2025 Conferences

[arthurrge]

Please post notices for / suggest gathering at conferences in 2025.

• SciTech 2025 (Jan 6-10)
• Dayton Digital Transformation Summit (May 6-8)

[arthurrge]

SciTech 2025

CoDADE update to DEIC

• Sci-Tech 25 F2F CoDADE Slides.pdf (prepared by Abhishek)

Panel: DGE-19: A Study of Guidelines for Enabling Interoperable and Responsible Use of AI

Description: This panel will report on the results of a survey aimed at understanding the impacts of computational methods, data, infrastructure, and organizational culture on the maturity of different non-deterministic methods (e.g., AI) with respect to interoperability and responsible use. The panelists will receive the results of this survey and they will provide their positions and responses to the survey results. The goal of the survey and the panel is to inform the development of additional guidelines aimed at promoting the responsible use of AI and maximizing interoperability.
Panel facilitator: Abhishek Ram, Draper

Steve Arnold (NASA)

• See Poster: Materials Informatics Enables Rapid Materials Discovery
• MicroNet = tool for labelling microstructures (like cats and dogs in photos) https://github.com/nasa/pretrained-microscopy-models
• Anecdote of 1 day file converter created via GPT tools
• Surrogates as means to support multiscale
• AIMAOS = Automatic Information Management Across Organizations and Scales
  • tool to query code/scripts of solvers at differing scales and { detect, track, propagate } changes to other scales
  • See https://ntrs.nasa.gov/citations/20240010796
• Tracking of intellectual debt
• JARIMIS = Just a Rather Intelligent Materials Interrogation System
  • tool to guide next experimentation parameters (like how we use iDACE)
  • See ICME Schema for Materials Data Management: An Executive Summary
  • See Materials Informatics: Information Management Toolset (slides)
  • See ICME: The Design of Fit-for-Purpose Materials (slides)
  • See Managing the digital thread for structural applications with fit-for-purpose materials (slides)

Chad Hall (Lockheed-Martin)

• Field expertise and explainable AI, NLP (natural language processing)

Alicia Kim (UCSD)

• Apply FAIR principles to engineering / science software to promote collaboration / community / interoperability
• "optimization exploits the weakness(es) in the problem statement(s)"
• Twins are always incomplete (as are all models)
  • and / or may be prohibitively expensive at desired fidelity / completeness
  • therefore they should respond "I don't know" if use attempted outside of (self-known) expertise
  • Therefore, important how their (output) data/information is structured
• Emergent behaviors are usually due to factors we do not understand (e.g., nonlinear couplings resulting in 3rd order effects becoming dominant)

Rick Arthur (GE Aerospace)

• AIpanel-final-RBA.pdf
• Goals/Opportunities
  • Multidisciplinary optimization { systems-of-systems, ASoT, change propagation/consistency, data exchange, functional interoperability }
  • E.g., Geometry/Mesh Generation & Validation, Model Building & Validation
• Legacy Challenges = { complexity, inconsistency, subjectivity, transdisciplinary-ness, tedious/labor-intensive tasks }
• Levers (technological advances)
  • Abundant + affordable { Compute, Storage, Network, Sensing } / HPC + IoT via Moore's Law
  • Abundant + affordable data leads to data-derived models (ML)
  • Standardized data + HPC leads to multi-{ physics, scale } models w/ more realism, fidelity, robustness
  • Measurements + models = confidence (via VVUQ processes)
  • Minds + machines = productivity (cognition at scale) + insight (perception at scale)
• AIAA DEIC Workforce Development paper (imminently to be released)
• Vision for model-view capabilities w/ automated mesh generation & validation (domain-dependent discretization, tolerancing, boundary conditions)

[arthurrge]

Note Rick panel (June 2025): The Convergence of AI, HPC, and Policy: Shaping the Future of Computing