Accelerating diffusions
Open accessThe DOI matches its Crossref record; 22 of the 23 references it cites were found in a registry or an index.
Abstract
Let U be a given function defined on ℝd and π(x) be a density function proportional to exp−U(x). The following diffusion X(t) is often used to sample from π(x), $$dX(t)=-\nabla U(X(t))\,dt+\sqrt{2}\,dW(t),\qquad X(0)=x_{0}.$$ To accelerate the convergence, a family of diffusions with π(x) as their common equilibrium is considered, $$dX(t)=\bigl(-\nabla U(X(t))+C(X(t))\bigr)\,dt+\sqrt{2}\,dW(t),\qquad X(0)=x_{0}.$$ Let LC be the corresponding infinitesimal generator. The spectral gap of LC in L2(π) (λ(C)), and the convergence exponent of X(t) to π in variational norm (ρ(C)), are used to describe the convergence rate, where λ(C)=Sup {real part of μ:μ is in the spectrum of LC, μ is not zero}, $$\rho(C)=\operatorname {Inf}\biggl\{\rho: \int \vert p(t,x,y)-\pi(y)\vert \,dy\le g(x)e^{\rho t}\biggr\}.$$ Roughly speaking, LC is a perturbation of the self-adjoint L0 by an antisymmetric operator C⋅∇, where C is weighted divergence free. We prove that λ(C)≤λ(0) and equality holds only in some rare situations. Furthermore, ρ(C)≤λ(C) and equality holds for C=0. In other words, adding an extra drift, C(x), accelerates convergence. Related problems are also discussed.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“In brief, the mixing afforded by solenoidal flow can render gradient descent more efficient [112, 113, 114, 115, 116].”
From The free energy principle made simpler but not too simple (2023) · cites this paper as [113]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Stochastic relaxation, gibbs distributions, and the bayesian restoration of images.
Geman, S., & Geman, D. (1984) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.1984.4767596 · Cited by 17,949 · cited 1×
“In Geman and Geman (1984), Amit and Grenander (1991) and Hwang and Sheu (1998) the convergence properties of some nonreversible Gibb samplers are studied.”
BackgroundPerturbation Theory for Linear Operators
Kato, T. (1995) · Classics in Mathematics · doi:10.1007/978-3-642-66282-9 · Cited by 16,290
Markov Chain Monte Carlo in Practice
(1995) · Chapman & Hall/CRC eBooks · doi:10.1201/b14835 · Cited by 7,757
General state space Markov chains and MCMC algorithms
Roberts, G. O., & Rosenthal, J. S. (2004) · Probability Surveys · doi:10.1214/154957804100000024 · Cited by 734 · cited 1×
“…Frigessi, Hwang, Sheu and di Stefano (1993), Hwang, Hwang-Ma and Sheu (1993), Amit (1996), Athreya, Doss and Sethuraman (1996), Gilks and Roberts (1996), Mengersen and Tweedie (1996), Stramer and Tweedie (1997), Chang and Hwang (1998), Hwang and Sheu (1998, 2000) and Roberts and Rosenthal (2004).”
BackgroundRepresentations of Knowledge in Complex Systems
Grenander, U., & Miller, M. I. (1994) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/j.2517-6161.1994.tb02000.x · Cited by 603 · cited 1×
“For applications one may consult Grenander and Miller (1994), Miller, Srivastava and Grenander (1995), Srivastava (1996) and references therein.”
BackgroundRates of convergence of the Hastings and Metropolis algorithms
Mengersen, K. L., & Tweedie, R. L. (1996) · The Annals of Statistics · doi:10.1214/aos/1033066201 · Cited by 553
Pointwise estimates and quasilinear parabolic equations
Trudinger, N. S. (1968) · Communications on Pure and Applied Mathematics · doi:10.1002/cpa.3160210302 · Cited by 254 · cited 2×
“Since ∇U and C are locally bounded, by a local Harnack inequality [Theorem 1.1 in Trudinger (1968)], ∀x in Rd,∀N > 0,∀ f with π(f) = 1 and f ≥ 0, Sup y∈B(x,N/2) T (s)f(y)≤C(N,x) Inf y∈B(x,N/2) T (2s)f(y),(8) where the constant C(N,x) depends only on N and x, and B(x,N/2) denotes a ball in Rd with…”
BackgroundInfluentialFrom Markov Chains to Non-Equilibrium Particle Systems
Chen, M. F. (1992) · WORLD SCIENTIFIC eBooks · doi:10.1142/1389 · Cited by 220 · cited 4×
“For the reversible case, the L1(π) to L1(π) exponential convergence rate is equivalent to the essentially uniform boundedness of g(x) in (6) [Chen (2002)].”
BackgroundInfluentialBogachev, V. I., Krylov, N. V., & Röckner, M. (2001) · Communications in Partial Differential Equations · doi:10.1081/pde-100107815 · Cited by 214
Weak Poincaré Inequalities and L2-Convergence Rates of Markov Semigroups
Röckner, M., & Wang, F.-Y. (2001) · Journal of Functional Analysis · doi:10.1006/jfan.2001.3776 · Cited by 160 · cited 2×
“By Lemma 2.2 in Röckner and Wang (2001), for s≤ t and π(f2) = 1, ‖T (s)f‖2 ≤ (‖T (t)f‖2)s/t ≤C(N,f)s/te2ρs.”
InfluentialAccelerating Gaussian diffusions
Hwang, C.-R., Hwang-Ma, S.-Y., & Sheu, S.-J. (1993) · The Annals of Applied Probability · doi:10.1214/aoap/1177005371 · Cited by 108
Self-citationConditional-mean estimation via jump-diffusion processes in multiple target tracking/recognition
Miller, M. I., Srivastava, A., & Grenander, U. (1995) · IEEE Transactions on Signal Processing · doi:10.1109/78.482117 · Cited by 103
On the convergence of the Markov chain simulation method
Athreya, K. B., Doss, H., & Sethuraman, J. (1996) · The Annals of Statistics · doi:10.1214/aos/1033066200 · Cited by 99
Frigessi, A., Stefano, P., Hwang, C.-R., & Sheu, S. (1993) · Journal of the Royal Statistical Society Series B (Statistical Methodology) · doi:10.1111/j.2517-6161.1993.tb01479.x · Cited by 97
Self-citationFrigessi, A., Hwang, C.-R., & Younès, L. (1992) · The Annals of Applied Probability · doi:10.1214/aoap/1177005652 · Cited by 83
Self-citationComparing sweep strategies for stochastic relaxation
Amit, Y., & Grenander, U. (1991) · Journal of Multivariate Analysis · doi:10.1016/0047-259x(91)90080-l · Cited by 73 · cited 2×
“Related works may be found in Amit (1991), Amit and Grenander (1991), Frigessi, Hwang and Younes (1992), Frigessi, Hwang, Sheu and di Stefano (1993), Hwang, Hwang-Ma and Sheu (1993), Amit (1996), Athreya, Doss and Sethuraman (1996), Gilks and Roberts (1996), Mengersen and Tweedie (1996), Stramer…”
BackgroundOn rates of convergence of stochastic relaxation for Gaussian and non-Gaussian distributions
Amit, Y. (1991) · Journal of Multivariate Analysis · doi:10.1016/0047-259x(91)90033-x · Cited by 64 · cited 2×
“Related works may be found in Amit (1991), Amit and Grenander (1991), Frigessi, Hwang and Younes (1992), Frigessi, Hwang, Sheu and di Stefano (1993), Hwang, Hwang-Ma and Sheu (1993), Amit (1996), Athreya, Doss and Sethuraman (1996), Gilks and Roberts (1996), Mengersen and Tweedie (1996), Stramer…”
BackgroundConvergence properties of the Gibbs sampler for perturbations of Gaussians
Amit, Y. (1996) · The Annals of Statistics · doi:10.1214/aos/1033066202 · Cited by 31 · cited 1×
“…(1991), Amit and Grenander (1991), Frigessi, Hwang and Younes (1992), Frigessi, Hwang, Sheu and di Stefano (1993), Hwang, Hwang-Ma and Sheu (1993), Amit (1996), Athreya, Doss and Sethuraman (1996), Gilks and Roberts (1996), Mengersen and Tweedie (1996), Stramer and Tweedie (1997), Chang and Hwang…”
BackgroundExistence of the spectral gap for elliptic operators
Wang, F.-Y. (1999) · Arkiv för matematik · doi:10.1007/bf02412223 · Cited by 26 · cited 3×
“(A2) The existence of a spectral gap for self-adjoint L has been studied extensively, for example, see Wang (1999).”
BackgroundInfluentialErgodic convergence rates of Markov processes--eigenvalues, inequalities and ergodic theoryPreprint
Chen, M.-F. (2003) · arXiv (Cornell University) · doi:10.48550/arxiv.math/0304367 · Cited by 11 · cited 2×
“For the reversible case, the L1(π) to L1(π) exponential convergence rate is equivalent to the essentially uniform boundedness of g(x) in (6) [Chen (2002)].”
BackgroundOn the Geometrical Convergence of Gibbs Sampler inRd
Hwang, C.-R., & Sheu, S.-J. (1998) · Journal of Multivariate Analysis · doi:10.1006/jmva.1997.1735 · Cited by 4
Self-citationA Remark on the Ergodicity of Systematic Sweep in Stochastic Relaxation
Hwang, C.-R., & Sheu, S. (1992) · Lecture notes in statistics · doi:10.1007/978-1-4612-2920-9_13 · Cited by 2
Self-citation
Research works printed but not linked: 4
- CHANG, H.-C. and HWANG, C.-R. (1998). On the average-case analysis of dynamic Monte Carlo schemes. In 3rd International Conference on Monte Carlo and Quasi-Monte Carlo Method.
- NAGEL, R. (1986). One-Parameter Semigroup of Positive Operators. Springer, Berlin.
- Srivastava, A. (1996). Inferences on transformation groups generating patterns on rigid motions. Ph.D. dissertation, Dept. Electrical Engineering, Washington Univ.
- Yosida, K. (1980). Functional Analysis, 6th ed. Springer, New York.
Cited by
Stochastic Processes and Applications: Diffusion Processes, the Fokker-Planck and Langevin Equations
Pavliotis, G. A. (2014) · Texts in Applied Mathematics · doi:10.1007/978-1-4939-1323-7 · Cited by 256
Partial differential equations and stochastic methods in molecular dynamics
Lelièvre, T., & Stoltz, G. (2016) · Acta Numerica · doi:10.1017/s0962492916000039 · Cited by 178
The free energy principle made simpler but not too simple
Friston, K., Da Costa, L., Sajid, N., Heins, C., Ueltzhöffer, K., Pavliotis, G. A., & Parr, T. (2023) · Physics Reports · doi:10.1016/j.physrep.2023.07.001 · Cited by 163
“In brief, the mixing afforded by solenoidal flow can render gradient descent more efficient [112–116].”
Generalisation of the Eyring–Kramers Transition Rate Formula to Irreversible Diffusion Processes
Bouchet, F., & Reygner, J. (2016) · Annales Henri Poincaré · doi:10.1007/s00023-016-0507-4 · Cited by 176
“A means to do so [28, 29, 34] is to consider the stochastic differential equation dX t = −∇U(X t )dt+ J∇U(X t )dt+ √ 2ǫdWt, (2.”
BackgroundOptimal non-reversible linear drift for the convergence to equilibrium of a diffusion
Lelièvre, T., Nier, F., & Pavliotis, G. A. (2013) · Journal of Statistical Physics · doi:10.1007/s10955-013-0769-x · Cited by 123
“As noticed in [24, 25], one way to accelerate the convergence to equilibrium is to depart from reversible dynamics (see also [10] for related discussions for Markov Chains).”
Metadata
- Authors
- Hwang, C.-R., Hwang-Ma, S.-Y., & Sheu, S.-J.
- Affiliations
- Soochow University (CN)
- Venue
- The Annals of Applied Probability
- Publisher
- Institute of Mathematical Statistics
- Published
- 2005-05-01
- Volume
- 15
- Issue
- 2
- Pages
- 1433-1444
- ISSN
- 1050-5164 (ISSN-L)
- 2168-8737
- OpenAlex
- W3037124231
- Cited by
- 113
- References
- 23
- Citations by year
- 2025: 10 · 2024: 10 · 2023: 8 · 2022: 12 · 2021: 10 · 2020: 10 · 2019: 8 · 2018: 7 · 2017: 5 · 2016: 11 · 2015: 5 · 2014: 4 · 2013: 7 · 2012: 2
- FWCI
- 0.00
- Type
- journal article
- Language
- en
- Open access
- bronze
- doi.org
- Locations
- The Annals of Applied Probability (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- Project Euclid (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- National Science Council
- Topics
- Physical Sciences › Mathematics › Statistics and Probability › Markov Chains and Monte Carlo Methods
- Physical Sciences › Mathematics › Mathematical Physics › Stochastic processes and statistical mechanics
- Physical Sciences › Mathematics › Applied Mathematics › Geometric Analysis and Curvature Flows
- Keywords
- accelerated convergence, diffusion, spectral gap, convergence rate, infinitesimal generator, Markov chain Monte Carlo
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.