**Abstract:** The search for axion dark matter (ADM) relies on ultra-sensitive microwave cavity experiments, typically exhibiting noisy signals complicated by systematic drifts. Current analysis methods often struggle to definitively distinguish ADM signals from these background fluctuations. This paper introduces a novel, fully automated anomaly detection system leveraging Optimal Transport (OT) and Gaussian Process Regression (GPR) to enhance ADM signal identification. Our system, termed "Axion Anomaly Scout" (AAS), learns a baseline noise profile, detects deviations from this profile using OT distance calculations, and characterizes the associated uncertainty using GPR.

**Abstract:** This paper introduces a novel system for the automated detection of creep anomalies and prediction of remaining useful life (RUL) in advanced ceramic composite (ACC) components, specifically focusing on silicon carbide (SiC) reinforced silicon carbide (SiC/SiC) composites. Addressing the limitations of traditional creep analysis methods which often rely on empirical power-law models, this framework utilizes a dynamic kernel regression (DKR) technique coupled with a multi-modal data ingestion and evaluation pipeline.

**Abstract:** This paper proposes a novel computational framework, Hyperdynamic Reactive Dynamics (HRD), for accelerating simulations of chemical reactions involving transition states and complex molecular systems. HRD combines hyperdimensional symbolic regression (HDSR) with an active learning strategy to construct surrogate models of potential energy surfaces (PESs) from sparse, high-fidelity quantum chemical calculations. The framework leverages hypervector computations to efficiently explore the PES landscape, identify critical regions requiring further sampling, and iteratively refine the surrogate model, resulting in a 10x speedup in reaction dynamic simulations with minimal loss in accuracy compared to traditional trajectory-based methods.

**Abstract:** Reconstructing past atmospheric carbon isotope composition (δ¹³C) is critical for understanding climate dynamics and the carbon cycle throughout Earth’s history. Traditional methods relying on manual spectral analysis of ancient organic matter are time-consuming and prone to subjective interpretation. This research proposes a novel automated framework, leveraging advanced spectral deconvolution algorithms coupled with deep learning regression, to objectively and rapidly quantify δ¹³C fractionation from complex spectral datasets, offering a 10x improvement in throughput and representing a significant advancement in paleoatmospheric reconstruction.