James Osborn
(Argonne National Lab)
17/06/2010, 14:30
Algorithms and machines
I will present an adaptive multigrid Dirac solver developed for Wilson
clover fermions which offers order-of-magnitude reductions in solution
time compared to conventional Krylov solvers. The solver incorporates
even-odd preconditioning and mixed precision to solve the Dirac
equation to double precision accuracy and shows only a mild increase
in time to solution for decreasing quark...
Nazario Tantalo
(INFN sez. Roma "Tor Vergata")
17/06/2010, 14:50
Algorithms and machines
In the solution of the lattice Dirac equation with heavy quarks masses the numerical precision obtained at
large time distances from the source may be significantly smaller with respect to the precision obtained at
small times. This happens for each gauge configuration because typical algorithms used to perform numerical
inversions require that a "global" residue is smaller than a...
Richard C. Brower
(Boston University)
17/06/2010, 15:10
Algorithms and machines
Computation of disconnected or non-valence quark loops presents a
computationally challenging problem. Absent these contributions
operators with zero flavor quantum numbers can not be studied. This is
a report on progress to measure the strange quark contributions to
nucleon form factors. The increasing importance of GPU architectures
and multigrid methods will be discussed, as will...
Alan O'Cais
(Cyprus Institute)
17/06/2010, 15:30
Algorithms and machines
The exact computation of the disconnected diagram contribution to a number of hadronic observables, such as the eta prime mass and the nucleon isoscalar form factors, is considered to explicitly expose the gauge noise associated with these diagrams. A Wilson action on lattices of SU(3) SESAM gauge field configurations with two dynamical flavors and a volume of 16^3X32 are utilized. An exact...
Francesco Virotta
(NIC, DESY Zeuthen)
17/06/2010, 15:50
Algorithms and machines
Simulations of QCD suffer from severe critical slowing down towards the continuum limit. This problem is known to be prominent in the topological charge, however, all observables are affected to various degree by these slow modes in the Monte Carlo evolution. We investigate the slowing down in high statistics quenched and dynamical fermion simulations and propose a new error analysis method,...
