PSI, Villigen, Switzerland
Tech-X, Boulder, Colorado, USA
LBNL, Berkeley, California, USA
Because of their ultra-high accelerating gradient, laser plasma based accelerators (LPA) are contemplated for the next generation of high-energy colliders and light sources. The upcoming BELLA project will explore acceleration of electron bunches to 10 GeV in a 1 meter long plasma, where a wakefield is driven by a PW-class laser. Particle-in-cell (PIC) simulations are used to design the upcoming experiments where boosted frame simulations are used to model the full scale stages. As criteria on energy spread and beam emittance become more stringent, PIC simulations become more challenging as high frequency noise artificially increases those quantities. We show that calculating the beam self-fields using a static Poisson solve in the beam frame dramatically reduces particle noise, allowing for more accurate simulation of the beam evolution. In particular, this method gets correct cancellation of the transverse self-electric and magnetic fields of the beam, eliminating artificial self-forces, which is usually not true when using the standard PIC algorithm based on the staggered (“Yee”) electromagnetic field solver.
Saint-Petersburg State University, Saint-Petersburg, Russia
We analyze radiation of a small bunch crossing a boundary between two dielectrics in a cylindrical waveguide. The total energy of radiation was studied earlier for such problem but dynamics of an energy loss as well as a field structure was not investigated. Meanwhile these questions are of essential interest for the wakefield acceleration technique and for new methods of generation of microwave radiation. Our research is based on original approach used previously for the case of the vacuum-plasma boundary*. The principal difference of presented work consists in generation of Cherenkov radiation in dielectric and so-called Cherenkov-transition radiation in vacuum. Algorithms of computations for the field and the energy loss are founded upon certain transformations of integration path. Comparison of analytical results with numerical ones shows a good coincidence. We consider two instances in detail: the bunch is flying from vacuum into dielectric and from dielectric into vacuum. In both cases we compare the energy losses by transition radiation and by Cherenkov one. Our investigation shows, for example, that energy loss can be negative at certain segments of the bunch trajectory.
* T.Yu. Alekhina and A.V. Tyukhtin, Phys. Rev. E. 83, 066401 (2011)
LBNL, Berkeley, California, USA
Numerical modeling of laser-plasma accelerators using the ponderomotive approximation allows efficient modeling of 10 GeV and beyond laser-plasma accelerators. The time-averaged ponderomotive force approximation also allows simulation in cylindrical geometry which captures relevant 3D physics at 2D computational cost. In this talk I will present the code INF&RNO (INtegrated Fluid & paRticle simulatioN cOde). The code is based on an envelope model for the laser while either a PIC or a fluid description can be used for the plasma. The effect of the laser pulse on the plasma is modeled with the time-averaged poderomotive force. These and other features, such as dynamical resampling of the phase space distribution to reduce on-axis noise and boosted-Lorentz-frame modeling capability, allow for a speedup of several orders of magnitude compared to standard full PIC simulations while still retaining physical fidelity. The code has been benchmarked against analytical solutions and 3D PIC simulations and a set of validation tests together with a discussion of the performances will be presented. Applications to the BELLA PW laser-plasma accelerator experiments at LBNL will be discussed.
LBNL, Berkeley, California, USA
LLNL, Livermore, California, USA
The Particle-In-Cell Code-Framework Warp originated in the Heavy Ion Fusion program to guide the development of accelerators that can deliver beams suitable for implosion of inertial fusion capsules. The range of application of Warp has considerably widened far beyond the initial area and it is now applied to the study and design of existing and next-generation high-energy accelerators, including, for example, the study of laser wakefield acceleration and electron cloud effects. We present an overview of Warp's capabilities, summarizing recent original numerical methods that were developed to address computational challenges such as space and time scale disparities, spurious numerical dispersion, efficient wideband digital filtering on parallel platforms, etc. The original methods include simulations in Lorentz boosted frames, an electromagnetic solver with tunable numerical dispersion and efficient stride-based digital filtering, Particle-In-Cell with Adaptive Mesh Refinement, a large-timestep ‘‘drift-Lorentz'' mover for arbitrarily magnetized species, and a relativistic Lorentz invariant leapfrog particle pusher. Selected examples of applications will be given.
DESY Zeuthen, Zeuthen, Germany
ANSL, Yerevan, Armenia
TEMF, TU Darmstadt, Darmstadt, Germany
GSI, Darmstadt, Germany
[1] http://www.txcorp.com/products/VORPAL/
[2] M. Passoni et al., Phys Rev E 69, 026411 (2004)
[3] P. Mora, Phys. Rev. Lett., 90, 185002 (2003)
LBNL, Berkeley, California, USA
While there is wisdom in the old adage "the two constants in life are death and taxes," there are unavoidable truths facing modern experimental and computational science. First is the growing "impedence mismatch" between our ability to collect and generate data, and our ability to store, manage, and gain understanding from it. The second is the fact that we cannot continue to rely on the same software technologies that have worked well for the past couple of decades for data management, analysis, and visualization. A third is that these complementary activities must be considered in a holistic, rather than balkanized way. The inseperable interplay between data management, analysis, visualization, and high performance computational infrastructure, are best viewed through the lens of case studies from multiple scientific domains, where teams of computer and accelerator scientists combine forces to tackle challenging data understanding problems.