BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Uncertainty Quantification from a Mathematical Perspective - Ralph
   Smith (North Carolina State University)
DTSTART:20180108T113000Z
DTEND:20180108T123000Z
UID:TALK97453@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:From both mathematical and statistical perspectives\, the fund
 amental goal of Uncertainty Quantification (UQ) is to ascertain uncertaint
 ies inherent to parameters\, initial and boundary conditions\, experimenta
 l data\, and models themselves to make predictions with improved and quant
 ified accuracy.&nbsp\; Some factors that motivate recent developments in m
 athematical UQ analysis include the following.&nbsp\; The first is the goa
 l of quantifying uncertainties for models and applications whose complexit
 y precludes sole reliance on sampling-based methods.&nbsp\; This includes 
 simulation codes for discretized partial differential equation (PDE) model
 s\, which can require hours to days to run.&nbsp\; Secondly\, models are t
 ypically nonlinearly parameterized thus requiring nonlinear statistical an
 alysis.&nbsp\; Finally\, there is often emphasis on extrapolatory or out-o
 f-data predictions\; e.g.\, using time-dependent models to predict future 
 events.&nbsp\; This requires embedding statistical models within physical 
 laws\, such as conservation relations\, to provide the structure required 
 for extrapolatory predictions.&nbsp\; Within this context\, the discussion
  will focus on techniques to isolate subsets and subspaces of inputs that 
 are uniquely determined by data.&nbsp\; We will also discuss the use of st
 ochastic collocation and Gaussian process techniques to construct and veri
 fy surrogate models\, which can be used for Bayesian inference and subsequ
 ent uncertainty propagation to construct prediction intervals for statisti
 cal quantities of interest.&nbsp\; The presentation will conclude with dis
 cussion pertaining to the quantification of model discrepancies in a manne
 r that preserves physical structures.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR