ESAF Euso Simulation and Analysis Framework Simulation, Reconstruction, Visulation, Analysis Simulates process from shower generation to light production, transport, detector response, reconstruction and analysis A package of C++, C and F77 programs organized as shared object Libraries, based on ROOT using its libraries with high modularity Air showers: CORSIKA, CONEX, built-in Slast shoower generator

Simulácia v ESAF

Rekonštrukcia v ESAF

Fake trigger background simulations JEM-EUSO detector is basically a large field of view UV camera, pointing toward the earth atmosphere, to detect and measure fluorescence light imprint produced by development at speed of light of Extensive Air Showers. For a 1020eV EAS, a few thousands photons are expected on detector focal surface (FS). However, the background photons are much more than those of signal. Therefore background reduction is essential for such observatory EECR. It is aim of the trigger to try to extract the signal from the background sea. Electronics will have to reject as much counts as possible without rejecting the signal itself. Fortunately signal has some peculiar characteristics that can be used to distinguish it. The shower generate a spot moving on focal surface. On the other hand, the background is distributed randomly. But it is necessary to assess, if the random processes do not produce fake patterns, which could be mistakenly interpreted as EECR events. For this purpose huge amount of measurements with only background events have to be simulated. Obtained results would be consequently analysed by several pattern recognition algorithms to verify the probability of registration a fake trigger events in several trigger conditions.

JEM-EUSO FOCAL SURFACE LAYOUT Typical air-shower seen by JEM-EUSO FS is covered by large numbers of photo-detector tubes structured in series of similar pieces, the one embedded in the others. Largest piece is PDM -> photodetector module. The whole FS is made of 137 such PDM’s. PDM structure is itself squared matrix of 3x3 smaller elements called elementary cells EC. Each EC is a squared matrix of 2 × 2 multianode photomultipliers. An EC is a 12 × 12 pixel matrix, corresponding to 144 channels. PDM is a 36 × 36 pixel matrix corresponding to 1296 channels. Each PDM probe a squared pad of 27km x 27km, enough large to contain a substantial part of the imaged trace under study JEM-EUSO

Outline of noise reduction capability. Úlohou triggra je detekovať výskyt signálu od reálneho eventu spomedzi extrémne vysokého pozadia ( ~ 1011 cts/s/FS) Signál postupne filtrovaný na viacerých úrovniach redukujúc trigger rate Schéma triggra zodpovedá segmentovaniu FS (Focal Surface) na moduly PDM (Photo Detector Module), ktoré sú dostatočne rozsiahle na to, aby obsiahli podstatnú časť zobrazenia skúmaného 'treku' Outline of noise reduction capability. Level Rate of signals/triggers at PDM level Rate of signals/triggers at FS level PDM level trigger Photon trigger ~9.2 × 108 Hz ~1.4 × 1011 Hz Counting trigger ~7.1 × 105 Hz ~1.1 × 108 Hz Persistency track trigger (PTT) ~7 Hz ~103 Hz PDM cluster level trigger (FS=144 PDM's) Linear track trigger (LTT) ~6.7 × 10-4 Hz ~0.1 Hz Expected rate of cosmic ray events ~6.7 × 10-6 Hz ~10-3 Hz

1st level PDM trigger 2nd level CCB trigger

FIRST LEVEL TRIGGER The 1st trigger level mainly operates to remove most of the background fluctuations by requiring a locally persistent signal above over a few GTU’s duration. GTU is gate time unit of the value 2.5μs -> temporal time resolution of detector electronics. In 1st level trigger -> PTT (Persistency Track Trigger) are pixels grouped in boxes of 3 × 3. Trigger is issued if for 5 consecutive GTU’s there is at least one pixel in box with an activity higher than a preset threshold and the total number of detected photoelectrons in the box is higher than a preset value. These two values are set as a function of the average noise level in order to keep the rate of triggers on fake events at a few Hz per PDM.

SECOND LEVEL TRIGGER Role of 2nd trigger level - Linear Track Trigger (LTT) is to find some track segments in 3dim from the list of pixels provided by the first level, for each GTU time bin. The track speed has to be compatible with a point travelling at speed of light in whatever direction it propagates. So it follows the movement of the EAS spot inside the PDM over some predefined time, to distinguish this unique pattern of an EAS from the background. From a PTT trigger, the PDM electronics will send a starting point, which contains the pixel coordinates and the GTU which generated the trigger. The LTT algorithm will then define a small box around it, move the box from GTU to GTU and integrate the photon counting values. When excess of integrated value above the background exceeds the threshold, an LTT trigger will be issued. It is foreseen to have a total of 67 starting points for the integration, which are distributed equally over time and position around this box. Each integration will be performed over ±7 GTU’s for a predefined set of directions. The background-dependent threshold on the total number of counts inside the track is defined to reduce the level of fake events to a rate of 0.1 Hz per FS. These two trigger levels combined together reduce therefore the rate of signals on the level of 109 at PDM level.

Motivácia a simulačný kód Veľmi vysoká úroveň očakávaného pozadia vyžaduje simuláciu obrovského počtu prípadov, aby sa dosiahla spoľahlivá filtrácia bg prípadov Vyvinutý a použitý rýchly C++ kód: - rozdelenie počtu fotoelektrónov pozadia sa riadi Poissonovým rozdelením so strednou hodnotou 500 ph/(m2 s sr) = 2.1 ph/pix/GTU - PDM = 6x6 pixlov -> M36 konfiguracia; resp 8x8 pixlov -> M64 konfiguracia - PTT algorithm 1 Hz/PDM - LTT algorithm 1mHz/PDM Kód je rýchly, ale potrebné vyprodukovať obrovskú štatistiku -> nutné paralelné počítanie Všetko počítané na našom (OKF) PC klastri v Košiciach Minimálnu potrebnú úroveň štatistiky možno dosiahnuť pre každú jednu z dvoch konfigurácií za rok nepretržitého počítania Ukladané dáta o prípadov, ktoré prešli úrovňami PTT aj LTT filtrácie Tieto analyzované pattern recognition

JEM-EUSO cluster 16 node Supermicro® SuperServer AS-1042G-MTF Configuration of node : - 4x Opteron 6134 (2,3GHz) - 16GB RAM - 600GB SATAII HDD (WD VelociRaptor) 2x master/disk server Supermicro® SuperServer AS- 1042G-MTF Configuration od server: - 1x Opteron 6134 (2,3GHz) - 4x 2TB SATAII HDD (WD RE4) All together: CPU: 64 + 2 @ 2.3GHz Cores: 512 + 16 RAM: 264 GB (4GB / CPU) Disk space: 16x600GB + 8x2TB = 25,2TB

Next step Fake trigger identification – pattern recognition methods to disentangle signal from background: Hough transformation Clustering Results from simulated UV background – optimize triggering and reconstruction algorithm in ESAF (EUSO simulation and analysis framework)