Accelerating Physics with ML@MIT

US/Eastern
Erik Katsavounidis (MIT), Michael Coughlin (University of Minnesota), Philip Coleman Harris (Massachusetts Inst. of Technology (US))
Description
Brain AI Rendering

 

The network of gravitational-wave detectors is rapidly approaching the next observing run (O4) in Spring 2023. Together with that, the prospects for multi-messenger observations in association with gravitational waves are re-opening all across the electromagnetic spectrum and with neutrinos. At the same time, the Large Hadron Collider(LHC) has just started the operations of Run 3. With the first year of the LHC running done, the next two years bode well to accumulate a substantial amount of data, more than doubling the existing dataset. 

This workshop is aiming to bring together developers and stakeholders with an interest in fully integrating machine learning-based tools in gravitational-wave physics, multi-messenger astrophysics, and particle physics from experiment to physics analyses and results.

Part of the presentations and discussions will focus on sharing experiences and challenges in applying machine learning algorithms to real experimental/observational data, including aspects of real-time implementation and the use of heterogeneous computing. We plan to devote time to tutorials on the use of machine learning-based algorithms in gravitational waves, optical astronomy, and particle physics. Additionally, we will go over examples of hardware acceleration within heterogeneous computing systems and their future large scale, deployment within next generation physics experiments. The ultimate goal of the workshop is to establish a clear pipeline toward the integration of Machine Learning within scientific computing in heterogeneous systems and to establish a clear set of goals for the large-scale adoption of these systems.   

The workshop will commence with hackathon sessions at 13:00 on January 28, 2023 and proceed with such sessions through the end of the day on January 29, 2023. These hackathon sessions will take place at the LIGO Laboratory at MIT, building NW22, room 268. This is located at 185 Albany Street in Cambridge, MA 02139. Access to the building is restricted to holders of a pass issued by MIT-- we have obtained such pass for all event participants. You can download it on your cellphone from here.

Presentation and discussion sessions will follow on January 30, 31 and through mid-day on February 1. These sessions will take place at the MIT Kavli Institute for Astrophysics and Space Research ("MKI"). It is located at 70 Vassar Street in Cambridge, MA 02139. We will be using the Marlar Lounge  (37-252) in the second floor of this building. The building is accessible to all (no special passes required).

The workshop will close with another set of hackathon sessions from mid-day February 1 through the end of the day on February 2.  These hackathon sessions will return to the LIGO Laboratory at MIT, building NW22, room 268 at 185 Albany Street in Cambridge, MA 02139. You will need a separate MIT pass to access the building on these two days;  you can download it on your cellphone from here.

Registration
Attendance (Virtual or In person)
Zoom Meeting ID
63451873099
Host
Philip Coleman Harris
Useful links
Join via phone
Zoom URL
    • 13:00 14:30
      Workshop Hackathon #1

      ML infrastructure and code tutorials, debugging and deployment

      Convener: Alec Gunny (MIT)
    • 14:30 15:00
      Coffee Break 30m
    • 15:00 17:00
      Workshop Hackathon #2
      Convener: Alec Gunny (MIT)
    • 09:00 10:30
      Workshop Hackathon #3
      Convener: Alec Gunny (MIT)
    • 10:30 11:00
      Coffee Break 30m
    • 11:00 12:30
      Workshop Hackathon #4
      Convener: Alec Gunny (MIT)
    • 12:30 14:00
      Lunch Break 1h 30m
    • 14:00 15:30
      Workshop Hackathon #5
      Convener: Alec Gunny (MIT)
    • 15:30 16:00
      Coffee Break 30m
    • 16:00 17:30
      Workshop Hackathon #6
      Convener: Alec Gunny (MIT)
    • 09:00 10:30
      Workshop Discussion #1: Introductions and software ecosystems for machine learning

      Software ecosystems for machine learning

      Convener: Erik Katsavounidis (MIT)
    • 10:30 11:00
      Coffee Break 30m
    • 11:00 12:30
      Workshop Discussion #2: Machine learning applications in gravitational-wave searches

      GW Overview

      Convener: Michael Coughlin
    • 12:30 14:00
      Lunch Break 1h 30m
    • 14:00 15:30
      Workshop Discussion #3: Machine learning in electromagnetic astronomy

      EM overview

      Convener: Daniel Muthukrishna
      • 14:00
        The Challenges of Identification, Classification and Inference for Time-domain Astrophysics with Big Data 30m
        Speaker: Ashley Villar
      • 14:30
        Multi-messenger astronomy infrastructure for optical follow-ups 30m
        Speaker: Konstantin Malanchev
      • 15:00
        Challenges for identifying and following-up the next kilonova 30m
        Speaker: Niharika Sravan
    • 15:30 16:00
      Coffee Break 30m
    • 16:00 17:30
      Workshop Discussion #4: Applications in High Energy Physics

      HEP Overview

      Convener: Shih-Chieh Hsu (University of Washington Seattle (US))
      • 16:00
        Software Infrastructure Challenges for IaaS with HEP (20'+10') 35m
        Speaker: Kevin Pedro (Fermi National Accelerator Lab. (US))
      • 16:35
        SONIC Inference as a Service for Point Clouds 25m
        Speaker: Yongbin Feng (Fermi National Accelerator Lab. (US))
      • 17:00
        ML GPU deployments for Neutrino Reconstruction 30m
        Speaker: Tingjun Yang (Fermi National Accelerator Lab. (US))
    • 09:00 10:30
      Workshop Discussion #5: Anomaly Detection applications and challenges

      Anomaly detection across all fields

      Convener: Deep Chatterjee
    • 10:30 11:00
      Coffee Break 30m
    • 11:00 12:30
      Workshop Discussion #6: Hardware acceleration

      HW acceleration

      Convener: Philip Coleman Harris (Massachusetts Inst. of Technology (US))
      • 11:00
        Point-Voxel CNN 30m
        Speaker: Zhijian Liu
      • 11:30
        Sparse GPU implementations 30m
        Speaker: Haotian Tang
      • 12:00
        Computing at the Massachusetts Green High Performance Center 30m
        Speaker: Chris Hill
    • 12:30 14:00
      Lunch Break 1h 30m
    • 14:00 15:30
      Workshop Discussion #7: HPCs

      Synergies, collaborations, other fields

      Convener: Philip Coleman Harris (Massachusetts Inst. of Technology (US))
      • 14:00
        ML Workflows on NSERC 30m
        Speaker: Steven Farrell (Lawrence Berkeley National Laboratory)
      • 14:30
        ML Algorithms at SDSC/NRP 20m
        Speaker: Frank Wurthwein (UCSD)
      • 14:50
        HPC Discussion 30m
    • 15:30 16:00
      Coffee Break 30m
    • 16:00 17:30
      Workshop Discussion #8: Break out session and white paper drafting

      Break out sessions for drafting report

      Conveners: Daniel Muthukrishna, Deep Chatterjee, Patrick McCormack
    • 09:00 10:45
      Workshop Discussion #9: Multidisciplinary opportunities

      reports from break out session

      Convener: Linqing Wen
      • 09:00
        Machine learning applications in Icecube 20m
        Speaker: Josh Peterson
      • 09:20
        Pointing to a Supernova 20m
        Speaker: Kate Scholberg
      • 09:40
        Binary neutron star early warning detection using ML 15m
        Speaker: Alistair Mcleod
      • 09:55
        Localization and waveform extraction for compact binary coalescences 15m
        Speaker: Chayan Chatterjee
      • 10:10
        Unsupervised Learning for TESS Variable Stars 20m
        Speaker: Jeroen Audenaert
      • 10:30
        Contextualized machine learning for applied data science 15m
        Speaker: Ben Lengerich
    • 10:45 11:30
      Coffee Break 45m
    • 11:30 12:30
      Workshop Discussion #10: Summaries and Actions out of the workshop discussions

      Workshop summaries and to-do's

      Convener: Erik Katsavounidis
    • 12:30 14:00
      Lunch Break 1h 30m
    • 14:00 15:30
      Workshop Hackathon #7
      Convener: Deep Chatterjee (MIT)
    • 15:30 16:00
      Coffee Break 30m
    • 16:00 17:30
      Workshop Hackathon #8
      Convener: Deep Chatterjee (MIT)
    • 09:00 10:30
      Workshop Hackathon #9
      Convener: Deep Chatterjee (MIT)
    • 10:30 11:00
      Coffee Break 30m
    • 11:00 12:30
      Workshop Hackathon #10
      Convener: Deep Chatterjee (MIT)
    • 12:30 14:00
      Lunch Break 1h 30m
    • 14:00 15:30
      Workshop Hackathon #11
      Convener: Deep Chatterjee (MIT)
    • 15:30 16:00
      Coffee Break 30m
    • 16:00 17:30
      Workshop Hackathon #12
      Convener: Deep Chatterjee (MIT)