Solving large sparse linear systems in end-to-end accelerator structure simulations

Lee, L.-Q.; Ge, Lixin; Kowalski, M.E.; Li, Z.; Ng, Cho; Schussman, Greg; Wolf, Michael M.; Ko, K.

doi:10.1109/ipdps.2004.1302909

Cited by 6 publications

(3 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, the emphasis was on creating new parallel or distributed algorithms, whereas our emphasis is on abstracting the fundamental underlying algorithm to produce a single generic version that encompasses both sequential and distributed computation. Lee et al [21] have found that the generic algorithms in the Iterative Template Library can instantiate to either parallel or sequential algorithms, using distributed vectors and matrices for distribution. Jézéquel [18] describes the use of Eiffel inheritance to build distributed algorithms and data structures from their sequential counterparts within the Eiffel Parallel Execution Environment (EPEE) [17], focusing on two core algorithms: a map() operation to transform one sequence into another and a reduce() operation to compute a single result from a sequence of values.…”

Section: Related Workmentioning

confidence: 99%

“…The separation of policy from mechanism has some farreaching and profound effects for high-performance computing. Because the cg() algorithm did not depend on the details of the application of A to x, the same cg() algorithm can be used to realize a parallel CG solver, simply by using distributed linear algebra objects [21]. For completeness we note that it is also necessary that the inner-product and norm functions make sense for the types of x and b.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Lifting sequential graph algorithms for distributed-memory parallel computation

Gregor

Lumsdaine

2005

Proceedings of the 20th Annual ACM SIGPLAN Conference on Object-Oriented Programming, Systems, Languages, and Applications

View full text Add to dashboard Cite

This paper describes the process used to extend the Boost Graph Library (BGL) for parallel operation with distributed memory. The BGL consists of a rich set of generic graph algorithms and supporting data structures, but it was not originally designed with parallelism in mind. In this paper, we revisit the abstractions comprising the BGL in the context of distributed-memory parallelism, lifting away the implicit requirements of sequential execution and a single shared address space. We illustrate our approach by describing the process as applied to one of the core algorithms in the BGL, breadth-first search. The result is a generic algorithm that is unchanged from the sequential algorithm, requiring only the introduction of external (distributed) data structures for parallel execution. More importantly, the generic implementation retains its interface and semantics, such that other distributed algorithms can be built upon it, just as algorithms are layered in the sequential case. By characterizing these extensions as well as the extension process, we develop general principles and patterns for using (and reusing) generic, object-oriented parallel software libraries. We demonstrate that the resulting algorithm implementations are both efficient and scalable with performance results for several algorithms.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Lifting sequential graph algorithms for distributed-memory parallel computation

Gregor

Lumsdaine

2005

Proceedings of the 20th Annual ACM SIGPLAN Conference on Object-Oriented Programming, Systems, Languages, and Applications

View full text Add to dashboard Cite

show abstract

“…A lot of research work is in progress in this direction. Lie-Quan Lee et al [7] have proposed a parallel hybrid solver for sparse linear [8]. Cai et al [9] have developed parallel iterative solvers for modern physical applications.…”

Section: Introductionmentioning

confidence: 99%

Non-stationary iterative solvers on a PC cluster

Kumar

Shalini

Mehra

2005

Advances in Engineering Software

View full text Add to dashboard Cite

Coupler Design for the LCLS Injector S-Band Structures

Li¹,

Chan²,

Bentson³

et al.

Proceedings of the 2005 Particle Accelerator Conference

View full text Add to dashboard Cite

The LCLS injector is required to provide a 1-nC, 10-ps bunch with a normalized rms transverse projected emittance of less than 1 micron. The LCLS beam is generated and accelerated in a 1.6-cell S-band RF gun at 120 MV/m up to 6 MeV. The gun is followed by two SLAC 3-m S-band accelerator structures to further accelerate the beam to 135 MeV which moves the beam out of the space-charge dominated regime. In the SLAC S-band structures, the RF power feed is through a single coupling-hole (single-feed coupler) which results in a field asymmetry. The time dependent multipole fields in the coupler induce a transverse kick along the bunch and cause the emittance to increase above the LCLS specification. To meet the stringent emittance requirements for the injector, the single-feed couplers will be replaced by a dual-feed racetrack design to minimize the multipole field effects. We will present detailed studies of the multipole fields in the SLAC linac RF coupler and the improvements with the dual-feed ractrack design using the parallel finite element S-parameter solver S3P.

show abstract

Solving large sparse linear systems in end-to-end accelerator structure simulations

Cited by 6 publications

References 8 publications

Lifting sequential graph algorithms for distributed-memory parallel computation

Lifting sequential graph algorithms for distributed-memory parallel computation

Non-stationary iterative solvers on a PC cluster

Coupler Design for the LCLS Injector S-Band Structures

Contact Info

Product

Resources

About