Radiation imprint of ultra- intense laser heating of solids

PowerPoint-PräsentationMarco Garten I HZDR – Research Group Computer Assisted Radiation Physics I www.hzdr.de/FWK
M. Garten1,2, A. Huebl1,2, R. Widera1, I. Goethel1,2, L. Obst-Huebl1,2, T. Ziegler1,2, K. Zeil1, T. Cowan1,2, U. Schramm1,2, M. Bussmann1, T. Kluge1
1 Helmholtz-Zentrum Dresden – Rossendorf, Germany 2 Technische Universität Dresden, Germany We acknowledge PRACE for awarding us access to Piz Daint at CSCS, Switzerland.
15th PRACE call, Project ID 2016163983.
Radiation imprint of ultra-intense laser heating of solids
Marco Garten I HZDR – Research Group Computer Assisted Radiation Physics I www.hzdr.de/CRP Page 2 EuroHPC Summit Week 2019 / PRACEdays19
Conventional ion acceleration at extreme scales
CERN Large Hadron Collider
Particle energy: 6.5 TeV
Conventional ion accelerators for medical application
Heidelberg Ion-Beam Therapy Center HIT Construction cost: 120 million € Particle energy: 50-430 MeV/nucleon
Alternative: Compact, laser-driven ion sources?
field strength: TV / m instead of ~ 10 MV / m
energy gain: MeV / µm
long accelerating structures
Concrete radiation shielding
Nobel prize in Physics 2018
Laser-driven proton sources for applications
S. Göde et al.: Phys. Rev. Lett. (2017), L. Obst-Huebl et al.: Nature Comm. (2018)
Beams of high dose (~ 1 Gy) + ultra-short (~ 1 ps) duration
desirable for
High-dose radiobiology
Translational research
in radiooncology
Proton radiography
Materials research
Laser-driven proton sources for applications
Target-normal sheath acceleration (TNSA) of Ions
C , H
, O c
on ta
m in
an t
ps – time scale forming of Debye sheath
Currents of 106-109 A
~ nm-µm
[2]
[1] Kluge et al. PRL 107, 205003 (2011), [2] Zeil et al, Appl. Phys. B 110.4 (2013) 437-444
2D
3D
What limits the performance?
Laser-driven ion acceleration - status
best performance ≠ usual key could be temporal pulse shape of the last picosecond
“cleaning” of temporal contrast with plasma mirror techniques can shoot thinner targets
Proton acceleration – stability in experiment
Measured laser contrast underlies limits in time resolution and dynamic range Shot-to-shot fluctuation
[1] L. Obst PPCF (2018)
Extract characteristic features
Understand interaction systematically
[1]
Simulation of plasma dynamics with particle-in-cell codes
+
-
-
+-
-
+
+
Repeated cycles allow simulating longer time durations.
Local operations only: well suited for parallelization.
Particle-in-cell algorithm:
perfect contrast perfectly sharp target rear surface best accelerating shield field
optimal thickness does not shift much with laser contrast
Would 3D simulations show the same trend?
2D Simulation pre-study at HZDR
15th PRACE call, Project ID 2016163983. We acknowledge PRACE for awarding us access to Piz Daint at CSCS, Switzerland.
currently 5th largest cluster in the world & former 1st regarding GPUs until June ‘18
PRACE Tier-0 1.6 million GPU hours ~4000³ cells @ 2400 GPUs > 1010 particles O(104) GPU hours per simulation 13 TB / raw output step
3D simulation campaign on Piz-Daint
Marco Garten I HZDR – Research Group Computer Assisted Radiation Physics I www.hzdr.de/FWK
The Result of 1.6 Million GPU Hours
& 4 PB of Data
Copper, dopt = 17 nm, T = 60 fs, ELaser = 30 J
Observe optimum thickness & optimum leading edge at non-ideal contrast
Completely different from 2D simulation results!
Garten et al., in preparation
Maximum proton energy – parameter scan
longitudinal proton phase space
Pre-acceleration due to radiation pressure
Acceleration mechanism
Equivalent to line focus
2D3D
2D and 3D are conceptionally different!
Marco Garten I HZDR – Research Group Computer Assisted Radiation Physics I www.hzdr.de/FWK
PIConGPU our HPC work horse to run fast
full 3D simulations
open source development
efficient parallel I/O
a fully relativistic 3D3V particle-in-cell code developed and maintained at
Particle-in-Cell Simulations for the Exascale Era
a fully relativistic 3D3V particle-in-cell code
bandwidth bottleneck
Richard Pausch, HZDR
Writing reduced data is still required.
Writing reduced data is still required.
Compression does not help per se Huebl A. et al., (2017) ISC High Performance (pp 15-29) DOI: 10.1007/978-3-319-67630-2_2
Reducing complex data requires computing: This is not only true for simulations but for experiments as well
Detector/Simulation Initial Reduction
Complex data becomes harder to reduce quickly
Storage of data requires initial knowledge extraction
Online analysis becomes essential
Offline Analysis of stored data
Complexity (per event/data set/image and community): Machines + detectors produce more and more complex data, more and more scientific communities are involved in analyzing this data
electrons
Bremsstrahlung and synchrotron radiation do not just follow electron distribution
Would require explicit trajectories TB / s output … doable but unfeasible Instead simulate photons as well and record virtual detector signal
Develop analytical model to predict signal in experiments
photons
Harvest the wealth or particle data: Data reduction via synthetic diagnostics
radiation simulation:
live visualization:
benchmarked at
Conclusion Optimum proton energies at non-optimal laser contrast
Full 3D simulations become inevitable
Efficient data throughput matters
Enormous amounts of data require in-situ analysis and visualization
Acknowledgments
This project has been enabled by many people in open-source and open-science communities. Great thanks to the communities and developers of: PIConGPU, Jupyter, yt, the SciPy ecosystem, ADIOS, HDF5, CMake, openPMD, Spack, ...
This project has received funding from the European Unions Horizon 2020 research and innovation programme under grant agreement No 654220.
We acknowledge PRACE for awarding us access to Piz Daint at CSCS, Switzerland. 15th
PRACE call, Project ID 2016163983.
Thank you to the organizers!
This research used resources of the Oak Ridge Leadership Computing Facility located in the Oak Ridge National Laboratory, which is supported by the Office of Science of the Department of Energy under Contract DE-AC05-00OR22725.
Slide Number 1
Slide Number 2
Slide Number 3
Slide Number 4
Slide Number 5
Slide Number 6
Slide Number 8
Slide Number 9
Slide Number 10
Slide Number 11
Slide Number 12
Slide Number 13
Slide Number 14
Slide Number 15
Slide Number 16
Slide Number 17
Slide Number 18
Slide Number 19
Slide Number 20
Slide Number 21
Slide Number 22
The main issue is throughput
The main issue is throughput
Reducing complex data requires computing:This is not only true for simulations but for experiments as well
Slide Number 26
Slide Number 27
Slide Number 28
Slide Number 29
Slide Number 30

Documents

Radiation imprint of ultra- intense laser heating of solids