Faculty of Physics, University of Munich (Germany)
Local Project ID:
HPC Platform used:
Researchers of the Faculty of Physics at the Ludwig-Maximilians-Universität München leveraged HPC system SuperMUC to a) investigate the proton driven laser-wakefield acceleration of electrons, and b) simulate the interaction of ultra-intense laser pulses with ultra-thin foils. The applied highly sophisticated Particle-In-Cell computer simulations were meant to contribute to new insights into laser driven proton acceleration. Further findings in this area could help to significantly cut costs for cancer therapy centers.
Due to the AWAKE project at CERN there is a great interest in proton driven wakefield generation, as they are considered to be a promising way of accelerating electrons to the TeV scale. This is investigated using carefully planned Particle-In-Cell (PIC) simulations at the cutting edge of high performance computing technology. At the same time the ELI-NP project is gaining momentum. ELI-NP promises to become an ultra-high field laser facility where novel high-field experiments can be carried out. Also in this project the interaction of short and ultra-intense laser pulses with ultra-thin foils and nano targets in cooperation with J. Schreiber and L. Veisz at the Max Planck Institute of Quantum Optics is studied. The resulting generation of fast particles in the MeV up to GeV range as well as strong electro-magnetic radiation can be employed in fundamental science and medical applications.
In both cases a detailed understanding of the expected physics, backed up by in-silico experiments, is needed and can strongly increase the return on the investment. Further results in laser driven proton acceleration can potentially help to cut costs in the order of €100M for cancer therapy centers by a factor of ten
Results and Methods.
The Plasma-Simulation-Code (PSC) is a general purpose framework to solve the extended Maxwell-Vlasov-Boltzmann system of equations via the PIC approach . Recent extensions comprise the self-field effects of radiation and electron-positron pair production in strong fields. The original FORTRAN version evolved to a modern modularized C simulation framework supporting bindings to FORTRAN as well as C/CUDA and features selectable field and particle pushers. The PIC approach is well-known for its good scaling capability via configuration space parallelization. A Hilbert-Peano space filling curve is used for efficient, dynamic and adaptive load and memory balancing allowing for complex and dynamic geometries.
In the AWAKE project the interaction of a 450 GeV proton beam of the SPS pre-accelerator at CERN with a 10 m long plasma is studied (Fig. 1). The required resolution is on the µm scale and therefore 735 billion grid cells are necessary (Fig. 3). Moving window technology allows for reducing the active memory footprint and the costs to about 3% of the total simulation, but it still takes several weeks on a large fraction of SuperMUC. A minimum of 10 full output steps produce around 300 TB. The researchers were able to demonstrate via a full 3D kinetic simulation that a maximum acceleration field of 700 MV is generated (Fig. 2), in accordance with 2D cylindrically symmetric codes , and therefore showing that PIC represents the physics correctly while opening research possibilities to important effects like beam filamentation.
Ultra-thin foils and nano targets
This project part aims for new particle as well as light sources. To produce atto-second X-ray pulses (AXP) via high harmonic generation (Fig. 3) of few cycle laser pulses enormous resolutions are necessary. E.g. the main simulation requires 2 nm resolution for a (16 µm)3box. Therefore 512 billion simultaneous grid cells (half a trillion) are necessary, that need up to 120 TB of main memory. The simulation took 30 hours on 16 islands (131,072 cores) of SuperMUC phase 1, i.e. four million core hours. Despite heavy inline data reduction still 100 TB of output data were generated and the frequent checkpoints of 30 TB were written with 105GB/s average I/O-throughput.
For Ion acceleration low emittance and high conversion efficiency from laser energy to fast ions are desirable. As the focus is not to resolve the electro-magnetic high harmonics, lower resolutions and therefore 105 core hours per shot are sufficient to cover the relevant physics.
The removal of electrons in the form of fs-bunches was studied as well as the subsequent Coulomb explosion, producing fast ions (up to 90 MeV for protons, 120 MeV for carbon, Fig. 4) with a very directed angular distribution. However, to develop better experimental designs a larger series of runs was carried out. Each run used 7168 cores for 12 hours and 5-10 TB disk space.
It is expected that the researchers' methods used in plasma physics (many body interaction with radiation) will be more and more adapted for medical diagnostics and treatments. For this research field we expect centimeter sized volumes with necessary resolutions of tens of micro meters resulting in boxes of >1012 voxels (100-200 TB) on a regular basis. In consequence the demand for computing time and especially for data storage and data handling capacities will also increase significantly.
Karl-Ulrich Bamberg, Patrick Böhl, Fabian Deutschmann, Constantin Klier, Nils Moschüring, Viktoria Pauw, Hartmut Ruhl (PI).
AWAKE Collaboration (CERN), Max Planck Institute of Quantum Optics (MPQ), ELI-NP
References and Links:
 Pauw, Ostermayr, Bamberg et al 2016, NIMA Section-A (10.1016/j.nima.2016.02.012)
 Caldwell & Lotov 2011, Phys. Plasmas 18, 103101
Computational & Plasma Physics
Theresienstr. 37, D-80333 München (Germany)
e-mail: Karl-Ulrich.Bamberg [at] physik.uni-muenchen.de