Документ взят из кэша поисковой машины. Адрес оригинального документа : http://antares.sinp.msu.ru/docs/Gracheva_Bamberg_22.09.11.pdf Дата изменения: Mon Dec 16 17:03:27 2013 Дата индексирования: Thu Feb 27 20:27:14 2014 Кодировка:

Data Selection For The Anisotropy Analysis
Gracheva Katya SINP-MSU

22.09.2011

Bamberg, Germany


Short reminder of previous analysis

The average number of muons N per minute for each run depends on number of active OMs (Npmts) during this run as ln N A+BNpmts. For different detector configuration and different triggers used during the run, parameters A and B vary it's necessary to split data on several general groups and process them separately. Splitting made by number of active lines and type of triggers (including or not GC, TQ, T2, SCAN&PRELIMS and reduced gains).



"Distance": the quality cut for run selection

"distance" difference between expected and real number of muons per minute with given configuration

In order to calculate optimal value of distance one should summarize in one axis all distance-histograms from each group, make a Gaussian approximation of total histogram and calculate the 3 value.


Further processing
Only 3N events of «good» runs of previous analysis are selected: 3N triggered events exist in every run Better accordance with MC for rates <150kHz (see elog 548 by Colas) GC triggered events excluded AaFit: > -6.5 (Vladimir's presentation explains reason) Zenith angle < 20° (in order to look only at events with vertical tracks)

After 2008 2009 2010

all fi year year year

ltres: -- 0.83107 -- 0.77107 -- 1.04107

-events

3N trigger > -6.5 Plot for one run only

Total: 2.6107 -events (comparing to 3108 for all triggers before filtering)


Spectrum analysis, Lomb method


Lomb method was elaborated for cases of unevenly sampled data Weights the data on «per point» basis instead of «per time interval»

NUMERICAL RECIPES IN C: THE ART OF SCIENTIFIC COMPUTING (ISBN 0-521-43108-5, p. 575)


(zenith < 20°, > -6.5, GC excluded, artificial added anisotropy)
I II

Example of the FastLomb output

time (sec)

Inverse frequency (sec)

III

IV

I -- input muon rate II -- null hypothesis probability vs inverse frequency III -- spectral power vs frequency IV -- null hypothesis probability vs frequency


Comparison of the Lomb analysis perfomed on the -rate with and without GC trigger after angle and lambda cuts
Lomb analysis of the -rate for the 2009 data with GC: Sharp peak in the null probability function with frequency 11.610-6 Hz corresponding to one sidereal day (23 h 56 min 4 sec)

Lomb analysis of the -rate for the 2009 data without GC: As soon as GC evens are filtred no peak is seen. No evidence of possible anisotropy either. But can it be observed this way?


Sensitivity of the Lomb method on possible muon anisotropy
I

I -- artificially added anisotropy 4%, peak is well-seen II -- artificiallly added anisotropy 2%, peak still seen III -- artificially added anisotropy 1%, no peak can be seen

Expected level of natural anisotropy is ~ 0.1% The method is helpless
II III


Conlusions






3 years of data are analysed and put into DB (2008-2010) Events with 3N trigger only are filtred Lambda cut ( > -6.5) for vertical downgoing muons is applied Lomb method is used to perform spectrum analysis for -rate during 2008-2010 years Analysis has shown it's impossible to see anisotropy using Lomb method (level of possible anisotropy is too low)


What's going on when the gain of PMTs changes? (Colas' question from Moscow)
3N+2T3 trigger, normal gain Black -- normal gain runs were before Red -- halfgain runs were before Blue -- gain/4 runs were before Green -- gain/8 runs were before

Colored runs in this plot possibly coud have another number of detected muons if calibration sets were not downloaded in time from the DB. But, apperantly, this effect is not seen in this analysis (even runs after switching between gain/8 or gain/4 and normal are lying in main group).