BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20241120T082409Z
LOCATION:HG E 1.1
DTSTART;TZID=Europe/Stockholm:20240604T173000
DTEND;TZID=Europe/Stockholm:20240604T180000
UID:submissions.pasc-conference.org_PASC24_sess109_msa121@linklings.com
SUMMARY:Julia-Based Multitask Surrogate Models for Heterogeneous Data Gene
 rated by Physical Models
DESCRIPTION:Minisymposium\n\nKatharine Fisher (Massachusetts Institute of 
 Technology), Michael Herbst (EPFL), and Youssef Marzouk (Massachusetts Ins
 titute of Technology)\n\nPhysical data is increasingly openly accessible t
 hough it may be challenging to definitively rank the accuracy of different
  information sources. We demonstrate that multitask Gaussian process regre
 ssion can leverage “datasets of opportunity” to efficiently construct surr
 ogate models. In particular, we consider training sets constructed from co
 upled-cluster (CC) and density functional theory (DFT) data generated with
  multiple exchange-correlation functional approximations. The cost of CC c
 alculation scales at a rate of N to the power of seven where N is the numb
 er of atoms in the system while DFT demonstrates relatively tractable N cu
 bed scaling. We report that multitask surrogates can predict at CC level a
 ccuracy with a reduction to data generation cost by over an order of magni
 tude. This interdisciplinary effort has been facilitated by Julia packages
  for atomistic computation and for the custom design of optimization and G
 aussian process models. If time permits, we will discuss the extension of 
 our computational models to produce calibrated uncertainty indicators for 
 each prediction.\n\nDomain: Chemistry and Materials, Computational Methods
  and Applied Mathematics\n\nSession Chair: Michael Herbst (EPFL)
END:VEVENT
END:VCALENDAR