Dynamic Data Driven Applications Systems

The DDDAS concept - and the term - was proposed by Frederica Darema starting in the early 80's and 90's; she discussed it in the 80’s referring to it as the “Gedanken Laboratory” and presented these ideas in 1990 at the Conference of the Society of Engineering Sciences.F. Darema, Parallel Applications and the Gedanken Laboratory, Conference of the Society of Engineering Sciences,(1990) Starting in 1999, Darema initiated the efforts within the National Science Foundation (NSF), and led the organization of  a [http://www.1dddas.org workshop in March 2000], where she designated as academic co-chairs of the workshop Profs Craig Douglas and Abhi Deshmukh.

Around 2008, Darema introduced the term Infosymbiotics or Infosymbiotic Systems to denote DDDAS. Many researchers in academia, industry, and labs were influenced to adopt the DDDAS concept and the term and conducted research under Dr. Darema's programs, starting from the mid-1990's, at DARPA, NSF (including multi-agency programs), and AFOSR. Dr. Blasch continued the program after he became Program Manager at AFOSR upon Dr. Darema becoming the Director of AFOSR in 2016. Thus, a community was formed and advanced systems capabilities and concepts are under the rubric of DDDAS.

Starting in 2000, Dr. Darema led the community in organizing several DDDAS forums; these include a series of DDDAS Workshops, Symposia, Panels, and other related activities, for example: in conjunction with the International Conference in Computational Sciences (ICCS) with Profs. Craig Douglas and Abani Patra; the International Parallel and Distributed Computing Symposium (IPDPS); the Winter Simulation Conferences (WSC). Profs. Dennis Bernstein, Puneet Singla, and Dr.  Sai Ravela organized sessions at the American Controls Conference (ACC) 2014. Dr. Ravela organized a related Dynamic Data-driven Environmental Systems Science conference, DyDESS 2014 (MIT), followed by the DDDAS 2016 (Hartford), which included participation by United Technologies Research Center, followed by DDDAS 2017 (MIT) and 2020 (Online) conferences, and hosted the 2022 (MIT) conference, organizing a new collocated [https://essg.mit.edu/cleps22 Earth, Planets, Climate, and Life] theme, CLEPS22. Since 2016, Dr. Blasch has organized numerous DDDAS and other associated forums (e.g., [https://researchr.org/publication/fusion-2015 Fusion2015] and follow-up Conference series). The 2024 conference [https://dddas2024.rutgers.edu/ DDDAS2024] was run by Prof. Dimitris Metaxas at Rutgers University (with more conferences planned in the future). The DDDAS conference proceedings are published by Springer. Other work is presented in the DDDAS Handbook series by Springer.{{Cite journal |date=2023 |editor-last=Darema |editor-first=Frederica |editor2-last=Blasch |editor2-first=Erik P. |editor3-last=Ravela |editor3-first=Sai |editor4-last=Aved |editor4-first=Alex J. |title=Handbook of Dynamic Data Driven Applications Systems |url=https://link.springer.com/book/10.1007/978-3-031-27986-7 |journal=SpringerLink |language=en |doi=10.1007/978-3-031-27986-7|isbn=978-3-031-27985-0 |url-access=subscription }} A more complete list of DDDAS forums and other activities is provided in the [http://www.1dddas.org DDDAS website]. The March 2023 US National Academies (NASEM) Report on “Foundational Research Gaps and Future Directions for Digital Twins",{{Cite book |url=https://nap.nationalacademies.org/catalog/26894/foundational-research-gaps-and-future-directions-for-digital-twins |title=Foundational Research Gaps and Future Directions for Digital Twins |date=2024-03-28 |publisher=National Academies Press |others=Committee on Foundational Research Gaps and Future Directions for Digital Twins, Board on Mathematical Sciences and Analytics, Committee on Applied and Theoretical Statistics, Computer Science and Telecommunications Board, Board on Life Sciences, Board on Atmospheric Sciences and Climate, Division on Engineering and Physical Sciences, Division on Earth and Life Studies, National Academy of Engineering, National Academies of Sciences, Engineering, and Medicine |isbn=978-0-309-70042-9 |location=Washington, D.C. |doi=10.17226/26894|pmid=39088664 }} which speaks about “predictive” capabilities in Digital Twins (DT) approaches, it identifies DDDAS as “an early concept that has all of the elements of a digital twin, including the physical, the virtual, and the two-way interaction via a feedback loop”. The NASEM Report states that (predictive) Digital Twins are holding "immense promise in accelerating scientific discovery and revolutionizing industries." Darema has used the term Dynamic Digital Twins, or DDDAS-based Digital Twins in the proceedings of the DDDAS2022 Conference{{Cite book |date=2024 |editor-last=Blasch |editor-first=Erik |editor2-last=Darema |editor2-first=Frederica |editor3-last=Aved |editor3-first=Alex |url=https://link.springer.com/book/10.1007/978-3-031-52670-1 |series=Lecture Notes in Computer Science |language=en |doi=10.1007/978-3-031-52670-1 |issn=0302-9743 |title=Dynamic Data Driven Applications Systems |volume=13984 |isbn=978-3-031-52669-5 }}), and in the subsequent DDDAS Handbooks series.

DDDAS-based approaches advance the state of the art over a number of somewhat related but more limited concepts, which have been proposed over preceding years, and which are subsets of the more comprehensive and powerful DDDAS paradigm of feedback-control between an executing model of a system with its instrumentation, as was also recognized in the above referenced March 2023 National Academies Digital Twins Report.  Prior to that, the 2006 the National Science Foundation Blue Ribbon Panel Report on Simulation-Based Science (SBES),Simulation-Based Engineering Science: Revolutionizing Engineering Science through Simulation, Report of the National Science Foundation, https://www.nsf.gov/pubs/reports/sbes_final_report.pdf also emphasizes the DDDAS aspect of “symbiotic feedback control system”, and cites DDDAS as a new paradigm, “that will rewrite the book on validation and verification of computer predictions”. The DDDAS concept goes beyond the traditional data assimilation  (which did not include the instrumentation  control aspect of DDDAS ) and is more powerful than cyber-physical systems (CPS), which is devoid of concrete definition, as to how to “integrate physical dynamics with software and networks". Other limited and superseded efforts include:

• Work in 1950s to 1970s, which were focusing on optimizing experimental methods  in a serialized (sequential) process with the human-in-the-loop making the decision on the next experiment to conduct,such as by Chertoff on [https://projecteuclid.org/journals/annals-of-mathematical-statistics/volume-30/issue-3/Sequential-Design-of-Experiments/10.1214/aoms/1177706205.full Sequential Design of Experiments], and by Fedorov on [https://onlinelibrary.wiley.com/doi/10.1002/prop.19680160602 Design of Experiments] (1970s), as depicted in [https://openlibrary.org/books/OL18496755M/Theory_of_optimal_experiments_by_V.V._Fedorov. Theory of Optimal Experiments], Diagram 1, Page 8 therein.
• Learning methods (starting in the 1980s and 1990s), such as the concept of [https://arxiv.org/abs/cs/9603104 active learning]. In distinction with the DDDAS approaches, the referenced active learning lacks systems-cognizant, first principles modeling. Active sampling strategies based on information gain are common in active and adaptive learning and relate to the [https://doi.org/10.1016/0893-6080(95)00137-9 design of experiments], e.g., [https://doi.org/10.1016/0893-6080(95)00137-9 Cohn's work] (1994); these methods, however, lack the system-cognizant design or control of instrumentation (experiments), a key capability in DDDAS.
• Reinforcement Learning (in the 90’s, and later than DDDAS) is a data-only driven approach and it’s not utilizing system-cognizant, first-principle models  (e.g., physics-based model); for example the Dyna algorithm{{Cite conference |last1=Sutton |first1=Richard |year=1990 |title=Integrated Architectures for Learning, Planning and Reacting based on Dynamic Programming |book-title=Machine Learning: Proceedings of the Seventh International Workshop}} by Sutton, which is a “trial and error” approach”, rather than the system-cognizant model premise in the DDDAS approaches and which lacks the "model-based control of the system instrumentation" of the DDDAS paradigm.
• The data assimilation concept where observation data are used to correct and constrain uncertainty in computed data-points (computed vector of a physical parameter in the model) is more limited than the DDDAS concept where the dynamic data inputs can replace a patch of the mesh and for multiple parameters. Moreover, the reverse aspect in the DDDAS feedback control loop – that is the model adaptively controlling the instrumentation, was also adopted later by the Data Assimilation concept.

• MacKay's Information-based [https://ieeexplore.ieee.org/document/6795562 Active Data Selection] (1991) employs Bayesian methods to determine expected informativeness of candidate measurements is used to select salient ones for learning, improving the expected informativeness. And, Information Retrieval (in the 90s), where queries generate searches, and the results refine the queries with relevance feedback; these approaches constitute a limited version of DDDAS.

• [http://www.1dddas.org/ 1DDDAS.org] Has a list of active projects and slides from the current DDDAS program and past contributions from NSF.

• Data assimilation
• Control systems
• Uncertainty Quantification
• Active Learning

Category:Theoretical computer science