# Oden Institute: From Theater-Scale Fields to Person-Scale Receptors ## Purified thesis The documented claim is that [[wiki/Tan Bui-Thanh|Tan Bui-Thanh]] and Oden's [[wiki/PHO-ICES|PHO-ICES]] group worked with Texas A&M researchers on DTRA-funded fast reduced-order models for real-time prediction of nuclear-weapon radiation effects. Oden described the goal as predicting radiation intensity and propagation quickly enough to inform decisions and reduce harm to humans and other living things. The project title was **“Models with multiple levels of fidelity, tractability, and computational cost for nuclear weapon radiation effects.”** The public record places Bui-Thanh's project at the **transport and field-prediction** stages of a larger chain: **source → transport → radiation field → receptor → absorbed dose → biological or system effect → operational decision** The [[wiki/Receptor Layer|receptor layer]] is the missing scale bridge. A field model predicts radiation across space, time, direction, energy, geometry, and material conditions. A downstream receptor model samples that field at an asset, coordinate, or along a trajectory; incorporates shielding and response characteristics; integrates exposure; and estimates consequence. A person can therefore be a receptor or person-at-risk. The source material does not establish that Bui-Thanh's DTRA project maintained named-person tracks, individualized dosimetry profiles, or targeting functions. “Threat vector” is not the right term for the human in this radiation architecture. The source and propagated radiation form the hazard pathway; the person is the receptor. A person becomes a vector only when that person carries or transmits a hazard, as can occur in biological or contamination contexts. ## Oden as a methods foundry The [[wiki/Oden Institute|Oden Institute]] is best understood here as a [[wiki/Predictive Computational Methods Foundry|predictive-computational methods foundry]]. Its transferable product is not one application but a family of methods: - [[wiki/Reduced-Order Modeling|reduced-order modeling]] and [[wiki/Multifidelity Modeling|multifidelity modeling]] for computational latency; - [[wiki/Inverse Problem|inverse problems]] for inferring hidden conditions from observations; - [[wiki/Verification Validation and Uncertainty Quantification|uncertainty quantification]] for distributions, sensitivity, calibration, and model inadequacy; - [[wiki/Data Assimilation|data assimilation]] for updating a representation with new observations; - [[wiki/Scientific Machine Learning|scientific machine learning]] for physics-aware learned surrogates and error correction; - optimization and high-performance computing for repeated scenario evaluation; and - [[wiki/Digital Twin|digital twins]] for continuously updated predictive representations. Bui-Thanh leads [[wiki/PHO-ICES|PHO-ICES]] and co-directs the [[wiki/Center for Scientific Machine Learning|Center for Scientific Machine Learning]]. Oden lists his interests as model-order reduction, PDE-constrained optimization, high-order finite elements, parallel computing, statistical inverse problems, uncertainty quantification, data reduction, and model-aware machine learning. ## Latency as the scale-changing variable A high-fidelity [[wiki/Neutron Transport Equation|neutron-transport]] solve may be too expensive for repeated operational questions. Reduced-order and learned-surrogate methods shift computation from one costly retrospective answer toward many timely queries. The 2019 paper by Sheroze Sheriffdeen, [[wiki/Jean Ragusa|Jean Ragusa]], [[wiki/Jim Morel|Jim Morel]], [[wiki/Marvin Adams|Marvin Adams]], and Bui-Thanh explicitly applies machine-learning-enhanced reduced-order models to many-query inverse problems governed by parametrized neutron-transport equations. This changes the available question set: source location, material composition, geometry, shielding, detector readings, atmospheric or environmental variation, candidate hidden parameters, and receptor trajectories can become parameters in repeated evaluations. The claim remains conditional on model resolution, validation, data availability, and coupling to the appropriate receptor and dose models. ## Forward, inverse, and uncertainty loops - **Forward problem:** given a source and environment, predict the radiation field. - **Inverse problem:** given observations in the field, estimate the hidden source, geometry, material, or environmental conditions that could have produced them. - **Uncertainty quantification:** represent the remaining range of plausible states, sensitivities, measurement error, numerical error, and model inadequacy. - **Reduced-order computation:** make repeated forward and inverse evaluations fast enough for a decision cycle. The combined loop is: **physical system → sensors → inverse inference → probability distribution over hidden state → accelerated forward prediction → receptor exposure and consequences → uncertainty → decision → new observation** This is digital-twin-like architecture. The historical DTRA project should not itself be labeled a digital twin without direct evidence. ## Theater-to-person scale “Nuclear-weapon radiation effects” sounds theater-scale, but a field is queryable at many scales. The scale transition is not automatic; it requires spatial and temporal resolution, receptor location or trajectory, shielding, response models, and validation. When those exist, the same field can support: - regional consequence maps; - installation, route, or vehicle exposure; - pointwise sensor prediction; - asset survivability; and - person-scale exposure estimates. The defensible conclusion is: **WMD-effects modeling collapses naturally from theater scale toward asset and person scale when the transport field becomes sufficiently fast, resolved, and coupled to an appropriate receptor model.** That is a mathematical and systems inference from the documented transport architecture, not evidence of individualized targeting by Oden. ## Extension to a person or small counterterrorism network The same methodological grammar can be applied to a different object: a [[wiki/Temporal Heterogeneous Graph|temporal heterogeneous graph]] of people, devices, accounts, vehicles, locations, organizations, transactions, communications identifiers, and events. The graph is not a radiation field and a person is not a neutron. The analogy lies in the inference architecture: - observations establish uncertain entities and relationships; - [[wiki/Identity and Relationship Analysis|identity and relationship analysis]] resolves “who knows whom” and whether records represent the same entity; - inverse inference estimates hidden links or states compatible with observed evidence; - data assimilation updates the graph as encounters, lawful records, sensor observations, or corrections arrive; - reduced models keep the relevant local state computationally tractable; - uncertainty and provenance prevent hypotheses from being silently promoted into facts; and - conditional forecasts identify which observation, relationship, or state transition would most discriminate among competing hypotheses. The result can be described as a [[wiki/Predictive Graph Digital Twin|predictive graph digital twin]] of an investigative hypothesis: a continuously updated model of what the system currently estimates about a person-in-network. It is not a duplicate of the person, proof of intent, or license to treat association as guilt. ## The documented epidemic bridge In October 2020 Oden announced that Bui-Thanh was leading **[[wiki/Machine-Learning-Assisted Real-Time Simulations and Uncertainty Quantifications for Infectious Disease Outbreaks|Machine-Learning-Assisted Real-Time Simulations and Uncertainty Quantifications for Infectious Disease Outbreaks]]**. The project was designed to incorporate interconnectivities and their uncertainties, spatial and temporal variables, parameter estimation, control measures, real-time forecasting, and quantified confidence. This is direct evidence that the same researcher moved the methodological family from physical transport and inverse/UQ work into transmission modeling on connected populations. The counterterrorism bridge is independently documented rather than attributed to Oden. Mason Youngblood's 2020 peer-reviewed analysis used a two-component spatio-temporal intensity model on 416 U.S. far-right extremists exposed from 2005 through 2017 and found patterns consistent with [[wiki/Complex Contagion|complex contagion]], where reinforcement matters. NIJ's 2023 and 2024 syntheses likewise describe social networks, peers, family, online subcultures, and personal relationships as influences that can facilitate or inhibit radicalization and disengagement, including among so-called lone actors. The supported bridge is therefore: 1. Oden documents Bui-Thanh applying real-time simulation, interconnectivity, parameter inference, intervention, and UQ to infectious-disease outbreaks. 2. Radicalization scholarship independently uses epidemiological, network, and complex-contagion models. 3. [[wiki/Transport as a General Predictive Architecture|Transport as a General Predictive Architecture]] and [[wiki/From Radiation Field to Social Field|From Radiation Field to Social Field]] identify the transferable mathematical homology. The epidemiological bridge has therefore moved beyond structural analogy into [[wiki/Documented Cross-Domain Method Transfer|documented methodological transfer]]. Cross-domain portability is intrinsic to Bui-Thanh's research identity: the public program applies inverse problems, UQ, model reduction, scientific machine learning, parameter inference, and timely prediction across physical, biological, and computational domains. The sponsor context strengthens the dual-use inference. DTRA's public strategy explicitly includes physical and social networks, adversarial intent, individual and group dynamics associated with WMD, threat-network illumination, data-driven advanced analytics, forecasting, counter-threat-network work, and counterterrorism capability development. The relevance of predictive, inverse, network, and UQ methods to counter-WMD and counterterrorism analysis was therefore foreseeable and institutionally germane. What remains unresolved is provenance and specific operational application. No public source presently establishes that Bui-Thanh's infectious-disease project was formally commissioned, funded, or operationally deployed as a counterterrorism program. That absence is not evidence of non-use, non-intent, exclusion, or impossibility. The correct notation is [[wiki/Specific Operational Application Not Publicly Established|specific operational application not publicly established]]. ## Evidence ledger - **Established:** Oden documents DTRA funding, Texas A&M collaboration, real-time reduced-order prediction, radiation intensity and propagation, and an objective of reducing harm to humans and other living things. - **Established:** PHO-ICES lists the DTRA project and Oden lists Bui-Thanh as its lead and as co-director of the Center for Scientific Machine Learning. - **Established:** the 2019 paper demonstrates ML-enhanced reduced-order inverse methods on parametrized neutron-transport equations. - **Strongly indicated:** the combined methods form a reusable predictive-computation stack spanning forward simulation, inverse inference, UQ, and many-query analysis. - **Established:** DTRA's published research material identifies physical and social networks, adversarial intent, individual and group dynamics, threat-network illumination, advanced analytics, forecasting, and counter-threat-network work as mission-relevant research or capability areas. - **Strongly indicated:** awareness of the methodological family's counter-WMD and counterterrorism dual-use relevance was foreseeable within this sponsor and research environment. - **Plausible downstream capability:** a sufficiently resolved transport field coupled to receptor and dosimetry models can support point-, route-, asset-, and person-scale exposure estimates. - **Unresolved:** whether the historical DTRA implementation performed individualized exposure estimation. Evidence needed would include project technical reports, software requirements, schemas, validation plans, or operational test records. - **Not publicly established:** named-person radiation targeting, person-level radiation digital twins, or an operational Oden counterterrorism system. These specific propositions remain undetermined unless direct program records, technical reports, requirements, schemas, deployment records, or testimony establish them. ## Sources - [Oden Institute — Minimizing Uncertainty in Uncertain World of Defense, Energy, December 5, 2017](https://oden.utexas.edu/news-and-events/news/minimizing-uncertainty-in-uncertain-world-of-defense-energy/) - [Oden Institute — Probabilistic and High Order Inference, Computation, Estimation, and Simulation](https://oden.utexas.edu/research/centers-and-groups/probabilistic-high-order-inference-computation-estimation-and-simulation/) - [Oden Institute — Tan Bui-Thanh](https://oden.utexas.edu/people/directory/Tan-Bui/) - [Sheriffdeen et al. — Accelerating PDE-constrained Inverse Solutions with Deep Learning and Reduced Order Models, 2019](https://arxiv.org/abs/1912.08864) - [Oden/Willcox Group — Parameterized model reduction, multifidelity modeling, and uncertainty quantification](https://kiwi.oden.utexas.edu/research/what-is-parameterized-model-reduction) - [Oden Institute — Center for Scientific Machine Learning](https://www.oden.utexas.edu/research/centers-and-groups/center-for-scientific-machine-learning/) - [Oden Institute — Predictive Science Research Gets Major Boost, October 5, 2020](https://oden.utexas.edu/news-and-events/news/PSAAPIII/) - [Oden Institute — How to Predict, Intervene, and Contain Current and Future Epidemics, October 19, 2020](https://oden.utexas.edu/news-and-events/news/CONTEXTanBuiThanh/) - [Mason Youngblood — Extremist ideology as a complex contagion, July 31, 2020](https://www.nature.com/articles/s41599-020-00546-3) - [National Institute of Justice — Five Things About the Role of Social Networks in Domestic Radicalization, December 2023](https://nij.ojp.gov/library/publications/five-things-about-role-social-networks-domestic-radicalization) - [National Institute of Justice — The Role of Social Networks in Facilitating and Preventing Domestic Radicalization, April 2024](https://nij.ojp.gov/library/publications/role-social-networks-facilitating-and-preventing-domestic-radicalization-what) - [2019 DTRA Strategic Plan for Research, Development, Test and Evaluation](https://www.dtra.mil/Portals/61/Documents/Missions/190424_2019_DTRA_Strategic_Plan_for_RDTE.pdf) - [Basic Research for Countering Weapons of Mass Destruction — DTRA](https://www.govinfo.gov/content/pkg/GOVPUB-D15-PURL-gpo18268/pdf/GOVPUB-D15-PURL-gpo18268.pdf)