BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Deterministic RBF Surrogate Methods for Uncertainty Quantification
 \, Global Optimization and Parallel HPC Applications - Christine Shoemaker
  (National University of Singapore)
DTSTART:20180208T100000Z
DTEND:20180208T110000Z
UID:TALK100204@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:<span>Co-author: Antoine Espinet		(Cornell University)        
 <br></span><br>This talk will describe general-purpose algorithms for glob
 al optimization These algorithms can be used to estimate model parameters 
 to fit complex simulation models to data\, to select among alternative opt
 ions for design or management\, or to quantify model uncertainty. In gener
 al the numerical results indicate these algorithms do very well in compari
 son to alternatives\, including Gaussian Process based approaches.. Prof. 
 Shoemaker&rsquo\;s group has developed open source (free) PySOT optimizati
 on software that is available online (18\,000 downloads) .  The algorithms
  can be run in serial or parallel. The focus of the talk will be on SOARS\
 , an Uncertainty Quantification  method for  using optimization-based samp
 ling to build a surrogate likelihood function followed by additional sampl
 ing The algorithms builds a surrogate approximation of the likelihood func
 tion based on simulations done during the optimization search. Then MCMC i
 s performed by evaluating the surrogate likelihood function rather than th
 e original expensive-to-evaluate function.  Numerical results indicate the
  SOARS algorithm is very accurate when compared to the posterior densities
  computed when using the expensive  exact likelihood function.   I also di
 scuss an application to a model of the underground movement of a plume of 
 geologically sequestered carbon dioxide.   The uncertainty in the paramete
 r values obtained from the MCMC analysis on the surrogate likelihood funct
 ion can be used to assess alternative strategies for identifying a cost-ef
 fective plan that will most efficiently give a reliable forecast of a carb
 on dioxide underground plume. This includes joint work with David Ruppert\
 , Antoine Espinet\, Nikolay Bliznyuk\, and Yilun Wang. <br><br>Related Lin
 ks<ul><li><a target="_blank" rel="nofollow" href="http://www-old.newton.ac
 .uk/cgi/https%3A%2F%2Fwww.isem.nus.edu.sg%20%20%20%20%20%20%20">https://ww
 w.isem.nus.edu.sg       </a> - NUS ISEM Department  website</li><li><a tar
 get="_blank" rel="nofollow" href="http://www-old.newton.ac.uk/cgi/https%3A
 %2F%2Fsites.google.com%2Fsite%2Fshoemakernusgroup%2Fhome">https://sites.go
 ogle.com/site/shoemakernusgroup/home</a> - Shoemaker group page</li></ul>
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR
