Broadening the Scope and Impact of escience. Frank Seinstra. Director escience Program Netherlands escience Center

Transcription

1 Broadening the Scope and Impact of escience Frank Seinstra Director escience Program Netherlands escience Center

2 Big Science & ICT

3 Big Science Today s Scientific Challenges are Big in many ways: Big Data Big Compute Big Analytics Big Budgets Big, Multidisciplinary Research Teams, Bruce Hevly, 1992 Big Scientific Challenges often driven by Big Societal Challenges Society increasingly demands science to provide solutions

4 Example (Societal): Circular Economy Industrial system restorative by design Renewable everything, Not: Take, Make, Dispose Big challenges in a.o.: Food Research / Green Genetics Advanced Materials Chemistry Logistics,

5 frequency Example (Scientific): Radio Astronomy Square Kilometer Array time

6 ICT / ICT-science ICT / ICT-Science required in vast majority Recent analysis: 15%-85% ICT in over 90% of large projects Drivers: More sensors / instruments More (varied and unstructured) data Increased distribution & heterogeneity Improved availability / access to HPC, high-end networking, data mgmt., e-infrastructure commoditizes (domain scientists agnostic )

7 escience

8 escience Bridges the gap between ICT- / Computer / Data Science AND Domain science (incl. industry) and ultimately also between domains! ICT Domain Bridge-idea simple & clear, but when on the bridge: look both ways! demand-driven aspect must be developed further

9 Accelerated escience In a perfect world: New demands from domain scientists drive new technological developments AND Domain Demand-driven technology-push ICT New technological developments generate new scientific ideas/demands Technology-driven demand-push A self-stimulating loop

10 But Beware! ICT-Science: ICT is everywhere You cannot make progress without us But no real incentives to truly partner with domain scientists Domain Science: We already use computers We have made progress for years / decades / centuries without you But.. often unaware of true potential of advanced ICT Domain scientists and ICT often competing for same funding

11 Intermezzo: A Recent Meeting Royal Dutch Academy of Sciences Dream Infrastructure (vision 2025) e.g. vessels for oceanography, telescopes, polar region Mars Rover, 100M+ EUR each Note: all have ICT components, but barely mentioned ICT Scientists Dreaming of 100M EUR e-infrastructure AND sustained funding While obviously domain-overarching, audience not at all convinced!

12 In other words Despite all good intentions, we re not there yet escience MUST involve both domain science & ICT science Positive trend at e.g. IEEE escience 2015 now also (more) domain scientists in PC focus day on Environmental Computing Balance what should be and what is achievable Period of mind-set change

13 Examples at NLeSC

14 Calls (1) 2015 Calls for Proposals Accelerating Scientific Discovery ICT Science Domain Science 5x 500K projects 50% in kind NLeSC funding Proposals must be driven by: domain questions PI from domain ICT-science may be involved

15 Calls (2) 2015 Calls for Proposals Disruptive Technologies ICT Science Domain Science 2x 500K projects 50% in kind NLeSC funding Proposals must be driven by: technological questions PI from ICT-science Domain use cases required

16 Projects (1): Computational Astrophysics Challenge: Understanding the Fundamental Physics of the Universe Prof. Simon Portegies Zwart University Leiden / Leiden Observatory Dr. Niels Drost Netherlands escience Center Demonstrated live at SC, Seattle, USA

17 Projects (1): Computational Astrophysics The AMUSE system (Leiden University) gravitational dynamics Early Star Cluster Evolution, including gas Most efficient architecture per model: stellar evolution AMUSE hydrodynamics Gravitational dynamics: GPU / GPU-cluster radiative transport Stellar evolution: Cluster / Cloud Hydro-dynamics, Radiative transport: Supercomputer Goals: Apply Jungle Computing (heterogeneous, distributed) Run simulations at unprecedented scale (new science)

18 Projects (1): Computational Astrophysics Demonstrated live at SC, Seattle, USA and now also by Prof. Portegies Zwart on a daily basis ( infra-agnostic )

19 Projects (2): Climate Research Challenge: Understand Impact of Melting Ice Sheets on Climate Change Prof. Henk Dijkstra University Utrecht / IMAU Dr. Jason Maassen Netherlands escience Center Demonstrated live at GLIF, Chicago, USA

20 Projects (2): Climate Research The CPL system or: The Community Earth System Model (CESM) Most efficient architecture per model: Ocean, Sea-ice: GPU / GPU-cluster Atmosphere, Land-vegetation: Cluster / Cloud / Supercomputer Goals: Apply Jungle Computing (heterogeneous, distributed) Run simulations at unprecedented level of detail (new science) landvegetation Now also: cross-domain software re-use & knowledge exchange! atmosphere CPL ocean sea-ice

21 Astrophysics meets Climate Research gravitational dynamics atmosphere Model 1 stellar evolution AMUSE landvegetation hydrodynamics OMUSE sea-ice Model 4 anymuse Model 2 radiative transport ocean Model 3 Apply benefits of AMUSE to CESM : OMUSE Interestingly: much of first phase of work without NLeSC or ICT-science! New Science: Climate Research: (easy) integration of ocean models with coastal models Astrophysics: atmospheric modeling on exo-planets

22 Enlighten Your Research Prize (SC 13) 10G 10G STAMPEDE (USA) EMERALD (UK) CARTESIUS (NLD) 10G SUPERMUC (GER)

23 EU / International IS-ENES / IS-ENES2 Distributed infrastructure of models, model data and meta data of the European Network for Earth System Modeling Enhance the development of Earth System Models for the understanding of climate variability and change Support high-end-simulations enabling a better understanding and prediction of climate variations and change Facilitate the application of Earth System Model simulations to better predict and understand climate change impact on society Partners, a.o: UK MetOffice, STFC, KNMI, DKRZ, MPI-Met,

24 Projects (3): Via Appia (Archaeology) Challenge: Understand ancient objects & their relationship to past landscapes wide variety of expertise: visualization, data mgmt, analytics,

25 Via Appia Via Appia Project Leader (Maurice de Kleijn, VU University): Internationally, I am now considered a leader in my field I now work on research questions I could not even have imagined a year ago => Demand-driven technology-push has led to new scientific ideas (and: demands) Methods & Tools created in Via Appia not unique for archaeology Potential for re-use explored Urban Development, Water Management (e.g. height maps),

26 Challenges for PLAN-E (Wilco Hazeleger)

27 Challenges for PLAN-E (1) Work as a consortium to set escience on the international agenda Collaborate in projects (not necessarily with entire consortium) Strengthen ties with scientific domains Support e-irg with their work of e-infrastructure be a stakeholder to them from scientific domains

28 Challenges for PLAN-E (2) Develop PLAN-E s own position as escience stakeholder for European Commission Connect to DG-Connect & DG-Research Have clear activities e.g. workshop in Brussels, and invite key persons from EC Robert Jan Smits in general, DGs per research domain from DG Research

29 The End