The slides that follow were presented at the PSAAP Bidder's Meeting May 1617, 2006 and represent the

1
The slides that follow were presented at the
PSAAP Bidder's Meeting May 16-17, 2006 and
represent the ASC Trilab authors and interests as
presented in the associated White Paper for this
subject area.
Predictive Science Academic Alliance Program
(PSAAP)
2
Predictive Science Academic Alliance
Program(PSAAP)

• Bob Voigt
• Representing the AST

3
Outline

• Goal
• Process
• Proposal requirements
• Review criteria
• Issues
• Management

4
Goal Establish Predictive Science via
Multidisciplinary Simulation Centers

• Create validated simulation capability in
  academia for prediction
• Scientific/engineering discovery
• National productivity
• Focus on a single, multiscale, multidisciplinary
  application of ASC/NNSA/National importance
• Demonstrate on most powerful ASC computing
  systems available

5
Goal

• Produce significant science/engineering results
• Produce new methodologies
• Verification
• Validation
• Uncertainty quantification
• Tight interplay of experiment simulation
• Improve tools and algorithms

6
Goal

• Integrate science/engineering, computational math
  and computer science into a focused research
  effort
• Increase Center/Lab interactions, e.g.
• Interplay between experiment simulation
• Tool or algorithm development
• Broaden the academic base of expertise in
  large-scale, multidisciplinary, simulation based
  predictability
• Increase the human resource pool exposed to
  predictive science

7
Process

• RFI to be released on May 26
• Review by panel of representatives from NNSA
  Labs, ASC HQ, academia, and other Labs and
  industry organized by AST
• Subset of proposers invited to submit full
  proposals in response to RFP
• All who respond to RFI will receive feedback

8
Process

• Proposals reviewed by panel of representatives
  from NNSA Labs, ASC HQ, academia, and other Labs
  and industry organized by AST
• After reviews a set of proposals deemed fundable
  submitted to ASC Execs for possible site visits
  and final selection
• Awards in place in October of 2007

9
Successful Proposals Will Include

• Focus on Science of Predictability
• A single, multi-physics, multi-scale problem of
  ASC/National interest
• Plan for VV and uncertainty quantification with
  plan for acquiring required data
• Predictive science research plan with clear,
  meaningful demonstration goal
• Convincing need for PFLOPS

10
Successful Proposals Will Include

• Convincing plan for integrating
  science/engineering, computational math and
  computer science focused on chosen problem
• Matching funds
• A plan for interacting with Labs
• A plan for attracting students capable of
  acquiring a DOE clearance

11
Review CriteriaDegree to which proposals address
the Goaland the quality of

• A plan for conducting research in the science of
  predictability via simulation with a focus on a
  single, multi-physics, multi-scale application
• Relevance of the problem to the NNSA Laboratories
• Importance of the demonstration goals to the
  broad science and engineering community
• Quality of proposed supporting disciplinary
  research
• A plan for utilizing existing, and developing
  new, software in an integrated simulation
  framework

12
Review Criteria

• A plan to assess the degree to which the proposed
  simulations agree with physical reality
• Innovativeness of the VV and UQ plans
• Plan for acquiring data for VV
• A convincing argument that all required
  disciplines are fully integrated into a coherent
  management structure
• Quality and appropriateness of the management
  team
• Plan for managing external partners if any
• A plan for interaction with the NNSA
  Laboratories
• A plan for attracting students and educating them
  in the science of predictability
• Matching funds.

13
Who Can Receive PSAAP Funding
14
Lab Interactions

• There are NO deliverables to the Labs
• Examples of interactions
• Visits by faculty/students/staff
• Experimental data
• Test problems
• VV UQ methodology
• Algorithm
• Independence

15
Matching Funds

• Minimum of 10 in matching funds is required
• Matching must be in the form of accountable
  dollars
• Examples include
• Labor for faculty, students, postdocs, staff, etc
• Cost of acquiring data for VV
• Reduction in university overhead rate
• Actual cost of new computing equipment
• NNSA Lab collaborators cannot be counted as
  contributing matching funds

16
Data for VV

• A modest fraction of funds may be used to support
  acquisition of data
• University experiment
• Off-site experiment
• Or actual cost may be counted towards matching
  funds
• No-cost interaction with NNSA Lab
• Be innovative

17
Multiple Universities

• Multiple universities may participate in a single
  proposal
• However
• Management can be challenging
• Contribution and responsibilities must be clear
• Proposal must make a convincing case for need
• See CRAs

18
Lab Advice

• Window of opportunity
• Talk to people
• Tri-Lab program
• 10,000 people with 20,000 points of view
• Buyer beware
• First VV exercise
• Contact the AST
• The more Lab interest the better

19
Management

• AST (Alliance Strategy Team)
• Overall management responsibility
• Member from each Lab HQ Contracting
• TST (Tri-Lab Sponsor Team)
• One for each MSC
• Support MSC and coordinate with Labs
• 2 Members from each Lab
• CRT (Computing Resource Team)
• Management of computing resources
• Member from each Lab and each MSC

20
Coming Attraction CRAs

• Integrated with one or more MSCs and/or one or
  more NNSA Labs
• Unique capability
• 200-800K per year for up to 3 years
• Require MSC endorsement
• New competition every 2-3 years beginning in 2008.

21

• Continue to visit the website
• http//www.llnl.gov/asci/alliances/psaap/
• Slides will be online or contact
• rvoigt_at_compsci.wm.edu