Introduction to Separable Simplicial Algebras
We study separable simplicial algebras and Azumaya simplicial algebras.
math.RA↗
arXiv subjects
Eric Zhang
Publications and source records attributed to Eric Zhang.
Evaluating a Large Language Model for Radiotherapy Patient Scheduling under Realistic Clinical and Operational Constraints: a Feasibility Study
Radiotherapy scheduling must balance clinical constraints and operational objectives. We evaluated the feasibility of using a large language model (LLM) to generate treatment schedules through simulation and retrospective clinical comparison. A one-year simulation modeled 1,400 new patients across 12 treatment categories and three linear accelerators (LINACs). Structured prompts provided ChatGPT with patient information, schedule status, clinical rules, and operational objectives. Scenarios assessed treatment-time consistency, LINAC continuity, gap reduction, and handling of infeasible requests. LLM schedules were compared with heuristic and constraint programming (CP-SAT) approaches under two objective settings. A six-week retrospective analysis included 2,043 clinical appointments. LLM-generated schedules met all predefined feasibility constraints in the evaluated scenarios, with 98.6-98.8% of fractions within preferred treatment-time windows. Prioritizing LINAC continuity reduced switching from 52.4 +/- 2.6% to 9.4 +/- 0.6%. Gap-constrained temporal relaxation reduced average daily gaps to 47.1 +/- 8.5 minutes and Friday gaps to 89.2 +/- 18.4 minutes while maintaining low overtime. The framework identified predefined violations in infeasible requests and suggested corrective actions. Heuristic and CP-SAT methods also generated feasible schedules, with performance dependent on objective weighting. In retrospective evaluation, all approaches scheduled appointments without overtime; LLM strategies reduced mean gaps by 50.8-77.1% and improved treatment-time consistency from 86.7% to 99.5-99.9%. These findings support further evaluation of structured LLM scheduling as a human-in-the-loop decision-support approach for radiotherapy.
physics.med-ph↗
AM-Bench: A Unified Taxonomy and Evaluation Suite for Agentic Misalignment
As LLM-based agents are deployed with increasing autonomy in real-world environments, a critical failure mode emerges: agentic misalignment, where agents choose harmful actions over failure. However, existing safety benchmarks primarily evaluate whether agents comply with explicitly harmful instructions, leaving the conditions that cause agents to choose harmful actions poorly understood. To address this limitation, we propose AM-Bench, a unified taxonomy and 35 scenario evaluation benchmark spanning seven industries for agentic misalignment grounded in the fraud triangle, a well-established criminological framework explaining goal-directed misconduct. We adapt this framework to agentic environments and construct our evaluation suite and detection systems along three dimensions: perceived threat, perceived opportunity, and reasoning. We operationalize each dimension into structured subcategories rooted in prior theory: five threat categories derived from the instrumental convergence thesis, seven opportunity categories drawn from the Emergent Strategic Reasoning Risks (ESRR) taxonomy, and a novel four-question binary reasoning framework synthesized from prior literature. Our evaluation of 10 frontier models reveals a 62-percentage-point gap in misalignment across scenarios, with Gemini 3 Flash and Gemini 3.1 Pro exhibiting the highest levels of agentic misalignment while Claude Sonnet 4.6 displays near-zero rates across all conditions. Our reasoning framework further reveals that models may exhibit harmful behaviours without explicit situational awareness or explanation, complicating both mitigation and detection. AM-Bench scenarios and code are available at https://github.com/AI4Collaboration/Agentic-Misalignment-Bench
cs.AI↗
Meter-scale quantum interconnects between large quantum processors
Modular quantum architecture requires long-distance quantum interconnects to stitch together individual processor modules. Such interconnects, long-distance couplers or L-couplers, must support coherent quantum operation over meter-scale distances and scale to multiple connections between any two modules. Here we demonstrate four L-couplers operating between two 120-qubit IBM Nighthawk R1 superconducting qubit processors. Each L-coupler joins two Nighthawk lattice qubits through two frequency-tunable L-qubits, two RF-SQUID based "i2m" couplers, and 80 cm of superconducting aluminum coax. Each cable installs perpendicular to the processor surface, to enable denser connections, and is joined to the chip by an indium-to-indium room temperature pressure bond made after both modules were mounted. L-coupler modes near 5 GHz, with internal quality factors of 0.86 to 1.09e6, carry relay-based state transfer at a mean network benchmarking fidelity of 99.00%, with a best link of 99.29 +/- 0.07%. We then realize distant CZ gates between lattice qubits in separate modules at a mean fidelity of 94.75%, with a best link of 96.42 +/- 0.28%, and use them to generate entangled states across the two processors, including a demonstration of simultaneous Bell pair generation using two adjacent L-couplers. These four high-fidelity L-couplers establish a practical path to scalable quantum interconnects for modular quantum computing.
quant-ph↗
The Entangling of Supernova Feedback Impacts with Coarsening Simulation Resolution
It is often understood that supernova (SN) feedback in galaxies is responsible for regulating star formation (SF) and generating gaseous outflows. However, a detailed look at the small-scale effects of SNe on the interstellar medium (ISM) in simulations shows that the macroscopic processes of SF suppression and outflow generation proceed in distinct channels. We demonstrate this finding in two independent simulations of isolated dwarf galaxies with very high (m_gas ~ Msun) numerical resolution, LYRA and RIGEL. Our findings suggest that the macroscopic effect of a given SN on the galaxy is best predicted by its local density. Outflows are driven by SNe in diffuse regions expanding to their cooling radii on large (~kpc) scales, while dense SF regions are disrupted in a localized (~pc) manner. However, these separate feedback channels are only distinguishable at very high resolutions capable of following mass scales \lesssim 10^2 \msun. When averaging on coarser scales, ISM densities are greatly mis-estimated, and variations between different SF and SNe-affected regions are severely washed out. It therefore cannot be __self-consistently__ determined, from coarse-resolution information __alone__, (1) whether a SN tends to contribute to outflows or direct SF suppression, and (2) the rate of SF in a given region. In particular, commonly used parameters in coarse-resolution (subgrid) models, such as the SN cooling radius and SF density threshold, may require more detailed treatments informed by high-resolution studies.
astro-ph.GA↗
Scalable Detection of Fossil Palynomorphs in Multifocal Digital Microscopy Images
Palynomorphs (microscopic, organic-walled fossils such as pollen, spores, and dinoflagellates) are important high-resolution records of past climates and are critical to the study of ancient ecosystems. Existing methods rely on manual analysis of high-resolution, multifocal digital microscopy images, which is slow and time-consuming and requires researchers to compromise on the scale of their investigations. To the best of our knowledge, our work proposes the first ever scalable end-to-end pipeline for automated palynomorph detection in whole slide images that addresses this bottleneck through: (1) efficient methods for decomposing and compressing digitized multifocal microscope slide images into tractable 2-dimensional tiles for analysis; (2) benchmarking modern object detection models, including RF-DETR, for the detection of palynomorphs, achieving an AP@50 of 0.879; (3) an efficient algorithm for the synthesis of detection outputs across large-scale, high-resolution images; and (4) an I/O optimization resulting in faster inference time. Our methods drastically reduce the time required for palynomorph detection in a single slide from often days of manual inspection to under one hour of automated analysis, enabling palynological research at a substantially greater scale.
cs.CV↗
Unforgettable Generalization in Language Models
When language models (LMs) are trained to forget (or "unlearn'') a skill, how precisely does their behavior change? We study the behavior of transformer LMs in which tasks have been forgotten via fine-tuning on randomized labels. Such LMs learn to generate near-random predictions for individual examples in the "training'' set used for forgetting. Across tasks, however, LMs exhibit extreme variability in whether LM predictions change on examples outside the training set. In some tasks (like entailment classification), forgetting generalizes robustly, and causes models to produce uninformative predictions on new task instances; in other tasks (like physical commonsense reasoning and scientific question answering) forgetting affects only the training examples, and models continue to perform the "forgotten'' task accurately even for examples very similar to those that appeared in the training set. Dataset difficulty is not predictive of whether a behavior can be forgotten; instead, generalization in forgetting is (weakly) predicted by the confidence of LMs' initial task predictions and the variability of LM representations of training data, with low confidence and low variability both associated with greater generalization. Perhaps most surprisingly, random-label forgetting appears to be somewhat insensitive to the contents of the training set: for example, models trained on science questions with random labels continue to answer other science questions accurately, but begin to produce random labels on entailment classification tasks. Finally, we show that even generalizable forgetting is shallow: linear probes trained on LMs' representations can still perform tasks reliably after forgetting. Our results highlight the difficulty and unpredictability of performing targeted skill removal from models via fine-tuning.
cs.LG↗
Shared Parameter Subspaces and Cross-Task Linearity in Emergently Misaligned Behavior
Recent work has discovered that large language models can develop broadly misaligned behaviors after being fine-tuned on narrowly harmful datasets, a phenomenon known as emergent misalignment (EM). However, the fundamental mechanisms enabling such harmful generalization across disparate domains remain poorly understood. In this work, we adopt a geometric perspective to study EM and demonstrate that it exhibits a fundamental cross-task linear structure in how harmful behavior is encoded across different datasets. Specifically, we find a strong convergence in EM parameters across tasks, with the fine-tuned weight updates showing relatively high cosine similarities, as well as shared lower-dimensional subspaces as measured by their principal angles and projection overlaps. Furthermore, we also show functional equivalence via linear mode connectivity, wherein interpolated models across narrow misalignment tasks maintain coherent, broadly misaligned behavior. Our results indicate that EM arises from different narrow tasks discovering the same set of shared parameter directions, suggesting that harmful behaviors may be organized into specific, predictable regions of the weight landscape. By revealing this fundamental connection between parametric geometry and behavioral outcomes, we hope our work catalyzes further research on parameter space interpretability and weight-based interventions.
cs.LG↗
Contrastive Dimension Reduction: A Systematic Review
Contrastive dimension reduction (CDR) methods aim to extract signal unique to or enriched in a treatment (foreground) group relative to a control (background) group. This setting arises in many scientific domains, such as genomics, imaging, and time series analysis, where traditional dimension reduction techniques such as Principal Component Analysis (PCA) may fail to isolate the signal of interest. In this review, we provide a systematic overview of existing CDR methods. We propose a pipeline for analyzing case-control studies together with a taxonomy of CDR methods based on their assumptions, objectives, and mathematical formulations, unifying disparate approaches under a shared conceptual framework. We highlight key applications and challenges in existing CDR methods, and identify open questions and future directions. By providing a clear framework for CDR and its applications, we aim to facilitate broader adoption and motivate further developments in this emerging field.
stat.ME↗
Deploying UDM Series in Real-Life Stuttered Speech Applications: A Clinical Evaluation Framework
Stuttered and dysfluent speech detection systems have traditionally suffered from the trade-off between accuracy and clinical interpretability. While end-to-end deep learning models achieve high performance, their black-box nature limits clinical adoption. This paper looks at the Unconstrained Dysfluency Modeling (UDM) series-the current state-of-the-art framework developed by Berkeley that combines modular architecture, explicit phoneme alignment, and interpretable outputs for real-world clinical deployment. Through extensive experiments involving patients and certified speech-language pathologists (SLPs), we demonstrate that UDM achieves state-of-the-art performance (F1: 0.89+-0.04) while providing clinically meaningful interpretability scores (4.2/5.0). Our deployment study shows 87% clinician acceptance rate and 34% reduction in diagnostic time. The results provide strong evidence that UDM represents a practical pathway toward AI-assisted speech therapy in clinical environments.
cs.SD↗
A Comparative Study of Controllability, Explainability, and Performance in Dysfluency Detection Models
Recent advances in dysfluency detection have introduced a variety of modeling paradigms, ranging from lightweight object-detection inspired networks (YOLOStutter) to modular interpretable frameworks (UDM). While performance on benchmark datasets continues to improve, clinical adoption requires more than accuracy: models must be controllable and explainable. In this paper, we present a systematic comparative analysis of four representative approaches--YOLO-Stutter, FluentNet, UDM, and SSDM--along three dimensions: performance, controllability, and explainability. Through comprehensive evaluation on multiple datasets and expert clinician assessment, we find that YOLO-Stutter and FluentNet provide efficiency and simplicity, but with limited transparency; UDM achieves the best balance of accuracy and clinical interpretability; and SSDM, while promising, could not be fully reproduced in our experiments. Our analysis highlights the trade-offs among competing approaches and identifies future directions for clinically viable dysfluency modeling. We also provide detailed implementation insights and practical deployment considerations for each approach.
cs.AI↗
Revisiting Rule-Based Stuttering Detection: A Comprehensive Analysis of Interpretable Models for Clinical Applications
Stuttering affects approximately 1% of the global population, impacting communication and quality of life. While recent advances in deep learning have pushed the boundaries of automatic speech dysfluency detection, rule-based approaches remain crucial for clinical applications where interpretability and transparency are paramount. This paper presents a comprehensive analysis of rule-based stuttering detection systems, synthesizing insights from multiple corpora including UCLASS, FluencyBank, and SEP-28k. We propose an enhanced rule-based framework that incorporates speaking-rate normalization, multi-level acoustic feature analysis, and hierarchical decision structures. Our approach achieves competitive performance while maintaining complete interpretability-critical for clinical adoption. We demonstrate that rule-based systems excel particularly in prolongation detection (97-99% accuracy) and provide stable performance across varying speaking rates. Furthermore, we show how these interpretable models can be integrated with modern machine learning pipelines as proposal generators or constraint modules, bridging the gap between traditional speech pathology practices and contemporary AI systems. Our analysis reveals that while neural approaches may achieve marginally higher accuracy in unconstrained settings, rule-based methods offer unique advantages in clinical contexts where decision auditability, patient-specific tuning, and real-time feedback are essential.
cs.AI↗
Contrastive CUR: Interpretable Joint Feature and Sample Selection for Case-Control Studies
Dimension reduction is an essential tool for analyzing high dimensional data. Most existing methods, including principal component analysis (PCA), as well as their extensions, provide principal components that are often linear combinations of features, which are often challenging to interpret. CUR decomposition, another matrix decomposition technique, is a more interpretable and efficient alternative, offers simultaneous feature and sample selection. Despite this, many biomedical studies involve two groups: a foreground (treatment or case) group and a background (control) group, where the objective is to identify features unique to or enriched in the foreground. This need for contrastive dimension reduction is not well addressed by existing CUR methods, nor by contrastive approaches rooted in PCAs. Furthermore, they fail to address a key challenge in biomedical studies: the need for selecting samples unique to the foreground. In this paper, we address this gap by proposing a Contrastive CUR (CCUR), a novel method specifically designed for case-control studies. Through extensive experiments, we demonstrate that CCUR outperforms existing techniques in isolating biologically relevant features as well as identifying sample-specific responses unique to the foreground, offering deeper insights into case-control biomedical data.
stat.ME↗
Can Gradient Descent Simulate Prompting?
There are two primary ways of incorporating new information into a language model (LM): changing its prompt or changing its parameters, e.g. via fine-tuning. Parameter updates incur no long-term storage cost for model changes. However, for many model updates, prompting is significantly more effective: prompted models can generalize robustly from single examples and draw logical inferences that do not occur under standard fine-tuning. Can models be modified so that fine-tuning does emulate prompting? This paper describes a method for meta-training LMs such that gradient updates emulate the effects of conditioning on new information. Our approach uses tools from gradient-based meta-learning but uses an LM's own prompted predictions as targets, eliminating the need for ground-truth labels. Subsequent gradient descent training recovers some (and occasionally all) of prompted model performance -- showing improvement on the ``reversal curse'' tasks, and answering questions about text passages after a single gradient update. These results suggest that, with appropriate initialization, gradient descent can be surprisingly expressive. Our results suggest new avenues for long-context modeling and offer insight into the generalization capabilities of gradient-based learning.
cs.CL↗
Bursty Star Formation in Dwarfs is Sensitive to Numerical Choices in Supernova Feedback Models
Simulations of galaxy formation are mostly unable to resolve the energy-conserving phase of individual supernova events, having to resort to subgrid models to distribute the energy and momentum resulting from stellar feedback. However, the properties of these simulated galaxies, including the morphology, stellar mass formed and the burstiness of the star formation history, are highly sensitive to numerical choices adopted in these subgrid models. Using the {\small SMUGGLE} stellar feedback model, we compute idealized simulations of a $M_{\rm vir} \sim 10^{10} \, \msun$ dwarf galaxy, a regime where most simulation codes predict significant burstiness in star formation, resulting in strong gas flows that lead to the formation of dark matter cores. We find that by varying only the directional distribution of momentum imparted from supernovae to the surrounding gas, while holding the total momentum per supernova constant, bursty star formation may be amplified or completely suppressed, and the total stellar mass formed can vary by as much as a factor of $\sim 3$. In particular, when momentum is primarily directed perpendicular to the gas disk, less bursty and lower overall star formation rates result, yielding less gas turbulence, more disky morphologies and a retention of cuspy dark matter density profiles. An improved understanding of the non-linear coupling of stellar feedback into inhomogeneous gaseous media is thus needed to make robust predictions for stellar morphologies and dark matter core formation in dwarfs independent of uncertain numerical choices in the baryonic treatment.
astro-ph.GA↗
Ragnarök: A Reusable RAG Framework and Baselines for TREC 2024 Retrieval-Augmented Generation Track
Did you try out the new Bing Search? Or maybe you fiddled around with Google AI~Overviews? These might sound familiar because the modern-day search stack has recently evolved to include retrieval-augmented generation (RAG) systems. They allow searching and incorporating real-time data into large language models (LLMs) to provide a well-informed, attributed, concise summary in contrast to the traditional search paradigm that relies on displaying a ranked list of documents. Therefore, given these recent advancements, it is crucial to have an arena to build, test, visualize, and systematically evaluate RAG-based search systems. With this in mind, we propose the TREC 2024 RAG Track to foster innovation in evaluating RAG systems. In our work, we lay out the steps we've made towards making this track a reality -- we describe the details of our reusable framework, Ragnarök, explain the curation of the new MS MARCO V2.1 collection choice, release the development topics for the track, and standardize the I/O definitions which assist the end user. Next, using Ragnarök, we identify and provide key industrial baselines such as OpenAI's GPT-4o or Cohere's Command R+. Further, we introduce a web-based user interface for an interactive arena allowing benchmarking pairwise RAG systems by crowdsourcing. We open-source our Ragnarök framework and baselines to achieve a unified standard for future RAG systems.
cs.IR↗
Versatile Diffusion: Text, Images and Variations All in One Diffusion Model
Recent advances in diffusion models have set an impressive milestone in many generation tasks, and trending works such as DALL-E2, Imagen, and Stable Diffusion have attracted great interest. Despite the rapid landscape changes, recent new approaches focus on extensions and performance rather than capacity, thus requiring separate models for separate tasks. In this work, we expand the existing single-flow diffusion pipeline into a multi-task multimodal network, dubbed Versatile Diffusion (VD), that handles multiple flows of text-to-image, image-to-text, and variations in one unified model. The pipeline design of VD instantiates a unified multi-flow diffusion framework, consisting of sharable and swappable layer modules that enable the crossmodal generality beyond images and text. Through extensive experiments, we demonstrate that VD successfully achieves the following: a) VD outperforms the baseline approaches and handles all its base tasks with competitive quality; b) VD enables novel extensions such as disentanglement of style and semantics, dual- and multi-context blending, etc.; c) The success of our multi-flow multimodal framework over images and text may inspire further diffusion-based universal AI research. Our code and models are open-sourced at https://github.com/SHI-Labs/Versatile-Diffusion.
cs.CV↗
Constraining a companion of the galactic center black hole, Sgr A*
We use 23 years of astrometric and radial velocity data on the orbit of the star S0-2 to constrain a hypothetical intermediate-mass black hole orbiting the massive black hole Sgr A* at the Galactic center. The data place upper limits on variations of the orientation of the stellar orbit (inclination, nodal angle, and pericenter) at levels between 0.02 and 0.07 degrees per year. We use a combination of analytic estimates and full numerical integrations of the orbit of S0-2 in the presence of a black-hole binary. For a companion IMBH whose semi-major axis $a_c$ is larger than that of S0-2 (1020 a.u.), we find that in the region between 1000 and 4000 a.u., a companion black hole with mass $m_c$ between $10^3$ and $10^5 M_\odot$ is excluded, with a boundary behaving as $a_c \sim m_c^{1/3}$. For a companion with $a_c < 1020$ a.u., we find that a black hole with mass between $10^3$ and $10^5 \, M_\odot$ is again excluded, with a boundary behaving as $a_c \sim m_c^{-1/2}$. These bounds arise from quadrupolar perturbations of the orbit of S0-2. However, significantly stronger bounds on the mass of an inner companion arise from the fact that the location of S0-2 is measured relative to the bright emission of Sgr A*. As a consequence, that separation is perturbed by the ``wobble'' of Sgr A* about the center of mass between it and the companion, leading to ``apparent'' perturbations of S0-2's orbit that also include a dipole component. The result is a set of bounds as small as $400 \, M_\odot$ at 200 a.u.; the numerical simulations suggest a bound from these effects varying as $a_c \sim m_c^{-1}$. We compare and contrast our results with those from a recent analysis by the GRAVITY collaboration.
astro-ph.GA↗