1. Title (10 sec) 2. Talk outline (10 sec) 3. Structure of downgoing simulation (self-explanatory) (40 sec) 1st click: alternative generators/propagators (20 sec) 2nd click: AMANDA-II simulation defaults (10 sec) 3rd click: New developments for CORSIKA and MMC (20 sec) total: (90 sec) 4. CORSIKA simulation settings (10 sec), advance the slide while explaining 1st click: 6 high-energy interaction models for the simulation of first interaction of cosmic rays (60 sec) features, default for AMANDA-II - QGSJET; VENUS and NEXUS produce similar results (from my talk during AMANDA meeting), but are much slower. All work for in the interesting energy range for downgoing flux estimation. HDPM is based on the phenomenological fits to accelerator data. Execution times are given for default AMANDA-II simulation configuration. 2nd click: Next bullet: curved treatment of the Earth's surface as well as the top of the atmosphere. 90 degrees in the detector coordinates are mapped into 88.67 degrees on the Earth's surface. This removes some artifacts of the original CORSIKA program at the horizon. Also two angles in the detector frame are mapped into the same angle on the surface. So the same compact CORSIKA file (after CORSIKA step but before the UCR shower core randomization) can be used to produce upgoing muons or neutrinos as well as downgoing. A small number of upgoing muons (zenith angles 90-95 degrees) is actually produced in the AMANDA-II simulation to make sure that downgoing muons possible produced by upgoing showers are accounted for. (120 sec) 3rd click: Next bullet: detector configuration is assumed cylindrical and will likely stay cylindrical for the IceCube simulation. Important to note that the ratio l/d is fixed after the CORSIKA run and the same ratio must be supplied during shower core randomization by UCR (30 sec) total: (220 sec) 5. CORSIKA run consists of CORSIKA run and UCR (one or more) runs. I have forgotten what UCR stood for (something like UC Reader), so the explanation is invented for this slide. (30 sec) 1st click: CORSIKA run is controlled by the card file shown here (10 sec) 2nd click: The zenith-angle dependent cut is used inside the CORSIKA program to make sure time is not spent on the primaries which are not likely to produce muons which reach the depth of the detector. The plot explains some of this cut (simplified): To produce a 1 TeV muon, energy of the primary must be at least 1.58 times the energy of the muon in 99.9% of all cases. For 99% this would be 2.06. These values will be explained in my thesis, and are currently available in the "dcorsika update report" available on my web site and on the ACID page (which is working again now) (60 sec) 3rd click: This should explain the argument that went into the 99% or 99.9% ratios, also determines muon energy cuts in the circled line of the card file. There is always a chance, however small, that a given energy muon will propagate to a given distance. Therefore one needs to decide on the acceptable ratio. 99.9% ratios are used for muon and primary energy cuts in the AMANDA-II simulation (40 sec) 4th click: Finally, full AMANDA-II simulation (with mam ice model, std kurt was also run; results for it are on my web page) and the minimum energy of the primary turns out to be higher than 273*1.58; less than 10^-6 fraction of the primaries producing events in AMANDA-II will have energies less than 600 GeV (40 sec) 5th click: output of "ucr -h" is shown. Here it is important to note that the values for the detector length and radius are chosen also based on some acceptable simulated signal loss (10^-6, default or possibly 10^-5). The plot the ratio of event-causing showers with cores outside the cylinder with given dimensions. This and next plots are symmetrical around r=0, and behavior of l may be counter-intuitive: for positive l, bottom of the detector is fixed at -400 (almost infinity), and ` for negative l the top is fixes at 400. The value of l is measured from the center of the detector (40 sec) 6th click: This plot shows the same thing as the previous one. The blue contours (and numbers on the right side) give the fraction values of unaccounted showers, and the green contours and numbers give "areasum", which is the integrated area times solid angle for the showers coming from the upper hemisphere. The values for in green are in 10^4 m^2 sr. The goal here is to minimize areasum (thus increasing simulated lifetime for the same number of showers) while keeping the fraction of unaccounted showers as small as possible (120 sec) total: (340 sec) 6. Now switching to MMC. Skip the features if no time or not necessary. However, to clear up some ignorance among IceCube (and even AMANDA) people, it may be useful to read them out. (120 sec) 1st click: output of "Amanda -h" (mmc also has "Frejus" and "Test" programs which can be used for different tests, cross section plots etc.) (20 sec) 2nd click: AMANDA geometry is implemented. All space (ice, ice and rock, or ice, rock and air, depends on the multi-media settings used) is divided into 3 parts: before the detector (fast propagation, no secondaries are recorded), inside the detector (all secondaries with energies above 500 MeV are recorded), and after the detector (very fast, propagation is done in one step to determine the average muon track length in the 3rd region) (60 sec) 3rd click: Settings for AMANDA are shown, they were just mentioned. Also there is a new option (not implemented in any other muon propagators) called "cont", which stands for "continuous randomization". (20 sec) 4th click: It randomizes the part of the muon spectrum which is otherwise treated continuously. There is always some part of energy losses treated continuously. 3 out of 4 cross sections (if not all) get very large at small energy losses, and if it were not for kinematical constraints, they would make the number of produced secondaries infinite. To speed up the calculation one has to optimize on the fraction of the energy loss below which produced secondaries can be treated continuously. "continuous randomization" feature allows one to choose this ratio at a much higher value, thus considerably speeding up the calculation. As implemented, it was shown to produce correct results for muon propagation over longer distances and v_cut from 0.1 to 0.001. Choosing v_cut outside this range when "cont" is enabled will make results inaccurate as shown in the MMC manual (see my web page or ACID page). Only used in the first propagation region for AMANDA. Spend some time explaining the plots, too. It should be obvious which v_cuts correspond to which curves. The smaller the v_cut, the more accurate the spectrum is. It converges to some final shape as v_cut is decreased. Enabling the "cont" option makes the curve shape converge faster. The single bins outstanding on the plot to the left consist of muons which suffered exactly zero interactions with energy loss fraction below v_cut. The bin is bigger and located at lower energy for higher v_cuts (obviously). (150 sec) total: (370 sec) 7. Parametrized functions are calculated at the run time (a couple of minutes) and stored in files (to speed up next calculations). Here you see different setting can be used for interpolation ("-g=" option) and parametrization ("-hifi" option). g stands for "romberG number", in this case means the number of points used for interpolation. The plot shows the difference between interpolated values and precise calculation (every cross section can be calculated precisely or interpolated, can be configured during compilation or at the run time, when the "-mediadef=" option is used for AMANDA, or "-intr" for "Frejus" or "Test" programs). Media definitions file allows to specify media, which detector lies outside of. This disables region 2 parameterizations for that medium, thus speeding up the initialization and reducing the run size. (60 sec) 1st click: show to demonstrate the last bullet: the particle gets "stuck" at high energy with low v_cut (500 MeV/10^15 GeV) because interaction length is too small to be added to the distance variable. Also shows MMC's region of applicability is much higher than that of other programs. Just in case this is asked, higher (up to 10^30 eV) energy of the muon is supported, but not with E_cut=500 MeV. Raise the e_cut or use v_cut instead. (40 sec) total: (100 sec) 8. Conclusions and discussion (60 sec) Total: 1600 sec.