In this blog, I...

-Will discuss yesterday's presentation, and the error I made in the calculation of the constants.

-Will answer a few questions I received on the list and by email. Including questions relating to the phi dependence

 

What's New Since the MIT Meeting

I made a few changes to the method of extracting the gains since the analysis meeting. The major changes that were made were;

1) The threshold on the number of entries in the histogram was increased to 1000, and changed from number of entries; to entries in the fit region.

2) A few of the strips that did not pass this threshold were erroneously fit with a function; this was due to the fact that parameters were set in the wrong place in the code. Notice the strips in the plot of extracted gains vs. pseudo-rapidity (from the analysis meeting) below, with gain of 0.03;

Plot 1

now, the corrected plot;

 

Plot 2

 

In yesterday's presentation, I showed a profile histogram of these gains;

 

Plot 3

 

However as Will, and a few others, pointed out; there is a mistake in the distribution of the gains in the plots after calibration. The shapes of the gains from -1 to 0 in rapidity appear to be flipped left-to-right;

 

Plot 4

 

As I feared, the problem lies in the ideal gain values chosen before I ran the calibration analysis; it must therefore be repeated. Specifically, the ideal gain distribution, which should be identical to Plot 3 (without the error bars); instead looked liked this;

 

Plot 5

 

I attribute this error, to the fact that I had two versions of this code floating around. The first version was used in the past, and averaged the positive and negative rapidity values; to generate 150 ideal gain values. The second version, was used when I expanded the ideal values to 300; in the full pseudo-rapidity range. Because of how I decided to treat the Phi strips (see below), I chose to edit the old code and run that (instead of editing the newer version). I forgot that Eta strips run from 0 to -1 and not from -1 to 0. This explains the change in shape of the distribution. I also neglected to perform the double-checks I usually do, to avoid such an error. Here are the correct ideal values;

 

Plot 6

 

I will re-launch the calibration now, and update again when I have the final-"final constants"... finally. Here are the distributions of constants for Eta and Phi with Strip Number (Soft id).

 

Plot 7 - Eta Constants

 

Plot 8 - Phi Constants

 

Changes to the Phi Strip Treatment

The pseudo-rapidity dependence of the phi gains was less than convincing (within the errors)- I therefore decided to go with a constant ideal value for the Phi strips (0.03).

Phi Dependence (Recap)

The phi dependence of the eta strips, are even less than convincing (within the errors) than the eta dependence of the phi strips;

Plot 9

The situation with the phi strips, is as expected;

Plot 10

I have therefore excluded this from consideration; in the calculation of my constants.

 

After Bug-Fix

I fixed the problem and re-ran the calibration test. The new "after calibration" plot follows;

Plot 11

This looks as expected. The calibration constants are ready for inclusion into the database. They are located here;

/star/u/martinc/FinalCalibrationConstants2007/

Again, there are two sets of constants.

1. Problem Strips
Problem strips are defined as those strips, in which the gain moved further away from the ideal, after calibration.

Total Problem Strips (Eta): 1,862  (10.3%)
Total Problem Strips (Phi): 1,253 (7.0%)

2. "Restrictive" Calibrated Strips
"Restrictive" Calibrated Strips (Recommended Constants), are those strips in which the problem strips, have a calibration constant of 0.

"Restrictive" Calibrated Strips (Eta): 12,008  (66.7%) EtaCalibrationConstants_Restrictive.txt
"Restrictive" Calibrated Strips (Phi): 13,666 (75.9%) PhiCalibrationConstants_Restrictive.txt

3. "Non-Restrictive" Calibrated Strips
"Non-Restrictive" Calibrated Strips, are those strips in which the problem strips, have a calibration constant of 1.

"Non-Restrictive" Calibrated Strips (Eta): 13,870  (77.1%) EtaCalibrationConstants_Non-Restrictive.txt
"Non-Restrictive" Calibrated Strips (Phi): 14,919 (82.9%) PhiCalibrationConstants_Non-Restrictive.txt

 Absolute Calibration Values (From previous db values - at mid-rapidity)

Eta Strips: 0 0.0080087231472135

Phi Strips: 0 0.0080087231472135

The average of our values at eta = 0, will be set to the old average at eta = 0. The shapes will differ however; and there will be 36,000 different constants.

 

*Blog will be updated tomorrow*

 

Martin Codrington