some point in the future....

Quick links to fields:
3/3m
4/4m
7/7m
Sub/Subm
Core/Corem
LPC/LPCm
FCJ/FCJm


Curious about that spike at small box sizes? Check out pixel intensity histograms for more info.


     
    



     
     
Curious about that spike at small box sizes? Check out pixel intensity histograms for more info.


     
     
Curious about that spike at small box sizes? Check out pixel intensity histograms for more info.


     
     

Curious about that spike at small box sizes? Check out pixel intensity histograms for more info.


     
     


    
          [10 PNe were used]
Or, a more complete view (manually entered positions - the red ring represents the PN who's given Dec was 11:05:10.42 but I changed to 12:05:10.42 just to see where it fell in the image - the original data was NOT altered)
[circles are larger too]   [ there are two more PNe off this view, and actually not in our field, to the right of this image ]
     
     


     
     


Back to Main Virgo Page