arXiv preprints from January 1, 2026 through July 28, 2026 — 17:41:34 EST
The inherent safety alignment of Large Language Models (LLMs) is prone to erosion during fine-tuning, even when using seemingly innocuous datasets. While existing defenses attempt to mitigate this via data selection, they typically rely on heuristic, instance-level assessments that neglect the global geometry of the data distribution...
Large language models exhibit impressive reasoning capabilities yet frequently generate plausible but incorrect solutions, a phenomenon commonly termed hallucination. This paper investigates the effect of training objective composition on reasoning reliability through Direct Preference Optimization. Two complementary training signals...
Quantum thermometry aims to measure temperature in nanoscale quantum systems, paralleling classical thermometry. However, temperature is not a quantum observable, and most theoretical studies have therefore concentrated on analyzing fundamental precision limits set by the quantum Fisher information through the quantum Cramer-Rao...
Post-training activation compression is essential for deploying Large Language Models (LLMs) on resource-constrained hardware. However, standard methods like Singular Value Decomposition (SVD) are gradient-blind: they preserve high-variance dimensions regardless of their impact on factual knowledge preservation. We introduce...
Einstein-bumblebee gravity, as a class of massive non-minimally coupled vector-tensor theories, provides a useful framework for constraining Lorentz symmetry breaking through astrophysical observations, largely due to the existence of exact static and spherically symmetric black hole solutions. These solutions are typically obtained...
For a set of permutations $S\subseteq S_n$, consider the quasisymmetric generating function $$Q(S): = \sum_{w\in S}F_{n, \mathrm{Des}(w)},$$ where $\mathrm{Des}(w) := \{i\mid w(i)> w(i+1)\}$ is the descent set of $w$ and $F_{n, \mathrm{Des}(w)}$ is Gessel's fundamental quasisymmetric function. A set of permutations is said to be...
After Chern's conjecture on the discreteness of the constant scalar curvatures of compact minimal submanifolds $M^n$ in unit spheres $\mathbb{S}^{n+q}$, Z. Q. Lu proposed a conjecture regarding the second gap, based on his ingenious refinement of the known first gap theorem. This refinement unifies Simons' first gap theorem for...
Attosecond science is an emerging topic where chirality plays a central role. Here we demonstrate subjecting iodoacetylene, a geometrically achiral molecule, to a pair of simulated non-ionizing ultrafast circularly polarized laser pulses at the highest time resolution to date, by two orders of magnitude (3.87 attoseconds), of the...
Graph-based Retrieval-Augmented Generation (GraphRAG) mitigates hallucinations in Large Language Models (LLMs) by grounding them in structured knowledge. However, current GraphRAG methods are constrained by a prevailing \textit{build-then-reason} paradigm, which relies on a static, pre-constructed Knowledge Graph (KG). This paradigm...
This paper introduces Lie groups in degenerate geometric (Clifford) algebras that preserve four fundamental subspaces determined by the grade involution and reversion under the adjoint and twisted adjoint representations. We prove that these Lie groups can be equivalently defined using norm functions of multivectors applied in the...
Large Language Model (LLM) agents struggle with long-horizon software engineering tasks due to "Context Bloat." As interaction history grows, computational costs explode, latency increases, and reasoning capabilities degrade due to distraction by irrelevant past errors. Existing solutions often rely on passive, external summarization...
We investigate the spatial organization of dengue cases in the city of Recife, Brazil, from 2015 to 2024, using tools from statistical physics and topological data analysis. Reported cases are modeled as point clouds in a metric space, and their spatial connectivity is studied through Vietoris-Rips filtrations and zero-dimensional...
The proliferation of AI-generated imagery and sophisticated editing tools has rendered traditional forensic methods ineffective for cross-domain forgery detection. We present ForensicFormer, a hierarchical multi-scale framework that unifies low-level artifact detection, mid-level boundary analysis, and high-level semantic reasoning...
Reproducibility remains a challenge in machine learning research. While code and data availability requirements have become increasingly common, post-publication verification in journals is still limited and unformalized. This position paper argues that it is plausible for journals and conference proceedings to implement...
Scanning diamond nitrogen-vacancy probe microscopy (SNVM) is an important tool for studying nanoscale condensed-matter phenomena. Ga$^+$-ion-focused-ion-beam (FIB) milling has been introduced as a method for fabricating SNVM, while the probe diameter is limited to a few micrometers due to Ga$^+$-induced damage. We report a method for...
This report summarizes changes in federal research funding to the medical physics community between FY24 and FY25. By linking the AAPM membership database with NIH RePORTER records, we quantified the distribution of NIH funding for projects led by AAPM researchers. Although total NIH funding to AAPM members remained relatively stable...
Robots need task planning methods to generate action sequences for complex tasks. Recent work on adversarial attacks has revealed significant vulnerabilities in existing robot task planners, especially those built on foundation models. In this paper, we aim to address these security challenges by introducing PROTEA, an LLM-as-a-Judge...
Large language models (LLMs) are increasingly deployed in security-sensitive applications, where they must follow system- or developer-specified instructions that define the intended task behavior, while completing benign user requests. When adversarial instructions appear in user queries or externally retrieved content, models may...
We study the mass spectra and various mass differences of heavy hadrons containing one or more bottom quarks using MILC's $N_f = 2+1+1$ HISQ gauge ensembles at three lattice spacings. For the valence quarks, we employ a combination of lattice actions: the NRQCD action is used for bottom quarks, the anisotropic Clover action for charm...
Multimodal Large Language Models (MLLMs) have demonstrated strong capabilities across a variety of vision-language tasks. However, their internal reasoning often exhibits a critical inconsistency: although deeper layers may attend to the correct visual regions, final predictions are frequently misled by noisy attention from earlier...
We introduce coefficient systems of pro-étale motives and pro-étale motivic spectra with coefficients in any condensed ring spectrum and show that they afford the six operations. Over locally étale bounded schemes, étale motivic spectra embed into pro-étale motivic spectra. We then use the framework of condensed category theory to...
The compact stellar clusters have emerged as particularly promising candidates for cosmic rays (CR) accelerators. The star clusters, recently observed in gamma-rays, are also known sources of non-thermal X-ray emission, which is due to synchrotron or inverse-Compton scattering of relativistic electrons. Thus, the search for the...
Passive acoustic mapping (PAM) is a key imaging technique for characterizing cavitation activity in therapeutic ultrasound applications. Recent model-based beamforming algorithms offer high reconstruction quality and strong physical interpretability. However, their computational burden and limited temporal resolution restrict their...
We study the robust matrix completion (RMC) problem subject to both sparse outliers and stochastic noise. A non-convex method termed Accelerated Robust Matrix Completion (ARMC) is proposed, which accelerates a prior non-convex approach by incorporating an explicit subspace projection step into the low-rank update, leading to...
We introduce MetaGlyph, a symbolic language for compressing prompts by encoding instructions as mathematical symbols rather than prose. Unlike systems requiring explicit decoding rules, MetaGlyph uses symbols like $\in$ (membership) and $\Rightarrow$ (implication) that models already understand from their training data. We test...