We use cookies to ensure that we give you the best experience on our website. You can change your cookie settings at any time. Otherwise, we'll assume you're OK to continue.

Durham University

Research & business

Research lectures, seminars and events

The events listed in this area are research seminars, workshops and lectures hosted by Durham University departments and research institutes. If you are not a member of the University, but  wish to enquire about attending one of the events please contact the organiser or host department.


November 2020
October 2020 December 2020
1 2 3 4 5 6 7
8 9 10 11 12 13 14
15 16 17 18 19 20 21
22 23 24 25 26 27 28
29 30

Events for 9 November 2020

Bledar Alexandros Konomi: On the Bayesian Analysis of Multifidelity Computer Models

1:00pm, ZOOM

We propose a multi-fidelity Bayesian emulator for the analysis of computer models when the available simulations are not generated based on hierarchically nested experimental design. The proposed procedure, called Augmented Bayesian Treed Co-Kriging, extends the scope of co-kriging in two major ways. We introduce a binary treed partition latent process in the multifidelity setting to account for non-stationary and potential discontinuities in the model outputs at different fidelity levels. Moreover, we introduce an efficient imputation mechanism which allows the practical implementation of co-kriging when the experimental design is non-hierarchically nested by enabling the specification of semi-conjugate priors. Our imputation strategy allows the design of an efficient RJ-MCMC implementation that involves collapsed blocks and direct simulation from conditional distributions. We develop the Monte Carlo recursive emulator which provides a Monte Carlo proxy for the full predictive distribution of the model output at each fidelity level, in a computationally feasible manner. We also extend our mode for high-dimensional output as well as for high number of observed inputs. The performance of our method is demonstrated on benchmark examples and compared against existing methods. The proposed method is used for the analysis of a large-scale climate modeling application which involves the WRF model as well as ADCIRC storm surge model.

Contact, for more information about this event.

Prof Harold Attridge: Training eyes to see

3:00pm, Zoom

Contact for more information about this event.