arXiv subjects

Ryan Kelly

Publications and source records attributed to Ryan Kelly.

5 recordsLinked to original sources

arXiv2025

Interaction of a Vortex Pair with a Polymeric Fluid Layer

The interaction of vortical structures with boundaries has been extensively studied in Newtonian fluids, where conditions such as no slip walls, free surfaces, or contaminated surfaces dictate whether vortices rebound, dissipate, or generate secondary structures. In this work, we investigate a related but fundamentally different problem: the interaction of a vortex pair with a finite, non uniform layer of polymeric fluid. Numerical simulations employing the finitely extensible nonlinear elastic Peterlin model are used to examine the effects of polymer concentration, relaxation time, polymer layer thickness, and maximum polymer extension on the evolution of kinetic energy and enstrophy. The results show that, while the polymeric fluid dissipates vortical motion, vortex polymer layer interactions can also generate new coherent structures. In particular, the formation of secondary and tertiary vortices coincides with transient increases in kinetic energy, a behavior absent in the Newtonian case. Unlike classical vortex boundary interactions, where the primary vortex survives, we find that under certain conditions it completely dissipates upon interaction with the polymer layer. These findings emphasize that fluids with non-uniform polymer concentrations, act not only as dissipative agents but also as sources of vorticity, extending the traditional view of polymer induced drag reduction and providing new insight into vortex polymer interactions.

physics.flu-dyn↗

arXiv2024

The Statistical Accuracy of Neural Posterior and Likelihood Estimation

Neural posterior estimation (NPE) and neural likelihood estimation (NLE) are machine learning approaches that provide accurate posterior, and likelihood, approximations in complex modeling scenarios, and in situations where conducting amortized inference is a necessity. While such methods have shown significant promise across a range of diverse scientific applications, the statistical accuracy of these methods is so far unexplored. In this manuscript, we give, for the first time, an in-depth exploration on the statistical behavior of NPE and NLE. We prove that these methods have similar theoretical guarantees to common statistical methods like approximate Bayesian computation (ABC) and Bayesian synthetic likelihood (BSL). While NPE and NLE methods are just as accurate as ABC and BSL, we prove that this accuracy can often be achieved at a vastly reduced computational cost, and will therefore deliver more attractive approximations than ABC and BSL in certain problems. We verify our results theoretically and in several examples from the literature.

stat.ML↗

arXiv2023

Generation of Laminar Vortex Rings by an Impulsive Body Force

It is shown that laminar vortex rings can be generated by impulsive body forces having particular spatial and temporal characteristics. The method produces vortex rings in a fluid initially at rest, and once generated, the flow field automatically satisfies the boundary conditions and is divergence-free. Numerical simulations and analytical models show that the strength of these rings can be accurately predicted by considering diffusion alone, despite the nonlinear nature of the generation process. A particularly simple model, which approximates the source of vorticity within vertical slabs, is proposed. This model predicts the ring circulation almost as accurately as a model which uses the exact geometry of the source of vorticity. It is found that when the duration of the force is less than a time scale based on the force radius and fluid viscosity, the ring circulation can be predicted accurately using an inviscid model.

physics.flu-dyn↗

arXiv2021

GAVIN: Gaze-Assisted Voice-Based Implicit Note-taking

Annotation is an effective reading strategy people often undertake while interacting with digital text. It involves highlighting pieces of text and making notes about them. Annotating while reading in a desktop environment is considered trivial but, in a mobile setting where people read while hand-holding devices, the task of highlighting and typing notes on a mobile display is challenging. In this paper, we introduce GAVIN, a gaze-assisted voice note-taking application, which enables readers to seamlessly take voice notes on digital documents by implicitly anchoring them to text passages. We first conducted a contextual enquiry focusing on participants' note-taking practices on digital documents. Using these findings, we propose a method which leverages eye-tracking and machine learning techniques to annotate voice notes with reference text passages. To evaluate our approach, we recruited 32 participants performing voice note-taking. Following, we trained a classifier on the data collected to predict text passage where participants made voice notes. Lastly, we employed the classifier to built GAVIN and conducted a user study to demonstrate the feasibility of the system. This research demonstrates the feasibility of using gaze as a resource for implicit anchoring of voice notes, enabling the design of systems that allow users to record voice notes with minimal effort and high accuracy.

cs.HC↗

arXiv2016

A Model-Based Approach to Wildland Fire Reconstruction Using Sediment Charcoal Records

Lake sediment charcoal records are used in paleoecological analyses to reconstruct fire history including the identification of past wildland fires. One challenge of applying sediment charcoal records to infer fire history is the separation of charcoal associated with local fire occurrence and charcoal originating from regional fire activity. Despite a variety of methods to identify local fires from sediment charcoal records, an integrated statistical framework for fire reconstruction is lacking. We develop a Bayesian point process model to estimate probability of fire associated with charcoal counts from individual-lake sediments and estimate mean fire return intervals. A multivariate extension of the model combines records from multiple lakes to reduce uncertainty in local fire identification and estimate a regional mean fire return interval. The univariate and multivariate models are applied to 13 lakes in the Yukon Flats region of Alaska. Both models resulted in similar mean fire return intervals (100-350 years) with reduced uncertainty under the multivariate model due to improved estimation of regional charcoal deposition. The point process model offers an integrated statistical framework for paleo-fire reconstruction and extends existing methods to infer regional fire history from multiple lake records with uncertainty following directly from posterior distributions.

stat.AP↗