BEGIN:VCALENDAR
CALSCALE:GREGORIAN
VERSION:2.0
METHOD:PUBLISH
PRODID:-//Drupal iCal API//EN
X-WR-TIMEZONE:America/New_York
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
DTSTART:20070311T020000
TZNAME:EDT
TZOFFSETTO:-0400
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
DTSTART:20071104T020000
TZNAME:EST
TZOFFSETTO:-0500
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
SEQUENCE:1
X-APPLE-TRAVEL-ADVISORY-BEHAVIOR:AUTOMATIC
UID:244041
DTSTAMP:20261009T104752Z
DTSTART;TZID=America/New_York:20261030T140000
DTEND;TZID=America/New_York:20261030T145000
URL;TYPE=URI:https://www.wpi.edu/news/calendar/events/harold-j-gay-lecture-
 mathematical-sciences-department-george-karniadakis
SUMMARY:Harold J. Gay Lecture Mathematical Sciences Department - George Kar
 niadakis
DESCRIPTION:Mathematical Sciences Department\nHarold J. Gay Lecture\nTitle:
  Agentic Scientific Machine Learning\nSpeaker: Prof. George Karniadakis\nA
 bstract: Scientific Machine Learning (SciML) integrates data-driven infere
 nce with physical modeling to solve complex problems in science and engine
 ering. However\, the design of SciML architectures\, loss formulations\, a
 nd training strategies remains an expert-driven research process\, requiri
 ng extensive experimentation and problem-specific insights. We introduce A
 THENA\, a collaborative multi-agent system in which about 20 specialized A
 I agents collaborate to propose\, critique\, and refine SciML solutions th
 rough structured reasoning and iterative evolution. The framework integrat
 es structured debate\, retrieval-augmented method memory\, and ensemble-gu
 ided evolutionary search\, enabling the agents to generate and assess new 
 hypotheses about architectures and optimization procedures. Across physics
 -informed learning and operator learning tasks\, the framework discovers s
 olution methods that outperform single agent and human-designed baselines 
 by orders of magnitude in error reduction. The agents produce novel strate
 gies -- including adaptive mixture-of-expert architectures\, decomposition
  based PINNs\, and physics-informed operator learning models -- that do no
 t appear explicitly in the curated knowledge base. They also prove error e
 stimates\, universal approximation theorems\, and discover singularities i
 n PDEs. These results show that collaborative reasoning among AI agents ca
 n yield emergent methodological innovation\, suggesting a path toward scal
 able\, transparent\, and autonomous discovery in scientific computing.
END:VEVENT
END:VCALENDAR