ASC Predictive Science Academic Alliance Program Verification and Validation Whitepaper

Klein, R ; Graziani, F ; Trucano, T

Lawrence Livermore National Laboratory

关键词: Forecasting;    99 General And Miscellaneous//Mathematics, Computing, And Information Science;    Specifications;    Simulation;    Verification;   

DOI  :  10.2172/928524
RP-ID  :  UCRL-TR-220342
RP-ID  :  W-7405-ENG-48
RP-ID  :  928524

美国|英语

来源: UNT Digital Library

【 摘 要 】

The purpose of this whitepaper is to provide a framework for understanding the role that verification and validation (V&V) are expected to play in successful ASC Predictive Science Academic Alliance (PSAA) Centers and projects. V&V have been emphasized in the recent specification of the PSAA (NNSA, 2006): (1) The resulting simulation models lend themselves to practical verification and validation methodologies and strategies that should include the integrated use of experimental and/or observational data as a key part of model and sub-model validation, as well as demonstrations of numerical convergence and accuracy for code verification. (2) Verification, validation and prediction methodologies and results must be much more strongly emphasized as research topics and demonstrated via the proposed simulations. (3) It is mandatory that proposals address the following two topics: (a) Predictability in science & engineering; and (b) Verification & validation strategies for large-scale simulations, including quantification of uncertainty and numerical convergence. We especially call attention to the explicit coupling of computational predictability and V&V in the third bullet above. In this whitepaper we emphasize this coupling, and provide concentrated guidance for addressing item 2. The whitepaper has two main components. First, we provide a brief and high-level tutorial on V&V that emphasizes critical elements of the program. Second, we state a set of V&V-related requirements that successful PSAA proposals must address.