High performance uncertainty quantification with parallelized multilevel Markov chain Monte Carlo

@article{Seelinger2021HighPU,
  title={High performance uncertainty quantification with parallelized multilevel Markov chain Monte Carlo},
  author={Linus Seelinger and Anne Reinarz and Leonhard Rannabauer and Michael Bader and Peter Bastian and Robert Scheichl},
  journal={Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis},
  year={2021}
}
  • L. Seelinger, A. Reinarz, Robert Scheichl
  • Published 30 July 2021
  • Computer Science
  • Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis
Numerical models of complex real-world phenomena often necessitate High Performance Computing (HPC). Uncertainties increase problem dimensionality further and pose even greater challenges. We present a parallelization strategy for multilevel Markov chain Monte Carlo, a state-of-the-art, algorithmically scalable Uncertainty Quantification (UQ) algorithm for Bayesian inverse problems, and a new software framework allowing for large-scale parallelism across forward model evaluations and the UQ… 

An adaptive strategy for sequential designs of multilevel computer experiments

TLDR
A novel method that combines the multilevel adaptive sequential design of computer experiments (MLASCE) in the framework of Gaussian process (GP) emulators, and makes use of reproducing kernel Hilbert spaces as a new tool for GP approximations of the increments across levels.

Multilevel Delayed Acceptance MCMC

TLDR
A novel Markov chain Monte Carlo method that exploits a hierarchy of models of increasing complexity to efficiently generate samples from an unnormalized target distribution is developed and shows that the algorithm satisfies detailed balance, hence is ergodic for the target distribution.

MUQ: The MIT Uncertainty Quantification Library

TLDR
The MIT Uncertainty Quantification library (MUQ) is a modular software framework for defining and solving uncertainty quantification problems involving complex models.

A general framework for quantifying uncertainty at scale and its application to fusion research

TLDR
The recently developed sensitivity-driven dimension-adaptive sparse grid interpolation strategy exploits the structure of the underlying model to enable efficient and accurate uncertainty quantification and sensitivity analysis at scale and enables studies previously considered infeasible.

References

SHOWING 1-10 OF 39 REFERENCES

Complexity analysis of accelerated MCMC methods for Bayesian inversion

TLDR
This work studies Bayesian inversion for a model elliptic PDE with an unknown diffusion coefficient and proposes and analyzes two strategies for reducing the computational complexity, including a sparse, parametric and deterministic generalized polynomial chaos (gpc) β€˜surrogate’ representation of the forward response map of the PDE over the entire parameter space.

Multilevel Dimension-Independent Likelihood-Informed MCMC for Large-Scale Inverse Problems

TLDR
A non-trivial integration of dimension-independent likelihood-informed (DILI) MCMC and the multilevel MCMC to explore the hierarchy of posterior distributions and suggests a new pooling strategy for allocating computational resources across different levels and constructing Markov chains at higher levels conditioned on those simulated on lower levels.

Multilevel Delayed Acceptance MCMC with an Adaptive Error Model in PyMC3

TLDR
Multilevel Delayed Acceptance with an Adaptive Error Model (AEM) is a novel approach, which alleviates uncertainty Quantification through Markov Chain Monte Carlo problem by exploiting a hierarchy of models, with increasing complexity and cost, and correcting the inexpensive models on-the-fly.

A stochastic collocation approach to Bayesian inference in inverse problems

We present an efficient numerical strategy for the Bayesian solution of inverse problems. Stochastic collocation methods, based on generalized polynomial chaos (gPC), are used to construct a…

Dimension-independent likelihood-informed MCMC

MCMC Methods for Functions: ModifyingOld Algorithms to Make Them Faster

TLDR
An approach to modifying a whole range of MCMC methods, applicable whenever the target measure has density with respect to a Gaussian process or Gaussian random field reference measure, which ensures that their speed of convergence is robust under mesh refinement.

Probabilistic programming in Python using PyMC3

TLDR
This paper is a tutorial-style introduction to PyMC3, a new open source Probabilistic Programming framework written in Python that uses Theano to compute gradients via automatic dierentiation as well as compile probabilistic programs on-the-fly to C for increased speed.

Tsunami modelling with adaptively refined finite volume methods*

TLDR
These issues are discussed in the context of Riemann-solver-based finite volume methods for tsunami modelling in a β€˜wellbalanced’ manner and also apply to a variety of other geophysical flows.

MCMC methods for diffusion bridges

TLDR
The novel algorithmic idea of the paper is that proposed moves for the MCMC algorithm are determined by discretising the SPDEs in the time direction using an implicit scheme, parametrised by θ ∈ [0,1].

Fast and Exact Simulation of Stationary Gaussian Processes through Circulant Embedding of the Covariance Matrix

TLDR
This paper shows that for many important correlation functions in geostatistics, realizations of the associated process over $m+1$ equispaced points on a line can be produced at the cost of an initial FFT of length $2m$ with each new realization requiring an additionalFFT of the same length.