Computational Science is applied to the field of Accelerator Physics to enable fundamental advances in theory and technology leading to new capabilities in particle accelerators.
Importance of Accelerator Physics theory and technology
Experiments based on Accelerator Physics theory and technology have made remarkable discoveries about the basic nature of matter. These discoveries include nuclear structure, the behavior of nuclear matter, quark dynamics, the nature of elementary particles and the fundamental forces, unified field theories, and cosmology. Future possible discoveries include quark plasmas, supersymmetric counterparts to the known existing particles, and the fundamental origin of mass.
At the same time, Accelerator Physics theory and technology have made substantial contributions to other branches of science and technology. Major existing applications include electron microscopy, microprobes, charged-particle-beam lithography, ion implantation, isotope production, particle beams for precision irradiation therapy, superconducting magnets and medical magnetic resonance imaging, neutral-beam heating of plasmas, synchrotron light sources, x-ray lithography, and free-electron lasers. Just this past year synchrotron light was used in the field of biology to determine how meters of DNA can be coiled and managed in cells, to determine the long-sought structure of bacteriorhodopsin, and to determine the largest x-ray crystal structure to date, that of the blue tongue virus made of more that 1000 separate proteins. Indeed, because of their great utility, some 26 new synchrotron light sources are anticipated worldwide.
Possible future applications of Accelerator Physics theory and technology include intense beams for inertial fusion, the production of tritium, the production of nuclear fuels, the transmutation of nuclear waste, and high-speed and high-resolution proton radiography for nuclear stockpile stewardship.
The principal new facility planned for the FY1999 budget by the US Department of Energy is the $1.3B Spallation Neutron Source. This collaborative project, involving several DOE laboratories, is based entirely on Accelerator Physics supported by modeling and simulation of the components and the entire facility.
The Role of Computation In Accelerator Physics
Accelerator Physics theory and technology make extensive use of modern numerical methods and computational facilities. First, extensive computation of electric and magnetic fields is required to design high quality magnets and high performance RF components. These computations involve fast and highly accurate 3-dimensional Maxwell solvers based both on mesh and finite-element methods.
Second, extensive computation is required to study the nature and long-term stability of single-particle orbits moving in these electromagnetic fields. For example, determining the stability of protons in the CERN Large Hadron Collider currently under design is equivalent in effort to determining the stability of the solar system since the Big Bang. Significant advances have been made in the study of single-particle orbits including the use of truncated power series algebra, Lie algebraic methods, symplectic integraters, normal forms, and the extensive use of maps. These methods, which require both extensive symbolic and numerical computation, have led to substantial improvements in both linear and circular accelerators.
Finally, extensive computation is required to treat both random effects and collective effects involving large numbers of particles. Random effects include synchrotron radiation reaction, RF noise, power supply noise, and ground motion. Collective effects include those arising from electromagnetic wake-fields produced by particles as they pass through various structures, space-charge effects in high current machines, laser-plasma-beam effects in laser or charged-particle driven acceleration in plasmas, and beam-beam effects in both linear and circular colliders. The current calculation of these effects often involves the use of specially written particle-in-cell codes.
Future Computational Needs
In order to realize the promise of future applications of Accelerator Physics theory and technology described above, and in order to realize possible new advances in Accelerator Physics such as laser or particle-driven acceleration in plasmas, acceleration in crystal lattices, and ultra-cold crystal beams, new computational resources will be required. The computation of electromagnetic fields and the computation of a limited number of single-particle orbits in these fields can be carried out by making extensive use of current computer hardware. However, in addition to computing a small number of single-particle orbits, in many cases it would be highly desirable to simulate the behavior of tens of thousands of noninteracting particles in order to make comparisons between theoretical predictions and the behavior of beams in existing accelerators. Such calculations are currently impossible, but are within reach of the next generation of large-scale (100 TByte and 100 Tflop) parallel computers. It is anticipated that the treatment of random effects, which are important for existing electron machines and the Large Hadron Collider, could also be carried out using such computers. By contrast, it is anticipated that a completely satisfactory treatment of long-term many-particle collective effects awaits the advent of computers with Peta-FLOPS speed. Finally, in additional to computational power, advances are required in visualization. For example, in understanding even single-particle phenomena, it would be very helpful to have ways of visualizing 4-dimensional (and even 6-dimensional) phase spaces.
Two Significant Problems
The requirements to proceed to the next level of sophistication in beam dynamics simulations can be represented by considerations of the so-called strong-strong beam-beam interaction. A sample distribution of several million particles in each beam must be tracked for many damping times, or at least 100,000 turns. At each turn, the fields from the beam distribution (which is highly elliptical and can contain important tail regions) must be found from the distribution and then the particles from the two counter-moving beams propagated through the fields. To approach this problem successfully would require a factor of 1000 more cpu power and the development of adequate and tested algorithms.
The complex physical phenomena that occur in the interaction region of an electron-positron linear collider can only be approximately modeled using present capabilities. Achieving the required luminosity for a future TeV linear collider requires on the order of 1010 particles per bunch in a collision area of roughly 5 by 400 nanometers. To successfully explore the physics from such an interaction requires separating the subnuclear particle production from the background created by the electromagnetic interactions of the colliding beams. Detailed simulation of these phenomena requires orders of magnitude greater computational power than exists in present facilities.
Summary
Compared to the beam dynamics in an accelerator or storage ring, the simulation capability that has been developed is still simplified and approximate. To understand many experimental observations, such as the beam-beam limit, beam lifetime, and particle backgrounds in detectors, we need the development of more sophisticated computational tools and better algorithms. We have made only the first crucial steps toward the goal of understanding and improving particle accelerators using computational science.