(2001) Data Filtering: UPDATE AMANDA/ICECUBE Collaboration Internal Progress Report Michael Stamatikos† [1/30/03] [† In collaboration with Dr. Kael Hanson & Dr. Paolo Desiati]
Goals for this presentation: Provide paths to (2001) filtered data, (2001) Monte Carlo, & comparative analysis plots. Briefly review the previous comparison [i.e. (2001) data and (2000) Monte Carlo]. Briefly describe the attributes of the new (2001) dcorsica Monte Carlo and (2001) data sample. Present comparative analysis plots of quantities of interests [e.g. nch, ndirc, etc.] for sample data runs. Acquire group feedback and analysis. Develop conclusions and a plan for future analysis.
Paths & Attributes: (2001) Data The (2001) filtered data can be found in path: /data/disk3/amanda/2001/filt-b/upmu The data folders within the directory are: This analysis is concerned with minimum bias (minbias). Hence, only the files with the extension [*.pscl.f2k.gz] were analyzed.
Paths & Attributes: (2001) Data Three runs were selected for analysis: Run (#) Day of Year (2001) (#) of Files (#) of Events 3150 69 173 9752 3270 180 173 8437 3390 285 183 8471 Selection was based on a sufficient number of files (>100) in order to provide the proper statistics and the day of year in order to properly span the (2001) data set.
Paths & Attributes: (2001) Data (MUNT) was used to create [*.hbook] files of the selected data runs. For example: cpfeil data/disk3/amanda/2001/filt/run3150/ab_2001_069_3150_***.pscl.f2k.gz | munt -yu=13 -XZWO minbias_3150.hbook In this manner, the following files were created: File Location (Path) minbias_3150.hbook /data/rellen/ms25/2001data/run3150 minbias_3270.hbook /data/rellen/ms25/2001data/run3270 minbias_3390.hbook /data/rellen/ms25/2001data/run3390
Paths & Attributes: (2001) MC Paolo created (200) (*.hbook) files of (2001) Monte Carlo using dcorsika. These files can be found in path: /data/disk4/simulation/2001/book The first dozen dcorsika files were chained in order to generate enough statistics for comparison (~ 8687 events). Each file has ( ~ 700) events
Paths & Attributes: Plots (PAW) was used to generate new (*.eps) plots using the following triggers: Quantity (2001) Data & MC Plot [*.eps] Number of channels (Linear Scale) nch 3**0nch2001 (Logarithmic Scale) nch 3**0nch2001log Number of direct hits ndirc(7) 3**0ndirc2001 of muon track Zenith of direct walk zenith(1) 3**0zendw2001 Zenith of muon track zenith(7) 3**0zenmu2001 Likelihood of cascade jkrchi(5) 3**0likcas2001 Likelihood of muon jkrchi(7) 3**0likmu2001 Velocity of line fit jkprob(3) 3**0vlf2001 Center of gravity (Z) cogz 3**0cogz2001
Paths & Attributes: Plots For Example: 3150nch2001.eps, 3270ndirc2001.eps, etc. The old plots may be found in the path: /data/rellen/ms25/Comparison_of_2001_Data_and_2000_Monte_Carlo/2001data/run3**0 The new plots may found in the path: /data/rellen/ms25/Comparison_of_2001_Data_and_Monte_Carlo or online at: http://amada.physics.wisc.edu/ms25. The convention used in all plots is that the (2001) data is the solid line, while the (2001) Monte Carlo is the dashed line. The old plots were truncated at (7615) events while the new plots have been truncated to (8437) events.
Old Number of Channels (Run 3150) (2001) Data (2000) MC Monte Carlo exhibited a systematic shift to the right.
The same trend appeared in all plots: It was anticipated that this phenomenon was attributed to the difference in the cleaning procedures between the (2001) and (2000) data. During cleaning, the number of (OM)’s is altered which can cause the above effect since (NCH # Clean Channels). See “Bad OM list for (2001)”, (~ 24) [D. Steele & M. Stamatikos]. General criteria: [average noise rate > 5000 Hz] or [standard deviation > 1000 Hz].
Cleaning: Good OM’s Gone Bad For some reason, run number (3275) was a turning point for some of our (OM’s):
Cleaning: Good OM’s Gone Bad For some reason, run number (3275) was a turning point for some of our (OM’s):
Cleaning: Good OM’s Gone Bad For some reason, run number (3275) was a turning point for some of our (OM’s):
Cleaning: Good OM’s Gone Bad Some (OM’s) are too noisy:
Cleaning: Good OM’s Gone Bad Other (OM’s) fluctuate too much:
New Number of Channels (Run 3150) (2001) Data (2001) MC Nice Agreement
New Number of Channels (Run 3150) (2001) Data (2001) MC (2000) MC Nice Agreement Recall, Monte Carlo exhibited a systematic shift to the right. The systematic shift has vanished in the new plot using (2001) MC.
New Number of Channels (Run 3150) (2001) Data (2001) MC Nice parallel slopes when plotted on a log scale (y-axis)
New Number of Channels (Run 3270) (2001) Data (2001) MC Same trend of good agreement is exhibited in the above plots
New Number of Channels (Run 3390) (2001) Data (2001) MC Same trend of good agreement is exhibited in the above plots
Comparative Analysis of (NCH) Old Old Old New New New
Old Number of Direct Hits of Muon Track (Run 3150) (2001) Data (2000) MC -15 ns < < +75 ns
The same trend appeared in all plots: The good agreement of the plots above demonstrated excellent simulation of minimal scattering within the Monte Carlo simulation (i.e near the OM’s).
New Number of Direct Hits of Muon Track (Run 3150) (2001) Data (2001) MC -15 ns < < +75 ns
New Number of Direct Hits of Muon Track (Run 3270) (2001) Data (2001) MC -15 ns < < +75 ns
New Number of Direct Hits of Muon Track (Run 3390) (2001) Data (2001) MC -15 ns < < +75 ns
Comparative Analysis of (NDIRC) Old New
Old Center of Gravity (Z) (Run 3150) Shift in event centroid was observed. (2001) Data (2000) MC
The same trend appeared in all plots: This has always been a problem and can be symptomatic of many different factors.
New Center of Gravity (Z) (Run 3150) (2001) Data (2001) MC Shift in event centroid is still observed.
New Center of Gravity (Z) (Run 3150) However, it is to a lesser degree than before. (2001) Data (2000) MC (2001) MC
New Center of Gravity (Z) (Run 3270) Shift in event centroid is still observed. (2001) Data (2001) MC
New Center of Gravity (Z) (Run 3390) Shift in event centroid is still observed. (2001) Data (2001) MC
Comparative Analysis of (COGZ) Old New
Old Velocity of Line Fit (Run 3150) (2001) Data (2000) MC Illustrated good agreement.
Velocity of Line Fit Historically the data and Monte Carlo have shown considerable disagreement in this plot. As illustrated in the plot to the right. Taken from Ignacio Jose Taboada Fermin’s thesis, “Search for High Energy Neutrino Induced Cascades with AMANDA-B10 Detector” [Figure (7.2) Page (86)]. Reflects the state of affairs in (1997). Red = Monte Carlo Black = Data
The same trend appeared in all plots: It is speculated that such good agreement was a coincidental phenomenon. Could be due to MAM model – see Paolo’s ice model plots.
New Velocity of Line Fit (Run 3150) (2001) Data (2001) MC Monte Carlo illustrates a slight systematic shift to the right.
New Velocity of Line Fit (Run 3270) (2001) Data (2001) MC Monte Carlo illustrates a slight systematic shift to the right.
New Velocity of Line Fit (Run 3390) (2001) Data (2001) MC Monte Carlo illustrates a slight systematic shift to the right.
Comparative Analysis of (VLF) Old New Old New Old New
Old Likelihood of Cascade (Run 3150) (2001) Data (2000) MC Monte Carlo exhibited systematic shift to the left.
The same trend appeared in all plots: It was suspected that the likelihood is sensitive to noise simulation and cleaning via the filtering process.
New Likelihood of Cascade (Run 3150) (2001) Data (2001) MC Data and Monte Carlo exhibit good agreement.
New Likelihood of Cascade (Run 3270) (2001) Data (2001) MC Data and Monte Carlo exhibit good agreement.
New Likelihood of Cascade (Run 3390) (2001) Data (2001) MC Data and Monte Carlo exhibit good agreement.
Comparative Analysis of (LIKCAS) Old New Old New Old New
Old Likelihood of Muon (Run 3150) (2001) Data (2000) MC
The same trend appeared in all plots: It was suspected that the likelihood is sensitive to noise simulation and cleaning via the filtering process.
New Likelihood of Muon (Run 3150) (2001) Data (2001) MC Data and Monte Carlo exhibit good agreement.
New Likelihood of Muon (Run 3270) (2001) Data (2001) MC Data and Monte Carlo exhibit good agreement.
New Likelihood of Muon (Run 3390) (2001) Data (2001) MC Data and Monte Carlo exhibit good agreement.
Comparative Analysis of (LIKMU) Old New Old New Old New
Old Zenith of Direct Walk (Run 3150) (2001) Data (2000) MC Data and Monte Carlo exhibited good agreement.
The same trend appeared in all plots: Attenuated features are due to preference given to zenith () and nadir (). Shows good agreement as a first guess method.
New Zenith of Direct Walk (Run 3150) (2001) Data (2001) MC Data and Monte Carlo still exhibit good agreement.
New Zenith of Direct Walk (Run 3270) (2001) Data (2001) MC Data and Monte Carlo still exhibit good agreement.
New Zenith of Direct Walk (Run 3390) (2001) Data (2001) MC Data and Monte Carlo still exhibit good agreement.
Comparative Analysis of (ZENDW) Old Old Old New New New
Old Zenith of Muon Track (Run 3150) (2001) Data (2000) MC Data and Monte Carlo exhibited good agreement.
The same trend appeared in all plots: Better quality of fit. Reveals angular distribution of the muons. Flux peaks ~ (30°) (80 °) < Mostly background (atmospheric muons)
New Zenith of Muon Track (Run 3150) (2001) Data (2001) MC Data and Monte Carlo still exhibit good agreement.
New Zenith of Muon Track (Run 3270) (2001) Data (2001) MC Data and Monte Carlo still exhibit good agreement.
New Zenith of Muon Track (Run 3390) (2001) Data (2001) MC Data and Monte Carlo still exhibit good agreement.
Comparative Analysis of (ZENMU) Old New
Conclusions Overall the agreement between the (2001) filtered data and (2001) Monte Carlo is very good, much better that the agreement between the (2001) filtered data and (2000) Monte Carlo. There is no longer a systematic shift of the Monte Carlo to the right in the (NCH) plot. Even though the Monte Carlo still systematically shifts to the right in the (COGZ) plot, it does so to a lesser degree. The Monte Carlo appears to be slightly systematically shifting to the right in the velocity of line fit plot. Previous agreement was attributed to coincidence. The likelihood plots of both cascade and muon now show good agreement between (2001) data and Monte Carlo. It is believed that these effects may be explained via the difference in the cleaning procedures between the (2000) and (2001) Monte Carlo simulations.