Email RIS file
Bookmark
Report an Issue

Monte Carlo sampling strategies for lattice gauge calculations

Author
Guralnik, G.
Published
United States : [publisher not identified], 1985
Springfield, Va.: National Technical Information Service, [approximately 1985]
Physical Description
microfiche : negative ; 11 x 15 cm
Additional Creators
Warnock, T. and Zemach, C.

Availability

Browse Nearby on Shelf
I Want It
I Want It

Finding items...

Summary
We have sought to optimize the elements of the Monte Carlo processes for thermalizing and decorrelating sequences of lattice gauge configurations and for this purpose, to develop computational and theoretical diagnostics to compare alternative techniques. These have been applied to speed up generations of random matrices, compare heat bath and Metropolis stepping methods, and to study autocorrelations of sequences in terms of the classical moment problem. The efficient use of statistically correlated lattice data is an optimization problem depending on the relation between computer times to generate lattice sequences of sufficiently small correlation and times to analyze them. We can solve this problem with the aid of a representation of auto-correlation data for various step lags as moments of positive definite distributions, using methods known for the moment problem to put bounds on statistical variances, in place of estimating the variances by too-lengthy computer runs.
Report Numbers
DE85010756; LA-UR-85-1317; CONF-8504110-10
Other Subject(s)
Collection
NTIS collection.
Note
DOE contract number: W-7405-ENG-36
OSTI Identifier 5859477
Research organization: Cray Research, Inc., Los Alamos, NM (USA).
Research organization: Los Alamos National Laboratory (LANL), Los Alamos, NM (United States).
View MARC record | catkey: 47353165