Diffusion limits of the random walk Metropolis algorithm in high dimensions (Q433896): Difference between revisions
From MaRDI portal
Created a new Item |
Set OpenAlex properties. |
||
(7 intermediate revisions by 5 users not shown) | |||
Property / author | |||
Property / author: Jonathan C. Mattingly / rank | |||
Property / author | |||
Property / author: Jonathan C. Mattingly / rank | |||
Normal rank | |||
Property / review text | |||
The authors establish a diffusion limit for a random walk Metropolis-Hastings (RWM) algorithm approximating samples from a measure on an infinite-dimensional space, more precisely, for the RWM algorithm applied to distribution families obtained as finite-dimensional approximations to a measure on an separable Hilbert space. In general, diffusion limits are used as a tool to measure the complexity of MCMC methods which have been applied to high-dimensional target measures with product structure. In that case, the individual components of the Markov chain satisfy an invariance principle with respect to a scalar stochastic differential equation. The present paper extends this approach to more general high-dimensional target measures naturally arising in applications. The established invariance principle is such that the entire Markov chain converges to an infinite-dimensional continuous-time stochastic process given by a Hilbert-valued stochastic differential equation, i.e., a stochastic partial differential equations (SPDE). In more detail, the authors consider target measures \(\pi\) on a real separable Hilbert space \(H\) which possess a Radon-Nikodym derivative with respect to a Gaussian measure \(\pi_0\) on \(H\) of the form \[ \frac{d\pi}{d\pi_0}=M_\Psi\exp(-\Psi(x)) \] for some real, measurable functional \(\Psi\). Then, a realisable implementation (necessarily in finite dimensions) of the RWM algorithm for \(\pi\) is obtained by applying an RWM algorithm for a projection \(\pi^N\) of \(\pi\) on an \(N\)-dimensional subspace of \(H\). The main result of the paper is that started in stationarity the piecewise constant linear interpolants of the Markov chains in the \(N\)-dimensional subspaces converge for \(N\to\infty\) weakly to the solution of an SPDE in a suitable subspace of \(H\) started at \(\pi\) which is the invariant measure for this SPDE. This is proved under a list of assumptions on the function \(\Psi\) and the trace class covariance operator of the Gaussian measure \(\pi_0\) which according to the authors are satisfied in many applications. The practical implications of this result are that at stationarity the work to explore that state space scales as \(\mathcal{O}(N)\) and the speed at which the invariant measure is explored can be maximized by tuning the acceptance rate probability to \(0.234\) analogous to the known case for measures of product structure. | |||
Property / review text: The authors establish a diffusion limit for a random walk Metropolis-Hastings (RWM) algorithm approximating samples from a measure on an infinite-dimensional space, more precisely, for the RWM algorithm applied to distribution families obtained as finite-dimensional approximations to a measure on an separable Hilbert space. In general, diffusion limits are used as a tool to measure the complexity of MCMC methods which have been applied to high-dimensional target measures with product structure. In that case, the individual components of the Markov chain satisfy an invariance principle with respect to a scalar stochastic differential equation. The present paper extends this approach to more general high-dimensional target measures naturally arising in applications. The established invariance principle is such that the entire Markov chain converges to an infinite-dimensional continuous-time stochastic process given by a Hilbert-valued stochastic differential equation, i.e., a stochastic partial differential equations (SPDE). In more detail, the authors consider target measures \(\pi\) on a real separable Hilbert space \(H\) which possess a Radon-Nikodym derivative with respect to a Gaussian measure \(\pi_0\) on \(H\) of the form \[ \frac{d\pi}{d\pi_0}=M_\Psi\exp(-\Psi(x)) \] for some real, measurable functional \(\Psi\). Then, a realisable implementation (necessarily in finite dimensions) of the RWM algorithm for \(\pi\) is obtained by applying an RWM algorithm for a projection \(\pi^N\) of \(\pi\) on an \(N\)-dimensional subspace of \(H\). The main result of the paper is that started in stationarity the piecewise constant linear interpolants of the Markov chains in the \(N\)-dimensional subspaces converge for \(N\to\infty\) weakly to the solution of an SPDE in a suitable subspace of \(H\) started at \(\pi\) which is the invariant measure for this SPDE. This is proved under a list of assumptions on the function \(\Psi\) and the trace class covariance operator of the Gaussian measure \(\pi_0\) which according to the authors are satisfied in many applications. The practical implications of this result are that at stationarity the work to explore that state space scales as \(\mathcal{O}(N)\) and the speed at which the invariant measure is explored can be maximized by tuning the acceptance rate probability to \(0.234\) analogous to the known case for measures of product structure. / rank | |||
Normal rank | |||
Property / Mathematics Subject Classification ID | |||
Property / Mathematics Subject Classification ID: 60J22 / rank | |||
Normal rank | |||
Property / Mathematics Subject Classification ID | |||
Property / Mathematics Subject Classification ID: 60J05 / rank | |||
Normal rank | |||
Property / Mathematics Subject Classification ID | |||
Property / Mathematics Subject Classification ID: 60F05 / rank | |||
Normal rank | |||
Property / Mathematics Subject Classification ID | |||
Property / Mathematics Subject Classification ID: 65C05 / rank | |||
Normal rank | |||
Property / Mathematics Subject Classification ID | |||
Property / Mathematics Subject Classification ID: 65C40 / rank | |||
Normal rank | |||
Property / Mathematics Subject Classification ID | |||
Property / Mathematics Subject Classification ID: 60H15 / rank | |||
Normal rank | |||
Property / zbMATH DE Number | |||
Property / zbMATH DE Number: 6053734 / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
Markov chain Monte Carlo | |||
Property / zbMATH Keywords: Markov chain Monte Carlo / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
random walk Metropolis-Hastings | |||
Property / zbMATH Keywords: random walk Metropolis-Hastings / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
difffusion limit | |||
Property / zbMATH Keywords: difffusion limit / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
optimal convergence | |||
Property / zbMATH Keywords: optimal convergence / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
scaling limits | |||
Property / zbMATH Keywords: scaling limits / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
convergence time | |||
Property / zbMATH Keywords: convergence time / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
stochastic partial differential equations | |||
Property / zbMATH Keywords: stochastic partial differential equations / rank | |||
Normal rank | |||
Property / reviewed by | |||
Property / reviewed by: Martin Georg Riedler / rank | |||
Normal rank | |||
Property / MaRDI profile type | |||
Property / MaRDI profile type: MaRDI publication profile / rank | |||
Normal rank | |||
Property / arXiv ID | |||
Property / arXiv ID: 1003.4306 / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Weak convergence of Metropolis algorithms for non-I.I.D. target distributions / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Asymptotic behaviour of a class of stochastic approximation procedures / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Optimal scalings for local Metropolis-Hastings chains on nonproduct targets in high dimensions / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: MCMC METHODS FOR DIFFUSION BRIDGES / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Q5190855 / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Pathwise accuracy and ergodicity of metropolized integrators for SDEs / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Optimal scaling of MaLa for nonlinear regression. / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: From Metropolis to diffusions: Gibbs states and optimal scaling. / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Stein's method for concentration inequalities / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Approximation of Bayesian Inverse Problems for PDEs / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Stochastic Equations in Infinite Dimensions / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Q3721531 / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Analysis of SPDEs arising in path sampling. II: The nonlinear case / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Q3015765 / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Analysis of SPDEs arising in path sampling. I: The Gaussian case / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Monte Carlo sampling methods using Markov chains and their applications / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Monte Carlo strategies in scientific computing. / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Q4828566 / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Weak convergence and optimal scaling of random walk Metropolis algorithms / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Optimal Scaling of Discrete Approximations to Langevin Diffusions / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Optimal scaling for various Metropolis-Hastings algorithms. / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Q4284032 / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Inverse problems: A Bayesian perspective / rank | |||
Normal rank | |||
Property / OpenAlex ID | |||
Property / OpenAlex ID: W2127836946 / rank | |||
Normal rank | |||
links / mardi / name | links / mardi / name | ||
Latest revision as of 10:29, 30 July 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Diffusion limits of the random walk Metropolis algorithm in high dimensions |
scientific article |
Statements
Diffusion limits of the random walk Metropolis algorithm in high dimensions (English)
0 references
8 July 2012
0 references
The authors establish a diffusion limit for a random walk Metropolis-Hastings (RWM) algorithm approximating samples from a measure on an infinite-dimensional space, more precisely, for the RWM algorithm applied to distribution families obtained as finite-dimensional approximations to a measure on an separable Hilbert space. In general, diffusion limits are used as a tool to measure the complexity of MCMC methods which have been applied to high-dimensional target measures with product structure. In that case, the individual components of the Markov chain satisfy an invariance principle with respect to a scalar stochastic differential equation. The present paper extends this approach to more general high-dimensional target measures naturally arising in applications. The established invariance principle is such that the entire Markov chain converges to an infinite-dimensional continuous-time stochastic process given by a Hilbert-valued stochastic differential equation, i.e., a stochastic partial differential equations (SPDE). In more detail, the authors consider target measures \(\pi\) on a real separable Hilbert space \(H\) which possess a Radon-Nikodym derivative with respect to a Gaussian measure \(\pi_0\) on \(H\) of the form \[ \frac{d\pi}{d\pi_0}=M_\Psi\exp(-\Psi(x)) \] for some real, measurable functional \(\Psi\). Then, a realisable implementation (necessarily in finite dimensions) of the RWM algorithm for \(\pi\) is obtained by applying an RWM algorithm for a projection \(\pi^N\) of \(\pi\) on an \(N\)-dimensional subspace of \(H\). The main result of the paper is that started in stationarity the piecewise constant linear interpolants of the Markov chains in the \(N\)-dimensional subspaces converge for \(N\to\infty\) weakly to the solution of an SPDE in a suitable subspace of \(H\) started at \(\pi\) which is the invariant measure for this SPDE. This is proved under a list of assumptions on the function \(\Psi\) and the trace class covariance operator of the Gaussian measure \(\pi_0\) which according to the authors are satisfied in many applications. The practical implications of this result are that at stationarity the work to explore that state space scales as \(\mathcal{O}(N)\) and the speed at which the invariant measure is explored can be maximized by tuning the acceptance rate probability to \(0.234\) analogous to the known case for measures of product structure.
0 references
Markov chain Monte Carlo
0 references
random walk Metropolis-Hastings
0 references
difffusion limit
0 references
optimal convergence
0 references
scaling limits
0 references
convergence time
0 references
stochastic partial differential equations
0 references
0 references