The ATLAS Simulation Infrastructure
- Aad, G. et al
- arXiv:1005.4568
 | noimg: Numbers of files, lines of code, and lines of comments in the ATLAS simulation project, by programming language for major contributors. External dependencies are not included.: Numbers of lines of code in each of the projects upstream of the ATLAS simulation project, versus the programming language. Most projects are dominated by C++ and \Python\ code. The most significant exception is the detector project, which contains 70,000 lines of XML and Java code. |
 | The flow of the ATLAS simulation software, from event generators (top left) through reconstruction (top right). Algorithms are placed in square-cornered boxes and persistent data objects are placed in rounded boxes. The optional pile-up portion of the chain, used only when events are overlaid, is dashed. Generators are used to produce data in HepMC format. Monte Carlo truth is saved in addition to energy depositions in the detector (hits). This truth is merged into Simulated Data Objects (SDOs) during the digitization. Also, during the digitization stage, Read Out Driver (ROD) electronics are simulated. |
 | ATLAS detector view. |
 | Commissioning layout of the detector used for cosmic ray data taking during 2008. The endcap toroidal magnets and beampipe are not yet installed. The ca\-lo\-ri\-me\-ter endcaps (purple) are shifted by 3.1~m and the muon endcaps (green) are shifted to provide access to the inner detector during installation. The barrel toroid magnets are shown in yellow, and the inner detector is shown in blue. |
 | Combined test beam setup from 2004. |
 | CPU time per event increases with varying cuts on the $|\eta|$ of primary particles. The time is normalized to CPU time for simulation of all primaries inside of $|\eta|<3.0$ and increases between three-fold and eight-fold for simulation of all primaries in $|\eta|<8.0$. The average of 200 simulated $\ttbar$\ and minimum bias events was taken. Linear fits are overlaid. |
 | An event display made with VP1. A Higgs boson decays into four muons (shown in red). Inner detector tracks are in green, and energy deposited in the ca\-lo\-ri\-me\-ter by the muons is shown in yellow. |
 | A Higgs boson decaying into four muons, with only the inner detector tracks and hits in the TRT being displayed by VP1. |
 | Distributions of CPU time for 250~$\ttbar$\ events in full, Fast G4, and \Atlfast-II simulations. Vertical dotted lines denote the averages of the distributions. |
 | width=3.2in |
 | Left, fast simulations (color) and full simulation (black) comparison of missing transverse energy along the $x$-axis in di-jet events with a leading parton $p_T$ between 560~and 1120~\GeV{}. Right, a comparison of jet $p_T$ resolution as a function of pseu\-do\-rapidity in $\ttbar$\ events for jets with $20<p_T^{True}<40$~GeV.Caption not extracted |
 | width=3.2in |
 | Fast simulations (color) and full simulation (black) comparison of reconstruct\-ed muon $p_T$ resolution as a function of muon $p_T$ for central ($|\eta|<1.2$) muons in $\Zmumu$\ events for muons reconstruct\-ed using only the muon spectrometer (left) and using both the inner detector and spectrometer (right). \Atlfast-I only provides one type of muon, which is included in the right plot.Caption not extracted |