Robert Ryne, Lawrence Berkeley National Laboratory
Kwok Ko, Zenghai Li, and Cho Ng, Stanford Linear Accelerator Center
Salman Habib and Ji Qiang, Los Alamos National Laboratory
Viktor Decyk and Warren Mori, University of California, Los Angeles
Panagiotis Spentzouris, Fermi National Accelerator Laboratory
Alex Dragt, University of Maryland
Gene Golub, Stanford University
Kwan-Liu Ma, University of California, Davis
Esmond Ng, NERSC

Top: Omega3P simulation of the PEP-II interaction region (from crotch to crotch) showing domain decomposition of a portion of the mesh around the interaction point. Bottom: Wall loss distribution of a trapped mode in this region of the vacuum chamber.

Research Objectives
The primary objective of this project is to establish a comprehensive terascale simulation environment for use by the U.S. particle accelerator community, enabling physicists and engineers to solve the most challenging problems in accelerator design, analysis, and optimization. Terascale simulation will help ensure the success of the next generation of particle accelerators by facilitating design decisions aimed at controlling and reducing cost, reducing risk, and optimizing performance.

Computational Approach
This project has three main physics-based focus areas: beam systems simulation (BSS), electromagnetic systems simulation (ESS), and beam/electromagnetic systems integration (BESI). The BSS component uses parallel particle-in-cell (PIC) techniques, particle managers, dynamic load balancing, FFT-based Poisson solvers, and techniques from the field of magnetic optics. The ESS component utilizes unstructured mesh generation, domain decomposition, adaptive mesh refinement, finite elements and sparse linear solvers (for eigenmode codes), and unstructured Yee grids (for time-domain codes). The BESI component involving particles in electromagnetic structures will utilize hybrid grids, with a structured mesh in the region of the beam and an unstructured grid near the structure boundaries.

Accomplishments
The Omega3P parallel electromagnetic eigenmode solver was run on NERSC platforms to help design and evaluate accelerating structures for the Next Linear Collider (NLC). In particular, simulations of structures comprised of roughly 50 cells were performed, which is close to the planned ~200 cells in an NLC traveling wave structure. Simulations of an alternative standing wave structure, which would be comprised of ~50 cells, have also begun and may hold the key to an improved NLC design. Omega3P was also used to study anomalous heating in the PEP-II B-factory interaction region. The heating has been identified as being caused by a "trapped mode," and methods to deal with this problem are under investigation.

The 3D parallel PIC code IMPACT was used to model the Low-Energy Demonstration Accelerator (LEDA) beam halo experiment, the Spallation Neutron Source linac, and the CERN superconducting proton linac. Extensive simulations were performed on the behavior of nonequipartioned beams, work that is relevant to almost all proposed high-intensity proton linac designs. IMPACT was also used to numerically generate, for the first time, 3D nonequipartioned self-consistent solutions of the Poisson/Vlasov equations for a beam in a constant focusing channel. This work involved a combination of analytical work to mathematically describe the equations governing the self-consistent solutions and large-scale computations (involving a parallel Langevin code) to numerically generate them and study the stability properties. Finally IMPACT was combined with GEANT3 to produce the first parallel simulation of a muon cooling channel.

Significance
The design of the next generation of accelerators will require a new level of simulation capability as accelerator designers push the boundaries of beam intensity, beam energy, and system complexity. All near- and far-future accelerators have very challenging modeling requirements that require high performance computing.

Publications
F. Gerigk, M. Vretenar, and R. Ryne, "Design of the superconducting SPL linac at CERN," in Proc. 2001 Particle Accelerator Conference (June 2001).

I. Hofmann, J. Qiang, and R. Ryne, "Collective resonance model of energy exchange in 3D nonequipartitioned beams," Phys. Rev. Lett. 86, 2313 (2001).

J. Qiang, R. Ryne, B. Blind, J. Billen, T. Bhatia, R. Garnett, G. Neuschaefer, and H. Takeda, "High resolution parallel particle-in-cell simulations of beam dynamics in the Spallation Neutron Source linac," Nucl. Instr. Meth. Phys. Res. A, 457, 1 (2001).

http://public.lanl.gov/ryne/gca.html