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:20210916T132451Z
LOCATION:Jean-Jacques Rousseau
DTSTART;TZID=Europe/Stockholm:20210707T143000
DTEND;TZID=Europe/Stockholm:20210707T150000
UID:submissions.pasc-conference.org_PASC21_sess146_msa226@linklings.com
SUMMARY:Topological Analysis of Simulation Data for Urgent Decision Making
DESCRIPTION:Minisymposium\n\nTopological Analysis of Simulation Data for U
 rgent Decision Making\n\nGuillou, Tierny\n\nNumerical simulations supporti
 ng urgent decision making need to provide detailed predictions of the poss
 ible outcomes of the simulated phenomenon, accompanied with advanced data 
 analytics enabling them to assess the characteristics of each class of out
 come, as well as an estimation of their statistical relevance. This proces
 s, usually referred to as "ensemble simulation", is in this context inhere
 ntly tied to strong time constraints, to allow deciders to act as quickly 
 as possible. This motivates innovative data analysis solutions, capable of
  providing detailed reports as quickly as possible. In this talk, we will 
 describe how we addressed this challenge in the context of the European pr
 oject VESTEC, dedicated to the development of HPC prototypes supporting ur
 gent decision making. In particular, we will present an overview of the re
 sults we obtained in this project for the automatic clustering of ensemble
  members. In this context, we show how topological analysis can be instrum
 ental for computing such a clustering "in-situ", i.e. computing on-the-fly
 , as the simulation is running, with negligible data IO. In particular, a 
 concise topological signature of each ensemble member, namely its Persiste
 nce diagram, is computed in-situ with TTK and Catalyst and stored in a Cin
 ema database. We show how these signatures can be clustered in-situ with o
 ur latest results in topological analysis for the computation of Wasserste
 in barycenters of Persistence diagrams. A complete prototype for the clust
 ering of space-weather simulations will be demonstrated. All the component
 s of this research have been implemented open source and are freely availa
 ble in the Topology ToolKit.\n\nDomain: CS and Math, Emerging Applications
END:VEVENT
END:VCALENDAR