Project 76: Wo Co 10 Prescott 2006: ============== Ford pointed out the need to initiate this "project" and plan for the next WoCo in three to four years. Members were urged to consider topics and hosts and to be prepared to make specific plans at our next meeting. Uppsala 2007: ============= This new project was introduced last year as the first step in planning and coordinating our next WoCo. Boisvert agreed to set up a small committee (as discussed earlier in the meeting) and report back at next years meeting. Toronto 2008: ============= This project was introduced two years ago as the mechanism for planning and coordinating our next WoCo. Boisvert has established a small committee of interested members and a report on recent progress is discussed in the "Future Meeting" section of the minutes, the proposed title is "Uncertainty Quantification in Scientific Computation". See also the link "Draft plans for Wo Co 10" on the projects page, under project 76. Raleigh 2009: ============= This project was introduced two years ago as the mechanism for planning and coordinating our next WoCo. Boisvert presented a detailed proposal for WoCo 10 on "Uncertainty Quantification in Scientific Computing". This proposal was developed by Ron and his colleague, Andrew Dienstfrey and was strongly supported by all members. The meeting would be held in Boulder possible dates were discussed. Early to mid-September 2011 seemed to be the most suitable time. Ron and Andrew will make a specific recommendation at next years meeting. A program committee has been established and potential invitees will be determined in the next few months. Members are urged to suggest names to Ron or Andrew. It is expected that there will be 25 WG members in attendance for the business meeting and an additional 35 researchers invited for the WoCo. Recall that participation in an IFIP working conference is by invitation and we all should participate actively in identifying the most suitable participants. Leuven 2010: ============ A detailed plan for this conference was presented. See also http://www.nist.gov/itl/math/ifip-woco-10.cfm Boulder 2011: ============= This project will now be terminated as it was introduced as the mechanism for planning and coordinating Wo Co 10. Boisvert's follow-up report on this successful meeting should be considered the final report. That report follows here. The IFIP Working Conference on Uncertainty Quantification (UQ) in Scientific Computing took place in Boulder, Colorado, USA on August 1-4, 2011. This was the 10th in a series of Working Conferences organized by Working Group 2.5. The 76 participants representing 10 countries were a unique mix of mathematicians, statisticians, computer scientists, application scientists and managers who gathered together to consider the role of measurement science in mathematical modeling and computer simulation. The latter techniques and tools are beginning to play a critical role as adjuncts to, or replacements for, direct physical measurement in both product design and societal decision making. The careful quantification of uncertainty in such computed results is believed to be a prerequisite for the effective use of modeling and simulation in this way. The conference program was divided into four theme areas: (1) Uncertainty Quantification Need: Risk, Policy and Decision Making, (2) Uncertainty Quantification Theory, (3) Uncertainty Quantification Tools, and (4) Uncertainty Quantification Practice. Keynote addresses were delivered by Pasky Pascual of the Environmental Protection Agency (US), Michael Goldstein of Durham University (UK), William Kahan of the University of California at Berkeley (US), Scott Ferson of Applied Biomathematics (US), and Maurice Cox of the National Physical Laboratory (UK). A panel discussion featured Sandy Landsberg of the US Department of Energy's Office of Science, Larry Winter, former Deputy Director of the US National Center for Atmospheric Research (NCAR), and Charles Romine, former Acting Deputy Director of the US National Institute of Standards and Technology (NIST). Antonio Possolo, Chief of NIST’s Statistical Engineering Division, Mark Cunningham of the US Nuclear Regulatory Commission, Alberto Paisanini of L'Ectricite de France, les Hatton of Kingston University (UK), and Peter Challenor of the National Oceanography Centre (UK) were among the notable plenary speakers. Three WG 2.5 members, Pat Gaffney, Brian Smith, and Wayne Enright, also gave well regarded plenary talks. Not only did the meeting raise the awareness of UQ within the numerical software community, it also established new connections between disparate groups who have heretofore worked on various aspect of this emerging problem area in isolation. Numerical analysts learned about statistical approaches, statisticians learned about computational techniques, computer scientists were exposed to new applications, and all learned of the needs and constraints of decision-makers. Talks aimed at broad audience enabled appreciation of diverse points-of-view, and the interaction of participants was further facilitated by the conferences moderate size, long Q&A sessions, and group breaks, lunches, and evening events. Finally, a lively Hot Topics session, organized after the start of the conference, provided seven participants with 15 minute slots to raise additional issues for group discussion. The conference uncovered many sources of uncertainty, including * Hardware: bit flips, floating-point anomalies * Systems: compilers, non-determinism (parallelism) * Software: defects * Algorithms: discretization, convergence * Model inputs: parameters, initial conditions * Model form: approximations, missing physics * Human judgment The conference provided a glimpse of a variety of application domains, including cosmology weather/climate modeling, remote sensing/atmospheric chemistry, hydrology, nuclear energy/safety, fluid dynamics, engineering mechanics, manufacturing processes, medicine, toxicology, and computer security. Speakers exposed a blizzard of UQ terminology, approaches and techniques, including * Aleatory vs. epistemic uncertainty * Verification & validation * Stochastic Differential Equations * Bayes Theorem * Propagation of probability density functions * Monte Carlo/importance sampling/Markov Chain Monte Carlo * Design of experiments, adaptive designs * Emulators * Sensitivity analysis/response surfaces * Stochastic expansions, e.g., polynomial chaos * Model calibration, data assimilation The conference participants identified many existing tools, including * Guidance: GUM, MUCM toolkit * Computing infrastructure: intervals, P-boxes * Software: DAKOTA toolkit, Test harness, R, OpenTURNS * Visualization techniques * Hardware: emerging architectures The Program Committee for the conference was comprised of * Andrew Dienstfrey, NIST, US, Chair * Ron Boisvert, NIST, US * Maurice Cox, National Physical Laboratory, UK * Bo Einarsson, Linköping University (retired), SE * Brian Ford, NAG Ltd. (retired), UK * Mac Hyman, Tulane University, US * Tony O'Hagan, University of Sheffield, UK * Michael Oberguggenberger, University of Innsbruck, AT * Bill Oberkampf, Sandia Labs (retired), US Brian Ford organized the conference's Hot Topics session and Mac Hyman organized the panel discussion. Boisvert, Ford, and Einarsson are members of WG 2.5. Social events held in conjunction with the conference included a reception (Mon Aug 1), a tour of the Boulder campus of NIST and NOAA (Wed Aug 3), and a conference banquet held at the Chautauqua Park in Boulder. Accompanying persons were treated to tours of the Celestial Seasoning's factory in Boulder, the Coors Brewery in Golden and the Denver Art Museum. They also participated in excursions to Pearl Street Mall, Flagstaff Mountain, the Leaning Tree Museum, and the Dushanbe Tea House. Rita Boisvert and Sydelle Dienstfrey organized the accompanying persons program. The meeting was organized on behalf of IFIP WG 2.5 by Andrew Dienstfrey and Ronald Boisvert of NIST's Applied and Computational Mathematics Division, which served as host and co-sponsor the the conference. Jim Pool of WG 2.5 also served on the Organizing Committee. Excellent administrative support was provided by Wendy McBride of NIST Conference Services, Lorna Buhse, Secretary of the NIST Information Technology Laboratory (Boulder), and Robin Bickel, Secretary of the NIST Applied and Computational Mathematics Division (Gaithersburg, MD). Further details can be found on the conference website, http://www.nist.gov/itl/math/ifip-woco-10.cfm Proceedings will be published. ---------- Ronald Cools reported 10 April 2012 The following extract from the recent SIAM unwrapped newsletter might be of interest, especially to those that attended the meeting last year in Boulder. > Report on Verification, Validation, and Uncertainty Quantification > > A report discussing Verification, Validation, and Uncertainty > Quantification (VVUQ) and surveying the science and methods in this > growing area has been produced by the Board on Math Sciences and their > Applications (BMSA) via its Committee on Mathematical Foundations of > VVUQ. Entitled "Assessing the Reliability of Complex Models: > Mathematical and Statistical Foundations of Verification, Validation, > and Uncertainty Quantification", it touches upon three main aspects: > Verification: How accurately does the computation solve the underlying > equations of the model for the quantities of interest? > > Validation: How accurately does the model represent reality for the > quantities of interest? > > Uncertainty Quantification (UQ): How do the various sources of error > and uncertainty feed into uncertainty in the model-based prediction of > the quantities of interest? > > A pre-publication version of this report is available through the > National Research Council at: > http://www.nap.edu/catalog.php?record_id=13395 > > The report appears as SIAM, in cooperation with the ASA and US ACM, > holds its first Conference on Uncertainty Quantification, > another indication of the growing importance of this area. Kind regards, /Ronald/ Santander 2012: =============== The proceedings have appeared from Springer 2012. The price is approximately 98 Euro including tax, but each attendee will receive a free copy. ISBN 978-3-642-32676-9 (Hardcover), 319 pages. It was decided to re-classify this project as inactive.