BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260202T201804Z
LOCATION:274
DTSTART;TZID=America/Chicago:20251117T115000
DTEND;TZID=America/Chicago:20251117T121000
UID:submissions.supercomputing.org_SC25_sess215_ws_ai4s116@linklings.com
SUMMARY:ROSE: RADICAL Orchestrator for Surrogate Exploration
DESCRIPTION:Aymen Alsaadi (Rutgers University); Tianle Wang (Brookhaven Na
 tional Laboratory); Sumeet Atul Vadhavkar (Rochester Institute of Technolo
 gy); Andrew Park (Rutgers University); Pradeep Bajracharya and Linwei Wang
  (Rochester Institute of Technology); Fanbo Sun (Indiana University); Sudi
 p Seal (Oak Ridge National Laboratory (ORNL)); Vikram Jadhao (Indiana Univ
 ersity); Geoffrey Fox (University of Virginia); and Shantenu Jha (Rutgers 
 University, Princeton University)\n\nScientific computing increasingly use
 s surrogate models to accelerate high-fidelity simulations, enable real-ti
 me predictions, and explore large design spaces. Building surrogates at sc
 ale is challenging: simulations are costly, data generation must be manage
 d, and surrogate learning involves large, heterogeneous, evolving workflow
 s. In active learning, where models guide data acquisition, these challeng
 es intensify due to tight coupling between simulation, inference, and trai
 ning. We present ROSE (RADICAL Orchestrator for Surrogate Exploration), a 
 flexible, portable, and scalable framework supporting the full surrogate m
 odeling lifecycle in HPC environments. ROSE integrates active learning wit
 h scalable orchestration, managing asynchronous execution across diverse r
 esources while minimizing user effort. It supports in-situ/ex-situ workflo
 ws, online/offline training, and adaptive sampling. Applied to three use c
 ases—electrolyte structure extraction, neutron diffraction structure recov
 ery, and colloid phase classification—ROSE sustains high throughput with l
 ow overhead on Polaris, Perlmutter, and Delta, achieving 4–8× end-to-end s
 peedups, with asynchronous orchestration delivering 1.5–3× gains over sync
 hronous baselines.\n\nRecording: Livestreamed, Recorded\n\nRegistration Ca
 tegory: Technical Program Reg Pass, Workshop Reg Pass\n\nSession Chairs: G
 okcen Kestor (Barcelona Supercomputing Center (BSC); University of Califor
 nia, Merced); Dong Li (University of California, Merced); and Murali Emani
  (Argonne National Laboratory (ANL))\n\n
END:VEVENT
END:VCALENDAR