SixTrack PROJECT: STATUS, RUNTIME ENVIRONMENT, AND NEW DEVELOPMENTS∗

SixTrack is a single-particle tracking code for high-energy circular accelerators routinely used at CERN for the Large Hadron Collider (LHC), its luminosity upgrade (HL-LHC), the Future Circular Collider (FCC), and the Super Proton Synchrotron (SPS) simulations. The code is based on a 6D symplectic tracking engine, which is optimised for long-term tracking simulations and delivers fully reproducible results on several platforms. It also includes multiple scattering engines for beam-matter interaction studies, as well as facilities to run integrated simulations with FLUKA and GEANT4. These features differentiate SixTrack from general-purpose, optics-design software like MAD-X. The code recently underwent a major restructuring to merge advanced features into a single branch, such as multiple ion species, interface with external codes, and high-performance input/output (XRootD, HDF5). This restructuring also removed a large number of build flags, instead enabling/disabling the functionality at run-time. In the process, the code was moved from Fortran 77 to Fortran 2018 standard, also allowing and achieving a better modularization. Physics models (beam-beam effects, RF-multipoles, current carrying wires, solenoid, and electron lenses) and methods (symplecticity check) have also been reviewed and refined to offer more accurate results. The SixDesk runtime environment allows the user to manage the large batches of simulations required for accurate predictions of the dynamic aperture. SixDesk supports CERN LSF and HTCondor batch systems, as well as the BOINC infrastructure in the framework of the LHC@Home volunteering computing project. SixTrackLib is a new library aimed at providing a portable and flexible tracking engine for singleand multi-particle problems using the models and formalism of SixTrack. The tracking routines are implemented in a parametrized C code that is specialised to run vectorized in CPUs and GPUs, by using SIMD intrinsics, OpenCL 1.2, and CUDA technologies. This contribution presents the status of the code and an outlook on future developments of SixTrack, SixDesk, and SixTrackLib. ∗ Research supported by the HL-LHC project † riccardo.de.maria@cern.ch ‡ Work supported by Google Summer of Code 2018 INTRODUCTION SixTrack [1,2] is a 6D single-particle symplectic tracking code able to compute the trajectories of individual relativistic charged particles in circular accelerators for studying dynamic aperture (DA) or evaluating the performance of beam-intercepting devices like collimators [3]. It can compute linear and non-linear optics functions, time-dependent effects, and extract indicators of chaos from tracking data. SixTrack implements scattering routines and aperture calculations to compute “loss maps”, i.e., leakage from collimators as a function of longitudinal position along the ring, and collimation efficiency [4]. Different from a general-purpose code like MAD-X [5,6], SixTrack is optimised for speed and numerical reproducibility. It can be also linked with the BOINC library to use the volunteering computing project LHC@Home [7]. SixTrack studies, such as estimation of dynamic aperture of large storage rings like the Large Hadron Collider (LHC) or the Future Circular Collider (FCC), require massive computing resources, since they consist of scans over large parameter spaces for probing non-linear beam dynamics over long periods. The SixDesk runtime environment manages SixTrack simulations from input generation, job queue management (using HTCondor or LSF in the CERN BATCH service and customised software in CERN Boinc server), to collecting and post-processing results. SixTrackLib is a new library built from scratch in C with the main aim of offering a portable tracking engine for other codes and offloading SixTrack simulation to GPUs. This paper summarises the main existing features of SixTrack, SixDesk and SixTrackLib and provide detail about the main development lines.