Minbias run with the EEMC SMD in the datastream. All towers were used in the following algorithm:
If there is a MIP (energy between 0.2 and 0.5 GeV) in the tower but no MIPs in
any adjacent channels, put one point in a 2dhisto for every pre-shower channel
that shows a hit, regardless of sector, phi, creed, etc...
Tower gains were taken from simulation, since tower gains do not yet exist for
this run.
1. Correlation between preshower 1 and tower, sectors 1 and 2
2. Correlation between preshower 1 and tower, sectors 3 and 4
3. Correlation between preshower 1 and tower, sectors 5 and 6
4. Correlation between preshower 1 and tower, sectors 7 and 8
5. Correlation between preshower 1 and tower, sectors 9 and 10
6. Correlation between preshower 1 and tower, sectors 11 and 12
6A. Correlation between preshower 1 and tower, sectors 11 and 12, FULL STATISTICS
7. Correlation between preshower 2 and tower, sectors 1 and 2
8. Correlation between preshower 2 and tower, sectors 3 and 4
9. Correlation between preshower 2 and tower, sectors 5 and 6
10. Correlation between preshower 2 and tower, sectors 7 and 8
11. Correlation between preshower 2 and tower, sectors 9 and 10
12. Correlation between preshower 2 and tower, sectors 11 and 12
13. Correlation between postshower and tower, sectors 1 and 2
14. Correlation between postshower and tower, sectors 3 and 4
15. Correlation between postshower and tower, sectors 5 and 6
16. Correlation between postshower and tower, sectors 7 and 8
17. Correlation between postshower and tower, sectors 9 and 10
18. Correlation between postshower and tower, sectors 11 and 12
19. Correlation between preshower 1 and tower
20. Correlation between preshower 2 and tower
21. Correlation between postshower and tower
The mapping from pre-shower to the towers seems to be correct for all channels that show up.
To do this properly, I had to suppress the background of random coincidences, and that might have
eliminated one or two points from each plot. I was fairly careful to select the baseline low enough
to keep points along the diagonal.
NEW METHOD THAT WORKS!!!
Briefly, the problem I'd had was that my "MIP" selection gave far too many
background counts. That, plus the fact that the number of hits is
strongly dependent on eta, made proving the mapping difficult (since there
were as many random coincidences between pre/postshower hits at low eta
and towers at high eta as there were with the "correct" pre/postshower
hits at high eta).
I used the following algorithm on the same MIP+background 2-d histogram:
(rows are pre/postshower, columns are towers)
Find row/column with highest total integrated number of hits (call this A)
Find highest hit in that row/column
If that hit's on the diagonal
else
Find the column/row that corresponds to A (if A is row, find column)
(for clarity, this has nothing to do with the highest hit)
Find the highest hit in that column/row
If that hit's on the diagonal
but also write out the high hit in the other row/column
this luckily never happens! If it did, write bad things
set the elements of row A *and* column A to zero
repeat until graph is empty
A list of all the "hot" off diagonal elements is given. Please note, that for
the row/column for *every* one of these elements, if you looked at the
corresponding column/row, the highest hit was on the diagonal. The
corresponding hit is given first, followed by the "bad" hit.
This algorithm is largely immune to hot towers, or different tower-to-tower MIP
rates. It is biased to expect the correlation to be on the diagonal. However,
if you look at the list of "bad" correlations, and you ignore channel 274 and
37, all of the "correlated" hits are far enough apart in number such that I
don't believe you could explain it via an electronics swap. Please correct me
if I'm wrong...
I promise to translate these into endcap notation in a day, but right now, I'm
out of time...
"hot tower" correlations were first...
min hit is row 625 and col 625
BAD hit is row 274 and col 625
min hit is row 121 and col 121
BAD hit is row 274 and col 121
min hit is row 385 and col 385
BAD hit is row 274 and col 385
min hit is row 421 and col 421
BAD hit is row 274 and col 421
min hit is row 97 and col 97
BAD hit is row 274 and col 97
min hit is row 61 and col 61
BAD hit is row 274 and col 61
min hit is row 274 and col 274
BAD hit is row 274 and col 37
min hit is row 287 and col 287
BAD hit is row 287 and col 37
min hit is row 290 and col 290
BAD hit is row 290 and col 37
and then a bunch of spurious random hits followed:
min hit is row 381 and col 381
BAD hit is row 493 and col 381
min hit is row 398 and col 398
BAD hit is row 398 and col 282
min hit is row 494 and col 494
BAD hit is row 494 and col 4
min hit is row 4 and col 4
BAD hit is row 5 and col 4
min hit is row 638 and col 638
BAD hit is row 638 and col 292
min hit is row 133 and col 133
BAD hit is row 133 and col 319
min hit is row 578 and col 578
BAD hit is row 578 and col 712
min hit is row 254 and col 254
BAD hit is row 254 and col 458
min hit is row 14 and col 14
BAD hit is row 14 and col 316
min hit is row 244 and col 244
BAD hit is row 530 and col 244
min hit is row 5 and col 5
BAD hit is row 626 and col 5
Lookup table: tower index vs. name