Physical Sciences and Mathematics Commons^™

Prediction-Powered Inference For Clinical Trials, Pierre-Emmanuel Poulet, Maylis Tran, Sophie Tezenas Du Montcel, Bruno Dubois, Stanley Durrleman, Bruno M. Jedynak

Mathematics and Statistics Faculty Publications and Presentations

Prediction-powered inference (PPI) [1] and its subsequent development called PPI++ [2] provide a novel approach to standard statistical estimation leveraging machine learning systems to enhance unlabeled data with predictions. We use this paradigm in clinical trials. The predictions are provided by disease progression models, providing prognostic scores for all the participants as a function of baseline covariates. The proposed method would empower clinical trials by providing untreated digital twins of the treated patients while remaining statistically valid. The potential implications of this new estimator of the treatment effect in a two-arm randomized clinical trial (RCT) are manifold. First, it leads …

Wind And Topography Underlie Correlation Between Seasonal Snowpack, Mountain Glaciers, And Late-Summer Streamflow, Elijah N. Boardman, Andrew G. Fountain, Joseph W. Boardman, Thomas H. Painter, Evan W. Burgess, Laura Wilson, Adrian A. Harpold

Geology Faculty Publications and Presentations

In a warming climate, net mass loss from perennial snow and ice (PSI) contributes a temporary source of unsustainable streamflow. However, the role of topography and wind in mediating the streamflow patterns of deglaciating watersheds is unknown. We conduct lidar surveys of seasonal snow and PSI elevation change for five adjacent watersheds in the Wind River Range, Wyoming (WRR). Between 2019 and 2023, net mass loss from PSI is equivalent to ~10-36% of August-September streamflow. Across 338 manually classified PSI features >0.01 km2, glaciers contribute 68% of the total mass loss, perennial snowfields contribute 8%, rock glaciers contribute 1%, buried …

Identifying The Effect Of Architecture Morphology On Wind Fields And Pm2.5 Dispersion At The Urban Block Scale Via A Combined Scheme, Bin Guo, Lu Li, Xiaowei Zhu, Zheng Wang, Miaoyi Chen, Lin Pei, Tengyue Guo

Mechanical and Materials Engineering Faculty Publications and Presentations

This study explores the potential of architecture morphology to mitigate PM2.5 pollution through enhanced urban ventilation. Despite the recognized influence of building designs on wind and air pollution distribution at the urban block scale, specific mechanisms remain poorly understood. To address this gap, we have developed an integrative approach combining in-situ observations, Geographic Information System (GIS), Remote Sensing (RS), Computational Fluid Dynamics (CFD), and statistical modeling. Here, CFD simulates the flow field and PM2.5 dispersion, using in-situ observations for turbulence model selection. GIS quantify architecture morphology parameters to establish accurate CFD boundary conditions, ensuring simulations reflect real-world conditions and precise …

Seedlings Of Pacific Northwest Plants: A Photographic Guide, Gabriel E. Campbell-Martinez, Roxy Olsson

Rae Selling Berry Seed Bank and Plant Conservation Program

This document presents a visual, non-technical field guide of native and exotic/invasive seedlings that grow wild in the Pacific Northwest. It contains over 550 taxa representing a broad range of families and ecosystems, focusing on species that occur in Oregon and Washington. Species were selected to reflect ecological importance, frequency of occurrence, and relevance to land management, restoration, and conservation. It is intended for a broad audience, including restoration practitioners, land managers, botanists, ecologists, students, naturalists, and others interested in identifying plants at early life stages.

Recovering Population Of The Southern Sea Otter Suppresses A Global Marine Invader, Rikke Jeppesen, Catherine E. De Rivera, Edwin D. Grosholz, M. Tim Tinker, Brent B. Hughes, Ron Eby, Kerstin Wasson

Environmental Science and Management Faculty Publications and Presentations

Understanding the role of apex predators on ecosystems is essential for designing effective conservation strategies. Supporting recovery of apex predators can have many benefits; one that has been rarely examined is control of invasive prey. We investigated whether a recovering apex predator, the southern sea otter (Enhydra lutris nereis), can exert local control over a global marine invader, the green crab (Carcinus maenas). We determined that southern sea otters in Elkhorn Slough estuary in California can consume large numbers of invasive green crabs and found strong negative relationships in space and time between otter and green crab abundance. Green crabs …

Collaborative Wildfire Governance Within The Southwest Idaho Wildfire Crisis Strategy Landscape: Survey Results 2025, Mahmood Muttaqee, Cody R. Evers, Lindsay K. Campbell, Max Nielsen-Pincus, Olivia Awbrey, Kathryn Hixson, Michelle Johnson, Erika Svendsen

Environmental Science and Management Faculty Publications and Presentations

Collaboration plays a critical role in wildfire risk mitigation efforts. This study examines collaborative wildfire governance networks within Southwest Idaho, a priority landscape designated in 2022 under the USDA Forest Service's Wildfire Crisis Strategy (WCS). Through survey data from 64 respondents, the Fueling Adaptation study aimed to identify the structure, capacity, and gaps in governance networks addressing wildfire risk in Southwest Idaho and understand how WCS investments have influenced wildfire management in the area. The findings reveal a growing governance network centered on the USDA Forest Service and the Idaho Department of Lands and supported by numerous non-governmental organizations and …

Rock: Riesz Occupation Kernel Methods For Learning Dynamical Systems, Victor W. Reilly, Chau Nguyen, Kamel Lahouel, Anthony Kolshorn, Nicholas Fisher, Bruno Jedynak

Mathematics and Statistics Faculty Publications and Presentations

We present a representer theorem for a large class of variational problems, generalizing regression, finite element methods, and providing techniques for learning partial and ordinary differential equations (PDEs and ODEs). We apply our formulation to the multivariate occupation kernel method (MOCK) for learning dynamical systems from data. Our generalized method outperforms the MOCK method on most tested datasets, while often requiring considerably fewer parameters. Experiments are presented for learning ODEs and PDEs.

Mock: An Algorithm For Learning Nonparametric Differential Equations Via Multivariate Occupation Kernel Functions, Victoria A. Rielly, Kamel Lahouel, Ethan Lew, Nicholas Fisher, Vicky Geneva Haney, Michael Wells, Bruno M. Jedynak

Mathematics and Statistics Faculty Publications and Presentations

Learning a nonparametric system of ordinary differential equations from trajectories in a -dimensional state space requires learning functions of variables. Explicit formulations often scale quadratically in unless additional knowledge about system properties, such as sparsity and symmetries, is available. In this work, we propose a linear approach, the multivariate occupation kernel method (MOCK), using the implicit formulation provided by vector-valued reproducing kernel Hilbert spaces. The solution for the vector field relies on multivariate occupation kernel functions associated with the trajectories and scales linearly with the dimension of the state space. We validate through experiments on a variety of simulated and …

The Matryoshka Doll Prior – Principled Multiplicity Correction In Bayesian Model Comparison, Andrew Womack, Daniel Taylor-Rodríguez, Claudio Fuentes

Mathematics and Statistics Faculty Publications and Presentations

This paper introduces a general and principled construction of model space priors with a focus on regression problems. The proposed formulation regards each model as a “local” null hypothesis whose alternatives are the set of models that nest it. Assuming constant ratio of prior probabilities for any “local” null and its alternatives provides a natural isomorphism of model spaces (like a matryoshka doll), constituting an intuitive way to correct for multiplicity in Bayesian model selection and averaging problems. This isomorphism yields the Poisson distribution as the unique limiting distribution over model dimension under mild assumptions. We compare this model space …

Foundation Models Boost Low-Level Perceptual Similarity Metrics, Abhijay Ghildyal, Nabajeet Barman, Saman Zadtootaghaj

Computer Science Faculty Publications and Presentations

For full-reference image quality assessment (FR-IQA) using deep-learning approaches, the perceptual similarity score between a distorted image and a reference image is typically computed as a distance measure between features extracted from a pretrained CNN or more recently, a Transformer network. Often, these intermediate features require further fine-tuning or processing with additional neural network layers to align the final similarity scores with human judgments. So far, most IQA models based on foundation models have primarily relied on the final layer or the embedding for the quality score estimation. In contrast, this work explores the potential of utilizing the intermediate features …

“Lost-In-The-Later”: Framework For Quantifying Contextual Grounding In Large Language Models, Yufei Tao, Adam Hiatt, Rahul Seetharaman, Ameeta Agrawal

Computer Science Faculty Publications and Presentations

Large language models (LLMs) are capable of leveraging both contextual and parametric knowledge but how they prioritize and integrate these sources remains underexplored. We introduce CoPE, a novel framework for systematically quantifying contextual grounding in LLMs. CoPE distinguishes between contextual knowledge (CK) and parametric knowledge (PK), enabling fine-grained attribution across languages and tasks. Using our newly created MultiWikiAtomic dataset in English, Spanish, and Danish, we analyze how LLMs integrate context, prioritize information, and incorporate PK in open-ended question answering. We find that across models and languages, only around 50 to 76 percent of outputs are grounded in the given context, …

Drta: Dynamic Reward Scaling For Reinforcement Learning In Time Series Anomaly Detection, Bahareh Golchin, Banafsheh Rekabdar, Kunpeng Liu

Computer Science Faculty Publications and Presentations

Anomaly detection in time series data is important for applications in finance, healthcare, sensor networks, and industrial monitoring. Traditional methods usually struggle with limited labeled data, high false-positive rates, and difficulty generalizing to novel anomaly types. To overcome these challenges, we propose a reinforcement learning-based framework that integrates dynamic reward shaping, Variational Autoencoder (VAE), and active learning, called DRTA. Our method uses an adaptive reward mechanism that balances exploration and exploitation by dynamically scaling the effect of VAE-based reconstruction error and classification rewards. This approach enables the agent to detect anomalies effectively in low-label systems while maintaining high precision and …

Scitopic: Enhancing Topic Discovery In Scientific Literature Through Advanced Llm, Pengjiang Li, Zaitian Wang, Xinhao Zhang, Ran Zhang, Lu Jiang, Pengfei Wang, Yuanchun Zhou

Computer Science Faculty Publications and Presentations

Topic discovery in scientific literature provides valuable insights for researchers to identify emerging trends and explore new avenues for investigation, facilitating easier scientific information retrieval. Many machine learning methods, particularly deep embedding techniques, have been applied to discover research topics. However, most existing topic discovery methods rely on word embedding to capture the semantics and lack a comprehensive understanding of scientific publications, struggling with complex, high-dimensional text relationships. Inspired by the exceptional comprehension of textual information by large language models (LLMs), we propose an advanced topic discovery method enhanced by LLMs to improve scientific topic identification, namely SciTopic. Specifically, we …

Unsupervised Oct Image Interpolation Using Deformable Registration And Generative Models, Shuwen Wei, Samuel W. Remedios, Bruno Jedynak, Peter Calabresi, Shiv Saidha, Aaron Carass, Multiple Additional Authors

Mathematics and Statistics Faculty Publications and Presentations

Optical coherence tomography (OCT) images are often acquired as highly anisotropic volumes, where the scanning step is dense along the fast axis but sparse along the slow axis. This affects image analysis, such as image registration for longitudinal alignment. To create more isotropic volumes, bicubic interpolation can be used along the slow axis, but it generally produces blurry features. Registration-based interpolation can reduce blurriness, but often fails to generate realistic OCT images. Deep generative models can sample realistic images, but lack the structural consistency constraints required for interpolation. In this paper, we propose an unsupervised image interpolation method that combines …

Descending An Inclined Plane With A Large Language Model, Justin C. Dunlap, Ryan Sissons, Ralf Widenhorn

Physics Faculty Publications and Presentations

[This paper is part of the Focused Collection in Artificial Intelligence Tools in Physics Teaching and Physics Education Research.] We present a study in which a version of a common conservation of mechanical energy introductory physics problem, an object released on an inclined plane, is given to OpenAI’s GPT-4 large language model (LLM). We investigate how different permutations of object, action verb, and property of the incline impact the responses of the LLM. The problem setup and prompting was left purposefully minimal, requiring the LLM to state multiple assumptions to justify the final answer. We specifically studied how different keywords …

A Proposed Ehrenfeucht-Fraïssé Game Model For Natural Language Processing Generative Adversarial Networks, Don Li

Anthós

Large Language Models (LLM’s) (e.g., ChatGPT) constitute both a significant research area and commercial application of AI. Current major LLM’s are built on Generative Pre-Trained Transformer (GPT) neural network architecture to perform natural language processing (NLP) tasks. Generative Adversarial Network (GAN) is another popular neural network architecture, which leverages a zero-sum game between constituent neural networks within the architecture to train the GAN, and is widely used for visual data applications. This article proposes a new GAN architecture for NLP: an EF-GAN whose underlying algorithm uses Ehrenfeucht–Fraïssé (EF) games, a game-theoretic approach from model theory to determine elementary equivalence of …

Consumption Pressure In Estuaries Peaks At Intermediate Salinities, Catherine E. De Rivera, Amy A. Larson, Benjamin G. Rubinoff, Luna R. Soto, Seth L. Wright, Edwin D. Grosholz, Gregory M. Ruiz, Andrew L. Chang

Environmental Science and Management Faculty Publications and Presentations

The nature and strength of biotic interactions change along stress gradients, but the importance of these interactions across estuarine gradients is under studied. Here, we examined how consumption varies across estuarine salinity gradients by deploying standardized baits (‘squidpops’) to measure consumption pressure along the gradients of five estuaries in Oregon, USA. The relationship between consumption and stress was nonlinear: consumption pressure peaked slightly at mid salinity and decreased at low salinity, especially as temperature increased, in the five estuaries studied. This finding does not support either of two existing models for consumption across gradients, including the Consumer Stress Model and …

Fueling Adaptation: Collaborative Wildfire Governance In The Colorado Front Range Landscape, Mahmood Muttaqee, Cody Evers, Lindsay K. Campbell, Max Nielsen-Pincus, Olivia Awbrey, Kathryn Hixson, Michelle Johnson, Erika Svendsen

Environmental Science and Management Faculty Publications and Presentations

Collaborative governance plays a critical role in wildfire risk mitigation efforts. This study examines collaborative wildfire governance networks within the Colorado Front Range, a landscape that was designated under the USDA Forest Service's Wildfire Crisis Strategy (WCS). Collaborative governance networks in the Colorado Front Range leverage partnerships across multiple organizational actors including federal, state, and local agencies, non-profits, fire districts, and research institutions. Through survey data collected from 281 respondents revealing a collaborative network of 401 organizations, the study aimed to identify the structure, capacity, and gaps in governance networks addressing wildfire risk in the Colorado Front Range. Survey results …

Urban Heterogeneity Drives Dissolved Organic Matter Sources, Transport, And Transformation From Local To Macro Scales, Rebecca L. Hale, Kristina G. Hopkins, Krista A. Capps, John S. Kominoski, Jennifer L. Morse, Allison H. Roy, Shuo Chen, Annika Quick, Andrew J. Blinn, Liz Ortiz Muñoz, Gwendolynn Folk

Environmental Science and Management Faculty Publications and Presentations

Urbanization reshapes dissolved organic matter (DOM) sources, transport, and transformations through changes in vegetation, hydrology, and management of waste and water. Yet the impacts of urbanization on DOM are variable within and among cities. Predicting heterogeneous responses to urbanization is challenged by diverse human activities and underlying biophysical variation along stream networks. Using data from the 486 largest urban areas in the continental United States and seven focal cities, we identified macro and local scale urban gradients in social, built, and biophysical factors that are expected to shape DOM. We used these gradients and the literature to develop hypotheses about …

Citizen’S Rare Plant Watch, Nora Dunkirk, Gabriel Campbell, Kris Freitag

Rae Selling Berry Seed Bank and Plant Conservation Program

Rare species data for plants is lacking, hindering land management decisions and conservation outcomes. With Citizen's Rare Plant Watch, we empower citizen scientists to provide the “eyes on the ground” and collect data needed for conservation decision making in Oregon.

Aquatic Invasive Species Survey And Treatment On Lake Umatilla And Lake Celilo 2023-2024 Report, Garbiel E. Campbell, Jacob Rose

Center for Lakes and Reservoirs Publications and Presentations

Flowering Rush (Butomus umbellatus) is an invasive aquatic plant in the Pacific Northwest that threatens salmon habitat. The Center for Lakes and Reservoirs staff surveyed for this and other aquatic species from 2023 and 2024 in the Columbia River in Lake Umatilla and Lake Celilo. This document summarizes their survey efforts including their protocols, data, and small-scale removal efforts.

Permafrost And Structural Controls On Holocene Bedrock Landslide Occurrence Around Eyjafjörður, North‐Central Iceland, Adam Booth, Halldór G. Pétursson

Geology Faculty Publications and Presentations

Rapid, transient, landscape‐scale changes associated with deglaciation can condition slopes for failure and trigger bedrock landslides. However, the mechanisms leading to paleo rock slope failures following the last glacial period are challenging to infer because observations of how both landsliding and potential driving factors were distributed in space and time are limited. Here, we map and analyze the spatiotemporal pattern of 676 post‐glacial bedrock landslides around Eyjafjörður in north‐central Iceland using 2‐m resolution digital elevation data generated from optical stereo satellite imagery. Frequency‐ratio analysis demonstrates that after controlling for slope, landslides are most overrepresented within 2.6 km horizontal distances from …

The Petrology Of The Ozerki Meteorite Constrained By Electron Backscatter Diffraction And X-Ray Computed Tomography, Alice Macente, Luke Daly, Sammy Griffin, Maria Gritsevich, Jarmo Moilanen, Josh Franz Einsle, Patrick Trimby, Chris Mulcahy, Jonathan Moffat, Alexander M. Ruzicka

Geology Faculty Publications and Presentations

Combining electron backscatter diffraction (EBSD) with X-ray computed tomography (XCT) offers a comprehensive approach to investigate shock deformation and rock texture in meteorites, yet such integration remains uncommon. In this study, we demonstrate the synergistic potential of XCT and EBSD in revealing deformation metrics, thereby enhancing our understanding of petrofabric strength and shock-induced deformation. Our analysis focuses on the Ozerki (L6, S4/5, W0) meteorite fall, which was instrumentally observed on June 21, 2018, and subsequently recovered by the Ural's branch of the Russian Fireball Network (UrFU) recovery expedition a few days later. The trajectory analysis conducted by the Finnish Fireball …

Global Patterns Of Coseismic Landslide Runout Mobility Differ From Aseismic Landslide Trends, Alex R. R. Grant, Natalie K. Culhane

Geology Faculty Publications and Presentations

Coseismic landslides significantly contribute to human and economic losses during and immediately following earthquakes, yet very little data on the runout of such landslides exist. While well-established behavior of aseismic (e.g., hydrologically triggered) landslide runout mobility suggests strong correlation between landslide size and mobility, limited studies of coseismic landslide runout find conflicting mobility trends. We present a global dataset of runout lengths produced from a new automated method for estimating landslide runout, developed and validated using 1726 manually mapped landslides from five unique earthquakes. We then apply the automated runout tool to 23 global earthquake-induced landslide inventories, producing a compiled …

The Geography Of Rock Glaciers In The American West, Allison Reese Trcka, Andrew G. Fountain, Bryce Allen Glenn

Geology Faculty Publications and Presentations

Rock glaciers are relatively common in conterminous United States west of the 100° meridian, yet their distribution and number are not well known. Using aerial and satellite imagery, we inventoried the population of rock glaciers, yielding 2,257 rock glaciers totaling 234.28 km² with an area of uncertainty of 7 percent. Of these, 1,561 (155.44 km²) are considered intact (moving) and 696 (76.84 km²) relict. Based on published estimates of ice content, total water-equivalent volume is roughly 2.52 km³. Rock glacier elevations increase from the northwest (Washington) to the southeast (Colorado), following a spatial …

The Tyrant King Of The Auction House: The Disastrous “Fine Artification” Of The Dinosaur, Amelia C. Kolstad

Anthós

This article examines how fine art auctioneers market dinosaur fossils as art pieces to wealthy collectors, depriving the paleontological field of specimens for scientific research. Drawing from recent examples of fossils sold at fine art auctions, this article notes how rising auction prices for specimens as ‘art’ prevent museums and research institutions from acquiring specimens for research and restoration. Kolstad concludes that the art auction houses’ marketing of fossils as art is detrimental to paleontological research. To protect fossils and their scientific value, Kolstad proposes alternatives, such as selling replicas to collectors and highlighting the importance of original specimens for …

Clea Labs For Planetary Astronomy Bottled For Silicon Macintosh, Yuktha Shetty, Hansika Reddy Bandela, Jack C. Straton

Physics Faculty Publications and Presentations

Project CLEA -- CONTEMPORARY LABORATORY EXPERIENCES IN ASTRONOMY -- develops laboratory exercises that illustrate modern astronomical techniques using digital data and color images.

These Labs were developed for Windows, but the authors have bottled them for Silicon Macintosh computers.

Clea Labs For Stellar And Cosmological Astronomy Bottled For Silicon Macintosh, Hansika Reddy Bandela, Yuktha Shetty, Jack C. Straton

Physics Faculty Publications and Presentations

Project CLEA -- CONTEMPORARY LABORATORY EXPERIENCES IN ASTRONOMY -- develops laboratory exercises that illustrate modern astronomical techniques using digital data and color images.

These Labs were developed for Windows, but the authors have bottled them for Silicon Macintosh computers.

Clea Labs For Planetary, Stellar, And Cosmological Astronomy Configured For 64-Bit Windows 11+ Machines, Itan Altamirano, Jack C. Straton

Physics Faculty Publications and Presentations

Project CLEA -- CONTEMPORARY LABORATORY EXPERIENCES IN ASTRONOMY -- develops laboratory exercises that illustrate modern astronomical techniques using digital data and color images.

These Labs were developed for 32-bit Windows machines but PSU student Itan Altamirano reconfigured installation programs to bypass steps that ask for DVDs and allowing these 32-bit programs to work on 64-bit Windows 11+ computers.

The Stochastic Occupation Kernel (Sock) Method For Learning Stochastic Differential Equations, M. L. Wells, Kamel Lahouel, Bruno Jedynak

Mathematics and Statistics Faculty Publications and Presentations

We present a novel kernel-based method for learning multivariate stochastic differential equations (SDEs). The method follows a two-step procedure: we first estimate the drift term function, then the (matrix-valued) diffusion function given the drift. Occupation kernels are integral functionals on a reproducing kernel Hilbert space (RKHS) that aggregate information over a trajectory. Our approach leverages vector-valued occupation kernels for estimating the drift component of the stochastic process. For diffusion estimation, we extend this framework by introducing operator-valued occupation kernels, enabling the estimation of an auxiliary matrix-valued function as a positive semi-definite operator, from which we readily derive the diffusion estimate. …