Loading…

The new Frontiers in Computing Systems group,  part of Columbia’s Data Science Institute, is hosting an inaugural full-day symposium, with leading speakers and panelists, to highlight the advances and grand challenges in Big Data infrastructure:  extreme-scale computing systems (hardware, parallel computing, software, databases) and their application to solve diverse cutting-edge problems in climate and ocean science, population-scale biomedical informatics, genomics, materials science, neuroscience, astrophysics and engineering.

The symposium includes an exciting keynote by Ruchir Puri, the chief architect of IBM’s Watson system, on "Engineering the Future of Cognitive Systems."  Other speakers include those developing state-of-art high-performance parallel computers and large-scale Python-based software platforms, as well as experts on computational problems in climate science, astrophysics, and protein folding simulation.

The event will include a keynote talk, lunch, two talk presentation sessions, a networking and poster session, and a panel of experts with audience participation.

 

Sign up or log in to bookmark your favorites and sync them to your phone or calendar.

Talks [clear filter]
Friday, March 24
 

1:15pm

Presentation Session I:  Systems
Session Chair:
Martha Kim, Associate Professor of Computer Science
Columbia University

  1. "Big Data, Streaming Graphs, and the Need for Innovation in Parallel Computer Architecture"
    Peter Kogge
    Chief Scientist and Founder, Emu Technology
    McCourtney Professor of Computer Science and Engineering, University of Notre Dame

    This talk will start with some insights gleaned from looking at real-world big data problems and how one example is affected by architecture. The Emu migrating thread architecture is then introduced and compared. A general template for integrated big graph batch and analytic processing is developed, and key graph operations, especially streaming, listed. The talk then ends with how the Emu architecture meshes well with such a dual-mode computing template.
     
  2. "Why Science Must Lead Computing in the Era of Big Data"
    Peter Wang
    CTO and Co-Founder, Continuum Analytics

    O
    ver the last 30 years, the PC and Internet booms have primarily driven funding and innovation in the for-profit business computing space.  Scientific computing has surely benefitted as an ancillary effect, but major technical computing initiatives (including supercomputers) have been a relatively niche pursuit in the broad technology ecosystem.

    Over the last few years, we've seen concurrent disruptions of hardware, software, networking, data, and algorithms -- all leading to a Cambrian explosion of approaches towards scaling massive computing on breathtakingly vast datasets.  The needs and interests of business computing and scientific simulation are colliding, leading to extremely exciting opportunities for innovation and discovery.

    In this talk, Peter will talk about the opportunity for scientific and technical computing to take the initiative in this unfolding Era of Data.  More importantly, he will discuss why it is of paramount importance for scientists to establish thought leadership not only in technical architecture and design, but also in areas such as reproducibility, data access, and ethics.

Speakers
avatar for Peter Kogge

Peter Kogge

Emu Technology, Chief Scientist and Founder, McCourtney Professor of Computer Science and Engineering, University of Notre Dame
PETER M. KOGGE received his Ph.D. in EE from Stanford in 1973. From 1968 until 1994 he was with IBM's Federal Systems Division, and is an IBM Fellow (1993). In 1994, he joined the University of Notre Dame as first holder of the McCourtney Chair in Computer Science and Engineering... Read More →
avatar for Peter Wang

Peter Wang

CTO and Co-Founder, Continuum Analytics
PETER WANG is the Chief Technology Officer and Co-Founder of Continuum Analytics.  He has been developing commercial scientific computing and visualization software for over 15 years, across a broad variety of areas, including 3D graphics, geophysics, financial risk modeling, large... Read More →


Friday March 24, 2017 1:15pm - 2:15pm
Davis Auditorium 530 West 120th Street New York, NY 10027 USA

3:00pm

Presentation Session II: Applications
Session Chair:
Nicholas Tatonetti, Assistant Professor of Biomedical Informatics
Columbia University

  1.  "Climate Science in the Age of Big Data"
    Gavin Schmidt
    Director, NASA Goddard Institute for Space Studies

    The amount of data from satellite observations, weather model reanalyses and climate model output is increasing rapidly and is now measured in 100’s of Petabytes and will soon reach Exabyte levels. The need to integrate these datasets and to evaluate and improve our models and our predictions has never been more acute, but our ability to query, test, and extract useful information from these datasets is actually decreasing. Neither bandwidth nor computer architectures are keeping up with the data flows and methods required to focus on the process-based diagnostics that are needed to improve models and provide better predictions. It remains an enormous challenge to come up with the next generation of tools and techniques that will be able to tame the climate science big data monster.
  2.  
  3. "How to Make a Galaxy from Scratch:  Challenges in Computational Galaxy Formation"
    Rachel Somerville
    Co-Lead, Galaxy Formation Group, Center for Computational Astrophysics, Simons Foundation
    Downsbrough Chair in Astrophysics, Rutgers University

    Understanding the formation of galaxies in a cosmological context is one of the great challenges of modern astrophysics. Galaxies form within a vast 'cosmic web' of filaments of dark matter, and are shaped by the energy released by the stars, supernovae, and supermassive black holes they harbor. As a result, large-scale numerical simulations of galaxy formation are extremely computationally demanding. They require not only grappling with approximately nine orders of magnitude in dynamic range, but also require the inclusion of a broad range of physical processes. I will discuss some of the recent progress in the area of computational galaxy formation, and some of the challenges that we face in the years to come.

  4. "Exploiting Hardware-Software Co-Design in a Special-Purpose Supercomputer to Speed Up Molecular Dynamics Simulation"
    Mark Moraes
    Head of Engineering, D.E. Shaw Research

    Anton 2 is the second generation of a family of massively-parallel special-purpose supercomputers for molecular dynamics simulations of proteins and other biological macromolecules.  The embedded software that runs on the Anton machines is co-designed with the underlying hardware, allowing the code to exploit unique capabilities of the hardware, and the hardware to make careful trade-offs for maximum performance, while preserving flexibility and programmability.  This closely coupled co-design is a major reason for Anton's large performance improvement -- as much as two orders of magnitude -- over commodity clusters and general-purpose supercomputers. This speedup enables the study of the structural changes underlying biological phenomena that occur on time scales far in excess of those previously accessible to computational study, with the ultimate aim of significantly advancing the process of drug discovery.

Speakers
avatar for Mark Moraes

Mark Moraes

Head of Engineering, D.E. Shaw Research
MARK MORAES is Head of Engineering at D.E. Shaw Research (DESRES), where he both leads and contributes directly to a large part of the engineering and systems development work.  He returned to the D.E. Shaw group in 2004 after spending several years designing systems management products... Read More →
avatar for Gavin Schmidt

Gavin Schmidt

Director, NASA Goddard Institute for Space Studies
GAVIN SCHMIDT is a climate scientist and Director of the NASA Goddard Institute for Space Studies (GISS) in New York, as well as an adjunct researcher at the Columbia University Earth Institute. He works on understanding past, present and future climate change and on the development... Read More →
avatar for Rachel Somerville

Rachel Somerville

Co-Lead, Galaxy Formation Group, Center for Computational Astrophysics, Simons Foundation, Downsbrough Chair in Astrophysics, Rutgers University
RACHEL SOMERVILLE is a Distinguished Professor and holds the George A. and Margaret M. Downsbrough Chair in Astrophysics in the Rutgers University Department of Physics and Astronomy. She is jointly co-leading the Galaxy Simulations group at the Center for Computational Astrophysics... Read More →


Friday March 24, 2017 3:00pm - 4:30pm
Davis Auditorium 530 West 120th Street New York, NY 10027 USA