Guided proposals for simulating multi-dimensional diffusion bridges
From MaRDI portal
Abstract: A Monte Carlo method for simulating a multi-dimensional diffusion process conditioned on hitting a fixed point at a fixed future time is developed. Proposals for such diffusion bridges are obtained by superimposing an additional guiding term to the drift of the process under consideration. The guiding term is derived via approximation of the target process by a simpler diffusion processes with known transition densities. Acceptance of a proposal can be determined by computing the likelihood ratio between the proposal and the target bridge, which is derived in closed form. We show under general conditions that the likelihood ratio is well defined and show that a class of proposals with guiding term obtained from linear approximations fall under these conditions.
The problem of Monte Carlo simulation of a multidimensional diffusion bridge, i.e., a diffusion conditioned to be at a specific point at a specific time in future, is considered. The actual conditional law involves the unknown density, so is unavailable for forming the proposal distribution. In view of this, a scheme is developed in order to derive a proposal distribution, in terms of an auxiliary diffusion, that can be shown to be absolutely continuous with respect to the target distribution under a set of assumptions. These assumptions are shown to hold for a class of linear diffusions. Numerical evidence in support of the proposed scheme is presented.
Recommendations
- Simulation of Multivariate Diffusion Bridges
- Exact simulation for diffusion bridges: an adaptive approach
- A piecewise deterministic Monte Carlo method for diffusion bridges
- MCMC METHODS FOR DIFFUSION BRIDGES
- Simulation of diffusions with boundary conditions
- Simulating diffusion processes in discontinuous media: benchmark tests
- A local interaction simulation approach to the solution of diffusion problems
- Incorporating Diffusion in Complex Geometries into Stochastic Chemical Kinetics Simulations
- Multi-grid methods for steady state diffusion in random media
- scientific article; zbMATH DE number 1830910
Cited in
(43)- Improved bridge constructs for stochastic differential equations
- Nonparametric Bayesian inference for reversible multidimensional diffusions
- Augmented pseudo-marginal Metropolis-Hastings for partially observed diffusion processes
- Bridge simulation and metric estimation on Lie groups
- Nonparametric statistical inference for drift vector fields of multi-dimensional diffusions
- Continuous-discrete smoothing of diffusions
- Efficient sampling of conditioned Markov jump processes
- Bayesian inference for Markov jump processes with informative observations
- Bayesian estimation of discretely observed multi-dimensional diffusion processes using guided proposals
- A geometric framework for stochastic shape analysis
- Nonparametric Bayesian methods for one-dimensional diffusion models
- The computational cost of blocking for sampling discretely observed diffusions
- SPDE bridges with observation noise and their spatial approximation
- Simulation of elliptic and hypo-elliptic conditional diffusions
- Sequential Monte Carlo with Highly Informative Observations
- Sampling conditioned diffusions
- Importance sampling techniques for estimation of diffusion models
- Score-Based Parameter Estimation for a Class of Continuous-Time State Space Models
- Online Smoothing for Diffusion Processes Observed with Noise
- Diffusion bridges for stochastic Hamiltonian systems and shape evolutions
- Bayesian estimation of incompletely observed diffusions
- Stability of sampling proposals for reducible diffusions over large time intervals
- Forward-reverse expectation-maximization algorithm for Markov chains: convergence and numerical analysis
- Simulation of Multivariate Diffusion Bridges
- Bridge.jl
- Conditioning continuous-time Markov processes by guiding
- An efficient method to simulate diffusion bridges
- Statistical inference for stochastic differential equations
- Stochastic shape analysis
- Inference for partially observed Riemannian Ornstein-Uhlenbeck diffusions of covariance matrices
- Simulation-based inference for stochastic nonlinear mixed-effects models with applications in systems biology
- Conditioning surface shape processes with neural operators
- Bayesian estimation of discretely observed diffusion processes using Wiener chaos expansion
- Guided smoothing and control for diffusion processes
- Simulating diffusion bridges with score matching
- Statistical algorithms for low-frequency diffusion data: a PDE approach
- On a class of exponential changes of measure for stochastic PDEs
- Nonparametric Bayesian estimation in a multidimensional diffusion model with high frequency data
- Controlled stochastic processes for simulated annealing
- Bayesian inference for non-synchronously observed diffusions
- Solution of stochastic boundary value problem in SDE using transition probability density function decomposition
- Simulating diffusion bridges using the Wiener chaos expansion
- Simulation of infinite-dimensional diffusion bridges
This page was built for publication: Guided proposals for simulating multi-dimensional diffusion bridges
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2405130)
