This wiki describes the zoology investigations of the false positives when looking at the images (not necessarily the flags).

== First observations ==

Many of the stack skycells show a bias (more false detections on the top or bottom half of a cell). These biases tend to run horizontally on a chip and often do not continue onto neighboring chips.  

== Plotting pixel/sqrt(variance) for cells and subset of cells ==


The stack that I investigated was stack 336346 (skycell.1316.089).  I chose a square that had a lot of false detections (green), and one that did not (red):

[[Image(falsdetbias.jpg)]]

I then chopped the cell that had a lot of false detections into two (top half = red, bottom half blue), and divided the pixel values by the sqrt of the variance.  A gaussian was fit to each of these parts, and it is in log space (y):

[[Image(differentfitsfalse.jpg)]]

* black = the cell with few false detections, 
* blue = the bottom half of the cell with the false detections
* red = the top half of the cell with the false detections

== plotting psf_mag vs chisq ==

These are plots of inst_psf_mag vs chisq for g band (if one of psf_mag or ap_mag is NULL, it is ignored). The other bands look very similar.  The first plot is for the true detections, the second for the false detections, and the third is for both (true = black, false = red).  The only pattern I see is that faint = fake.

[[Image(false.det.pix-0.jpg)]]
[[Image(false.det.pix-1.jpg)]]
[[Image(false.det.pix-2.jpg)]]

2d histogram plots of these (ignore the axis):

 * x axis goes from -20mag to 0 in increments of (bins of) .1
 * y axis goes from 0 to 10000 in bins of 100

[[Image(falsedet.color-0.png)]] 
[[Image(falsedet.color-1.png)]] 



== plotting psf_mag vs psf_mag - ap_mag ==

These are plots of inst_psf_mag vs inst_psf_mag - ap_mag for g band. Similar to the other plots, the entries with one of psf_mag or ap_mag = NULL are ignored, and it looks very similar for all bands. 

[[Image(false.det.pix-3.jpg)]]
[[Image(false.det.pix-4.jpg)]]
[[Image(false.det.pix-5.jpg)]]

== repeating this experiment for other bands (same location) == 

Initially, I did the histogram plots as above, for the same skycell. This was not interesting: Only g band had lots of extra detections. I went hunting around in r band until I found a skycell in lots of cells, and checked that there were lots of extra detections in the other bands (for the same skycell).  Except for y, all the bands had the same chip/orientation for the skycell.  For g,r,i,z, it suggests that the false detections often lie on the same cells on a chip (XY27).  If you look (from the top right) down, the rows are:

 * XY37: row 6
 * XY37: row 7

 * XY27: row 0 (pattern.row = on)
 * XY27: row 1 (pattern.row = on)
 * XY27: row 2 (pattern.row = on) 
 * XY27: row 3 (pattern.row = on)
 * XY27: row 4 
 * XY27: row 5 
 * XY27: row 6
 * XY27: row 7 (pattern.row = on)

I can't find a pattern between where there are false detections and whether pattern.row was run or not.

I will make histogram plots (similar to above) tomorrow.
  

g

[[Image(336336.g.png)]]

r

[[Image(336377.r.png)]]

i

[[Image(336440.i.png)]]

z

[[Image(329679.z.png)]]

y

[[Image(330343.y.png)]]

== histogram plots of some interesting cells from above == 

The colors are as follows:

 * top cell = lots of false detections (pattern.row run on it) = blue
 * middle cell = few false detections (pattern.row run on it ) = black
 * bottom cell = lots of false detections, no pattern.row  = red

 * red/blue are consistently shifted to the right from the one with few false detections. 

|| g || [[Image(gband.jpg)]]  ||[[Image(gbandhist.jpg)]] || 
|| r ||  [[Image(rband.jpg)]]  ||[[Image(r.hist.jpg)]]  || 
|| i || [[Image(iband.jpg)]]  ||[[Image(i.hist.jpg)]]  ||
|| z || [[Image(zband.jpg)]]  || [[Image(z.hist.jpg)]]  ||