Hybrid Monte Carlo

Duane, S; Kennedy, A.D.; Pendleton, B.J.; Roweth, Duncan

doi:10.1016/0370-2693(87)91197-x

Cited by 3,526 publications

(3,113 citation statements)

References 11 publications

Supporting

Mentioning

3,022

Contrasting

Unclassified

Order By: Relevance

“…The subsets of gauge fields, where µ is far below the central value of the distribution, occur with such a small probability in this case that their contributions to the common physical observables can be safely neglected. Numerical simulations, using a preconditioned Hybrid Monte Carlo (HMC) algorithm [13], for example, will then normally run smoothly and produce a representative sample of field configurations as expected.…”

Section: Sources Of Instabilitymentioning

confidence: 99%

Stability of lattice QCD simulations and the thermodynamic limit

Debbio¹,

Giusti²,

Lüscher³

et al. 2006

J. High Energy Phys.

148

253

View full text Add to dashboard Cite

We study the spectral gap of the Wilson-Dirac operator in two-flavour lattice QCD as a function of the lattice spacing a, the space-time volume V and the current-quark mass m. It turns out that the median of the probability distribution of the gap scales proportionally to m and that its width is practically equal to a/ √ V . In particular, numerical simulations are safe from accidental zero modes in the large-volume regime of QCD.

show abstract

Section: Sources Of Instabilitymentioning

confidence: 99%

Stability of lattice QCD simulations and the thermodynamic limit

Debbio¹,

Giusti²,

Lüscher³

et al. 2006

J. High Energy Phys.

148

253

View full text Add to dashboard Cite

show abstract

“…In sampling step (7), we use Hamiltonian Monte Carlo (HMC) [18,19] to update the absorption coefficients µ m conditioned on the current segmentation z (l) . HMC is an efficient method to generate random samples from probability distributions over continuous variables.…”

Section: Hamiltonian Monte Carlomentioning

confidence: 99%

Autonomous reconstruction and segmentation of tomographic data

Wollgarten

Habeck

2014

Micron

View full text Add to dashboard Cite

A Bayesian approach to reconstruction and segmentation of tomographic data is outlined and further detailed for the case of absorption tomography. The algorithm allows the quantification of reconstruction errors and segmentation confidence. Calculation results for various experimental settings (number of projections, incident dose, different materials) are shown and discussed.We are pleased to dedicate this paper to the memory of David J. H. Cockayne. HighlightsSimultaneous tomographic reconstruction and segmentation algorithm Fully self-contained Probabilistic error metrics for both reconstructed densities and segmentation on a voxel basis Highlights (for review)Autonomous reconstruction and segmentation of tomographic data We are pleased to dedicate this paper to the memory of David J. H. Cockayne.

show abstract

“…Since the posterior distribution of θ is generally of complicated form, it is desirable to use MCMC methods to draw random samples from p(θ|X, y). Particularly appropriate for this task is the Hamiltonian Monte Carlo 1 method [34,35]. The detailed algorithm for Hamiltonian MC is given in the Appendix.…”

Section: Gaussian Process Regressionmentioning

confidence: 99%

“…The Hamiltonian Monte Carlo (HMC) is a family of MCMC methods based on the concept of dynamic systems in physics [34]. Recently HMC has been shown in various applications to converge significantly faster than the Metropolis-Hastings algorithm, since the dynamic method avoids the random walk behavior inherent in conventional approaches [34,35,33].…”

Section: Appendix: the Hamiltonian Monte Carlo Algorithmmentioning

confidence: 99%