Surrogate models for UQ in complex systems

Workshop theme

Our modern treatment of predicting the behavior of physical and engineering problems relies on approximating solutions in terms of physical and

stochastic domains, particularly in the case of uncertainty quantification (UQ), where the input data (coefficients, forcing terms, boundary conditions, geometry, etc) are affected by randomness. For higher resolution and accuracy, simulations must increase the number of deterministic and stochastic variables, and expend more effort resolving smooth or even discontinuous behavior within each individual component. Simulating the entire complex system at the level of fidelity that truly predicts the solution, remains elusive outside of idealized situations. To combat this resulting explosion in computational effort, surrogate models are employed to quickly and efficiently predict the input-output map of many complex simulations. As such, this workshop will be organized around a set of themes in which surrogate models have become an essential tool in enabling and accelerating UQ for a variety of problems, including:

1. High-dimensional stochastic systems;

2. Inference, optimization, and control in the presence of uncertainty;

3. Stochastic multi-scale and multi-physics problems; and

4. Design of physical and computational experiments;

However, constructing accurate surrogates for such problems and computing their predictions along with the uncertainties they introduce, constitute additional challenges. This workshop will explore a variety of approaches that have been developed to address such challenges, including Gaussian processes, sparse interpolation and projection, discrete least squares, compressed sensing, dimension reduction, multi-fidelity formulations, Bayesian inference, homogenization techniques, and atomistic to continuum coupling methods.

Applications for this workshop are now closed

The Institute kindly requests that any papers published as a result of this programme’s activities are credited as such. Please acknowledge the support of the Institute in your paper using the following text:

The author(s) would like to thank the Isaac Newton Institute for Mathematical Sciences, Cambridge, for support and hospitality during the programme Surrogate models for UQ in complex systems, where work on this paper was undertaken. This work was supported by EPSRC grant EP/K032208/1 and EP/R014604/1.

Timetable

Monday 5th February 2018
10:50 to 11:20	Registration & Morning Coffee	No Room Required
11:20 to 11:30	Welcome from David Abrahams (INI Director)	No Room Required
11:30 to 12:30	Catherine Powell University of Manchester Adaptive Stochastic Galerkin Finite Element Approximation for Elliptic PDEs with Random Coefficients	Room 1
12:30 to 13:30	Lunch @ Churchill College	No Room Required
13:30 to 14:30	Michael Goldstein Durham University Emulation for model discrepancy	Room 1
14:30 to 15:30	Ralph Smith North Carolina State University Active Subspace Techniques to Construct Surrogate Models for Complex Physical and Biological Models	Room 1
15:30 to 16:00	Afternoon Tea	No Room Required
16:00 to 17:00	Christoph Schwab ETH Zürich Domain Uncertainty Quantification	Room 1
17:00 to 18:00	Welcome Wine Reception at INI	No Room Required

Tuesday 6th February 2018
09:00 to 10:00	Hoang Tran Oak Ridge National Laboratory Recovery conditions of compressed sensing approach to uncertainty quantification	Room 1
10:00 to 11:00	Maurizio Filippone Random Feature Expansions for Deep Gaussian Processes	Room 1
11:00 to 11:30	Morning Coffee	No Room Required
11:30 to 12:30	Lorenzo Tamellini Università degli Studi di Pavia Multi-Index Stochastic Collocation (MISC) for Elliptic PDEs with random data	Room 1
12:30 to 13:30	Lunch @ Churchill College	No Room Required
13:30 to 14:30	John Paul Gosling University of Leeds Modelling discontinuities in simulator output using Voronoi tessellations	Room 1
14:30 to 15:30	Guannan Zhang Oak Ridge National Laboratory A domain-decomposition-based model reduction method for convection-diffusion equations with random coefficients	Room 1
15:30 to 16:00	Afternoon Tea	No Room Required
16:00 to 17:00	Poster Session	No Room Required

Wednesday 7th February 2018
09:00 to 10:00	Martin Eigel Weierstraß-Institut für Angewandte Analysis und Stochastik Aspects of adaptive Galerkin FE for stochastic direct and inverse problems	Room 1
10:00 to 11:00	Elaine Spiller None / Other Emulators for forecasting and UQ of natural hazards	Room 1
11:00 to 11:30	Morning Coffee	No Room Required
11:30 to 12:30	Panel comparisons: Challenor, Ginsbourger, Nobile, Teckentrup and Beck	Room 1
12:30 to 13:30	Lunch @ Churchill College	No Room Required
13:30 to 17:00	Free Afternoon	No Room Required
19:30 to 22:00	Formal Dinner at Trinity College	No Room Required

Thursday 8th February 2018
09:00 to 10:00	Ben Adcock Simon Fraser University Polynomial approximation of high-dimensional functions on irregular domains	Room 1
10:00 to 11:00	Christine Shoemaker National University of Singapore Deterministic RBF Surrogate Methods for Uncertainty Quantification, Global Optimization and Parallel HPC Applications	Room 1
11:00 to 11:30	Morning Coffee	No Room Required
11:30 to 12:30	Aretha Teckentrup University of Edinburgh Surrogate models in Bayesian Inverse Problems	Room 1
12:30 to 13:30	Lunch @ Churchill College	No Room Required
13:30 to 14:30	David Ginsbourger None / Other; Universität Bern Positive definite kernels for deterministic and stochastic approximations of (invariant) functions	Room 1
14:30 to 15:30	Raul Fidel Tempone King Abdullah University of Science and Technology (KAUST) Uncertainty Quantification with Multi-Level and Multi-Index methods	Room 1
15:30 to 16:00	Afternoon Tea	No Room Required
16:00 to 17:00	Maria Adamou University of Southampton Bayesian optimal design for Gaussian process model	Room 1

Friday 9th February 2018
09:00 to 10:00	Olivier Roustant Mines Saint-Étienne Group covariance functions for Gaussian process metamodels with categorical inputs	Room 1
10:00 to 11:00	Daniel Williamson University of Exeter Nonstationary Gaussian process emulators with covariance mixtures	Room 1
11:00 to 11:30	Morning Coffee	No Room Required
11:30 to 12:30	Oliver Ernst Technische Universität Chemnitz High-Dimensional Collocation for Lognormal Diffusion Problems	Room 1
12:30 to 13:30	Lunch @ Churchill College	No Room Required
13:30 to 14:30	Robert Gramacy Virginia Polytechnic Institute and State University Replication or exploration? Sequential design for stochastic simulation experiments	Room 1
14:30 to 15:30	Future directions panel	Room 1

Uncertainty quantification for complex systems: theory and methodologies

Surrogate models for UQ in complex systems

Organisers

Serge Guillas University College London
Lindsay Lee University of Leeds
Fabio Nobile EPFL - Ecole Polytechnique Fédérale de Lausanne
Clayton Webster University of Tennessee; Oak Ridge National Laboratory

Speakers

Maria Adamou University of Southampton
Ben Adcock Simon Fraser University
Martin Eigel Weierstraß-Institut für Angewandte Analysis und Stochastik
Oliver Ernst Technische Universität Chemnitz
Maurizio Filippone
David Ginsbourger None / Other; Universität Bern
Michael Goldstein Durham University
John Paul Gosling University of Leeds
Robert Gramacy Virginia Polytechnic Institute and State University
Catherine Powell University of Manchester
Olivier Roustant Mines Saint-Étienne
Christoph Schwab ETH Zürich
Christine Shoemaker National University of Singapore
Ralph Smith North Carolina State University
Elaine Spiller None / Other
Lorenzo Tamellini Università degli Studi di Pavia
Aretha Teckentrup University of Edinburgh
Raul Fidel Tempone King Abdullah University of Science and Technology (KAUST)
Hoang Tran Oak Ridge National Laboratory
Daniel Williamson University of Exeter
Guannan Zhang Oak Ridge National Laboratory

Show All

Participants

Maria Adamou University of Southampton
Ben Adcock Simon Fraser University
Alexander Archibald University of Cambridge
Marco Ballesio King Abdullah University of Science and Technology (KAUST)
Joakim Beck King Abdullah University of Science and Technology (KAUST)
Juan Pablo Carbajal ETH Zürich
Peter Challenor University of Exeter
Xin Chen Cranfield University
Samuel Coveney University of Sheffield
Daan Crommelin Centrum voor Wiskunde en Informatica (CWI); Universiteit van Amsterdam
Adam Crowder University of Manchester
Martin Eigel Weierstraß-Institut für Angewandte Analysis und Stochastik
Simen Eldevik None / Other
Oliver Ernst Technische Universität Chemnitz
Ionut-Gabriel Farcas Technische Universität München
Maurizio Filippone
Sanmitra Ghosh University of Oxford
David Ginsbourger None / Other; Universität Bern
Mark Girolami Imperial College London; The Alan Turing Institute
Michael Goldstein Durham University
John Paul Gosling University of Leeds
Robert Gramacy Virginia Polytechnic Institute and State University
Serge Guillas University College London
Max Gunzburger Florida State University
Tobias Jahnke Karlsruhe Institute of Technology (KIT)
Jill Johnson University of Leeds
Louise Kimpton University of Exeter
Jonas Latz Technische Universität München
Lindsay Lee University of Leeds
Hyung-Chun Lee Ajou University
Laura Mansfield University of Reading; Imperial College London
Theodoros Mathikolonis University College London
Gildas Mazo INRA - Institut National de la Recherche Agronomique
Hossein Mohammadi University of Exeter
PAVANAKUMAR MOHANAMURALY Queen Mary University of London
Franck Monmont Schlumberger Cambridge Research
Helena Nandi Formentin Universidade Estadual de Campinas; Durham University
Fabio Nobile EPFL - Ecole Polytechnique Fédérale de Lausanne
Tony O'Hagan University of Sheffield
Akwum Onwunta Max Planck Institute Magdeburg
Oluwole Oyebamiji University of Newcastle upon Tyne
Ludger Paehler Munich University of Technology
Jim Parr McLaren Applied Technologies
Christopher Pope University of Leeds
Catherine Powell University of Manchester
Olivier Roustant Mines Saint-Étienne
Benjamin Sanderse Centrum voor Wiskunde en Informatica (CWI)
Claudia Schillings University of Warwick
Christoph Schwab ETH Zürich
Pranay Seshadri University of Cambridge
Christine Shoemaker National University of Singapore
David Silvester University of Manchester
Ralph Smith North Carolina State University
Elaine Spiller None / Other
Björn Sprungk Universität Mannheim
Benny Stein Karlsruhe Institute of Technology (KIT)
Jonathan Sykes University of Sheffield
Lorenzo Tamellini Università degli Studi di Pavia
Aretha Teckentrup University of Edinburgh
Raul Fidel Tempone King Abdullah University of Science and Technology (KAUST)
Hoang Tran Oak Ridge National Laboratory
Giang Tran None / Other
Hemant Tyagi University of Edinburgh
Clayton Webster University of Tennessee; Oak Ridge National Laboratory
Daniel Williamson University of Exeter
Soeren Wolfers King Abdullah University of Science and Technology (KAUST)
Henry Wynn London School of Economics
Guannan Zhang Oak Ridge National Laboratory