Papers

Large language model (LLM) serving is increasingly constrained by the GPU memory consumed by key-value (KV) caches. Existing compression, eviction, and offloading techniques alleviate this pressure, but serving runtimes typically treat only the configured target KV representation as execution-ready. Under memory pressure, this target-only contract can turn KV shortage into request stalls and preemptions. We present ElasticKV, a mixed-fidelity KV runtime built on the observation that target fidelity need not gate execution. ElasticKV introduces a compact intermediate KV state, making fidelity a runtime-managed execution property. To realize this state in a paged serving runtime, ElasticKV combines (i) a pair-structured layout that turns fidelity reduction into reusable GPU capacity, (ii) a dual-mode attention backend that directly consumes the compact state while preserving the native target-only path, and (iii) pressure-aware fidelity management that adapts KV fidelity to memory pressure. Our extensive evaluation across diverse workloads, model families and scales, and GPU platforms demonstrates the effectiveness and generality of ElasticKV. Under high concurrency, ElasticKV achieves 3.8-4.0$\times$ lower time-to-first-token (TTFT) and 9.1$\times$ lower P90 TTFT than vLLM while preserving generation quality.

We study the problem of computing how close a given election is to being group-separable, measuring proximity by swaps of adjacent candidates in the votes. We also consider several other domains, including caterpillar group-separable, balanced group-separable, single-peaked, and single-crossing ones. Our problem is generally intractable, but we find practical FPT algorithms parameterized by the number of candidates or swaps. For the latter case, our algorithm applies to all domains characterized by finite forbidden subelections, resolving a well-established open problem. We supplement our theoretical findings with experimental analysis.

Large language models (LLMs) inevitably generate some incorrect or outdated content, necessitating efficient and precise mechanisms for continual knowledge updates. However, existing model editing methods struggle to sequentially edit unstructured long-form knowledge, suffering from severe edit forgetting and degradation of general capabilities. To address these challenges, we reframe knowledge editing from a manifold perspective, viewing it as a localized displacement of an edit sub-manifold within the global knowledge manifold. Under this formulation, the problem can be decomposed into two key questions: (i) how to identify representative edit points that effectively anchor the edit sub-manifold, and (ii) how to preserve the remaining manifold structure during the sub-manifold displacement process. Based on this perspective, we propose ManiEdit, a novel manifold-aware autoregressive editing framework consisting of two core components. Pivot Localization addresses the mediocre-point dilemma by identifying high-leverage pivots to anchor the edit sub-manifold. Manifold-Aware Preservation preserves different knowledge types through an energy-weighted penalty combined with recursive null-space alignment. Experiments on two base LLMs and four unstructured editing benchmarks demonstrate that ManiEdit achieves state-of-the-art performance, outperforming the strongest baseline by up to +27.81 BERTScore and +8.50 ROUGE-L, while maintaining near-original general capabilities across six representative downstream tasks. Our code is available at: https://github.com/Areyliu/ManiEdit

Transformers have emerged as powerful architectures for learning solution operators of physical systems. Empirically the prediction error has been observed to decrease when the data size and model size increase, suggesting neural scaling behavior. Yet a theoretical understanding of such scaling laws for transformer-based operator learning remains limited. In this work, we develop a theoretical framework for characterizing the approximation and generalization errors of transformer-based operator learning. Our analysis builds on a local-to-global approximation principle that is naturally aligned with the softmax attention mechanism and yields discretization-invariant output functions. On approximation theory, we derive a universal approximation error of transformer-based operator learning for Hölder-regular operators. On generalization theory, we establish a power scaling law between the prediction error and the training data size. The rate of convergence represented by the scaling exponent explicitly reflects the dimensions of the input and output domains, the regularity of the underlying functions and operators, and crucially, the intrinsic dimension of the input function class. By exploiting this intrinsic low-dimensional structure, our analysis yields a power-law generalization rate for operator learning, in contrast to the logarithmic-type power-law rates appearing in existing analyses of operator learning with feedforward neural networks. Numerical experiments validate the predicted power-law scaling and confirm that the convergence rate varies systematically with the intrinsic dimension of the input function class.

Reproducing a provenance-based intrusion detector's score does not establish what that score says about its emitted alarms or the information its encoder uses. We audit nine released implementations, execute four detectors using their own code, and isolate three measurement effects. First, a fixed-alert comparison separates label choice from neighbourhood credit: ThreaTrace reports precision 0.938 with neighbourhood credit, although only seven of its 994 alarms carry its own attack label. Second, removing test-label checkpoint selection lowers attack detection precision (ADP) by 0.16 to 0.33 across four forty-member word2vec configurations without changing detector order. Third, a buffer-reuse defect gives a linear encoder unintended degree-dependent inputs. Correcting it lowers type-only ADP in every identical-input initialization pair on two hosts, while historical word2vec effects depend on the host. These findings qualify claims from the inspected implementations about alarm precision, performance magnitude and static-attribute sufficiency. STRICT connects them to six checkable reporting requirements. Because the comparisons condition on benchmark targets, they neither validate those labels nor establish a universal detector ranking.

Let $(R,\mathfrak m,k)$ be a Noetherian local ring, let $M$ be a nonzero finitely generated $R$-module, and let $J_1,\ldots,J_s$ be arbitrary proper ideals of positive height on $M$. We introduce the notion of directional mixed multiplicities, obtained from the multigraded quotients $J^{\mathbf u}M/J^{\mathbf u+\mathbf e_j}M$ by retaining one distinguished direction $j$. These invariants refine the mixed multiplicity sequences of arbitrary ideals while remaining compatible with the two classical extremes: for one ideal they recover the positive components of the Achilles–Manaresi multiplicity sequence, and for $\mathfrak m$-primary ideals their top-degree terms recover the usual mixed multiplicities individually. We establish additivity, an associativity formula, superficial reduction, and an exact decomposition of the diagonal coefficients into directional components. Over an infinite residue field, every positive directional coefficient is realized as an eventual length by a sufficiently general directional $(FC)$-system. We then identify an additional synchronization condition under which the directional coefficient becomes the Hilbert–Samuel multiplicity of a parameter ideal, yielding a Rees-type theorem and recovering the classical theorem in the $\mathfrak m$-primary case. In addition, we show a multihomogeneous product formula for $J_1^{n_1}\cdots J_s^{n_s}$, which specializes to the classical Bhattacharya–Rees–Teissier formula.

Natural Language Understanding (NLU) plays a crucial role in various applications, yet its performance suffers from weaknesses in handling the complexities of human languages, ranging from lexical ambiguity to high-level reasoning difficulties. Analyzing errors across diverse linguistic phenomena is crucial for NLU improvement, as it will help humans get insights to comprehensively assess models' limitations and capabilities, so optimizing models' generalization. Notably, several benchmarks contain diagnostics datasets designed for investigation and fine-grained error analysis. When highlighting the gaps in the state-of-the-art, we noted that there is no naming convention for macro and micro categories or even a standard set of linguistic phenomena that should be covered. To overcome this gap, we propose an initial hierarchy for Cross-Lingual NLU error analysis. Moreover, we propose a methodology to create an NLI hierarchical framework and applied a case study on Arabic NLU. Moreover, this paper introduces E-CONAN diagnostics dataset, a freely available dataset manually-annotated with coarse-grained and fine-grained categories based on our proposed Arabic hierarchy. E-CONAN dataset helps NLU designers better understand their models by doing error analysis and in-depth investigation. We used E-CONAN to investigate the performance of 9 pretrained language models and 5 LLMs. Results indicate that LLMs outperform pretrained models in world knowledge and commonsense reasoning macro-category, and underperform pretrained models in syntactic macro-category. Moreover, the hardest phenomena for all models is Reasoning, and the easiest phenomena for all pretrained models is Syntactic, and the easiest for LLMs is Lexico-Syntactic.

We study the elementary continued fractions of Ramanujan. The entries considered are precursors of his hypergeometric and $q$-hypergeometric continued fractions. They appear in the first third of Chapter 12 of Ramanujan's second notebook.

Critical slowing down (CSD), the slowing of recovery from perturbations as a system approaches a critical transition, is widely used for early warning, but whether such signals appear in environmental transitions subject to rapidly changing external forcing has rarely been tested with observations. Here we show, using 2,427 icing-onset events in 1-min observations from 121 South Korean road weather stations, that such signals do not consistently precede road icing. Icing began from different prior surface states, but conditions at onset varied continuously with weak separation among pathways. Cold dose, deposition dose and freeze-thaw accumulation at onset exceeded those at non-icing control times, with cold-dose discrimination highest at decay timescales of 30 min or less. However, the increases in variance and lag-1 autocorrelation expected under CSD did not distinguish icing from control times; the same procedure recovered these increases in synthetic series with a declining recovery rate, and the null result was replicated for 3,091 events at 127 independent stations. Predictive information was nevertheless present: adding recent changes and cumulative antecedent conditions to current conditions improved discrimination, and surrounding meteorological fields increased the area under the ROC curve by 0.030-0.032 for forecast windows of 1 to 12 h and reduced false alarms by 62-67% at 3 and 6 h. These results suggest that road icing is more consistent with recent meteorological forcing moving the road surface into icing-favourable conditions than with a progressive decline in internal stability, and they indicate where early warning should look when a transition is externally forced.

Parameter space symmetries are important for understanding neural networks' loss landscape, training dynamics, and generalization. However, systematically identifying these symmetries remains a challenge. In this paper, we formalize data-dependent parameter symmetries and characterize loss invariance and the group-action axioms through infinitesimal conditions, which provide objectives for jointly learning group generators and nonlinear action maps. Our framework systematically uncovers parameter symmetries, including previously unknown ones. To study larger networks, we establish conditions under which subnetwork symmetries extend to the full model. The same construction gives an explicit family of finite-batch symmetries, providing both analytical examples and a foundation for discovery through small subnetworks. Using the infinitesimal characterization and subnetwork construction, we implement a framework for automated discovery of parameter symmetries, and successfully uncovered symmetries in various architectures, including pretrained transformer models.

The inducibility problem asks for the maximum number of induced copies of a fixed graph among all graphs with a prescribed number of vertices. Inducibility has been an active area of research in extremal combinatorics, but determining all extremal graphs for explicitly defined patterns remains challenging, particularly when the description is required to hold at every host order. A graph $H$ is called a fractalizer if, for every positive integer $n$, every $n$-vertex graph maximizing the number of induced copies of $H$ is a balanced iterated blow-up of $H$, obtained by recursively repeating the same pattern in parts whose sizes differ by at most one. Previous probabilistic results show that large random graphs and random abelian Cayley graphs are fractalizers with probability tending to one, establishing their abundance without directly providing explicit families. We prove that every sufficiently large prime-order Paley graph is a fractalizer. Thus these classical arithmetic patterns determine the exact recursive structure of every extremal host, at every host order and without any algebraic assumptions on the host. We also construct a second explicit infinite family of nontrivial Cayley fractalizers, for which the fractalizer property admits a simpler proof than in the Paley case. Together, these results resolve the explicit-construction question discussed at the 2025 American Institute of Mathematics workshop "Flag Algebras and Extremal Combinatorics."

AI-generated synthetic personas are increasingly used for content planning and virtual-user simulation, yet the digital content use contexts embedded in their narratives remain underexamined. Using all 1,000,000 records of NVIDIA's Nemotron-Personas-Korea, this study coded mentions of five engagement modes-viewing, discovering and sharing, interacting, reading, and listening-with an openly released Korean coding dictionary and applied latent class analysis with a three-step approach accounting for classification error. Four classes emerged: viewing-centered (20.8%), reading-centered (3.8%), listening-centered (2.1%), and low-mention (73.3%). Age showed the strongest association: the odds of the viewing-centered versus low-mention class were multiplied by 0.15 per 10-year increase, while mentions of any non-viewing mode declined from 49.4% among those under 30 to 2.3% among those aged 70 and over. Sex differences in predicted class probabilities were below 1 percentage point, and provincial differences reached at most 5.0 percentage points. These classes describe narrative configurations, not consumer segments, and provide a basis for assessing diversity in synthetic persona use contexts.

Self-evolving agents aim to turn research feedback into reusable skills, tools, and research rules. Whether these accumulated capabilities continue to improve later research requires controlled evaluation. Long-horizon alpha discovery provides a state-dependent setting: once a new factor enters the portfolio, the predictive information already covered changes, so the value of the same candidate or experience may change over time. We introduce EverMine, an empirical framework for studying self-evolving research capabilities in long-horizon alpha discovery. EverMine decomposes the research state into history (Hist), the current factor portfolio (Frontier), and reusable capabilities (Cap). Under matched resource limits, we compare complete runs with fixed or evolving Cap, and replace Cap while holding Hist and Frontier fixed to estimate the conditional value of accumulated capabilities. We also combine full trajectories with historical-state replay to examine how experience-based decisions affect candidate selection and portfolio outcomes. Across 18 long-horizon trajectories, end-to-end comparisons show no consistent gain from Cap evolution. Across 48 continuation branches from shared Hist and Frontier states, accumulated Cap also does not consistently outperform the initial Cap. Parameter tuning of existing factor structures can still improve the portfolio. In an exploratory replay of two screening batches from one Evolving trajectory, some screened-out candidates have positive marginal value at the original state, yet submitting all screened-out candidates sequentially slightly lowers final portfolio IC in both batches. These results show that candidate value depends on the evolving portfolio and submission order, and motivate evaluating self-evolving research capabilities through end-to-end outcomes, conditional capability value, and the consequences of experience-based decisions.

We present In-Token Learning, an image restoration framework that adapts a pretrained diffusion transformer using conditional rectified flow matching. Clean targets paired with degraded inputs supervise transport from Gaussian noise to restored images. Spatially aligned degraded-image tokens are fused with evolving latent tokens along the channel dimension, preserving the image-token count at a given resolution. Direct Low-Quality Guidance (DLG) combines frozen degraded-image embeddings with a fixed task prompt through the native conditioning pathway, without a trainable ControlNet-style branch or image captioning. We evaluate super-resolution and denoising on DIV2K, LSDIR, FFHQ, RealLQ250, and RealPhoto60, and automatic colorization on DIV2K and LSDIR. The tasks use separately trained checkpoints under the same framework. Results show competitive fidelity and perceptual quality under the evaluated protocols, with weaker generalization on RealLQ250. We report full-image QHD ($2560{\times}1440$) inference and a tiled $12$K restoration demonstration of Along the River During the Qingming Festival. Attention cost still increases with resolution. This technical report preserves the early broader study underlying Fill2SR, which subsequently developed the real-world super-resolution direction.

Manual visual inspection on assembly lines is a persistent manufacturing bottleneck: operator fatigue over extended shifts lowers defect-detection rates. This paper presents the design, integration, and field deployment of an AI-assisted collaborative inspection cell at the Silverline kitchen-appliance factory, developed within the AI-PRISM project. The cell couples a Universal Robots UR 10e cobot carrying a machine-vision defect-detection pipeline with a Comau Racer-5 cobot for functional tests, coordinated through ROS 2 Humble on an Ubuntu 22.04 LTS server. Multi-modal data (Basler camera imagery, TIA microphone acoustics, and SPS electrical-safety measurements) are logged locally and visualised in real time with Grafana. We report the practical deployment challenges (close-proximity safety, AI robustness under glare and reflections, ROS 2 namespace collisions across two cobots, and operating-system and dependency issues) together with the engineering solutions adopted, and structure the integration through a four-level Human-Robot Interaction analysis. The deployed cell cuts per-unit quality-check time from 82 s to 61 s (about 25%), raises final-control resource efficiency from 0.75 to 0.88, reduces operator visual-inspection viewing time by 82%, and significantly lowers operator mental demand (p = 0.005, NASA-TLX).

Trust evaluation plays a critical role in cybersecurity by supporting risk mitigation and decision-making. A variety of trust evaluation methods have been proposed, with learning-based approaches offering high accuracy and automation. However, they often require substantial ground truth, suffer from low training efficiency, lack support for basic trust properties, and provide limited explainability. Large Language Models (LLMs) offer a compelling alternative due to their strong zero-/few-shot reasoning abilities and broad knowledge. To this end, we propose LLM4Trust, the first benchmark framework that systematically explores the capabilities of LLMs for trust evaluation. We first construct diverse trust graphs to model five basic trust properties and design corresponding property understanding tasks. We then assess the ability of eight representative LLMs to understand these properties under nine prompt methods. Based on this exploration, we identify the most effective LLM-prompt combinations and apply them to five real-world datasets for validating LLMs' trust evaluation capability. During this process, we propose two strategies to extract key information from large-scale trust graphs, addressing the context window limitations of LLMs. Extensive experiments show that LLMs can effectively understand basic trust properties and have great potential for real-world trust evaluation, particularly under limited supervision. However, they remain vulnerable to attacks targeting trust graphs and demonstration examples used in few-shot prompting, and incur high inference costs. Accordingly, we propose a defense mechanism and batch inference to improve the robustness and efficiency of LLM-based trust evaluation. The source code of LLM4Trust is available at https://github.com/Jieerbobo/LLM4Trust

Weakly-supervised thoracic disease detection in chest X-rays (CXR) is challenging due to subtle appearances and complex anatomical overlap, motivating anatomy-aware modeling for improved localization. However, prior anatomy-aware methods typically rely on coarse region proxies or static spatial priors, which may restrict dynamic instance discovery and limit precise localization of small abnormalities. We propose Anatomy-Structured Hierarchical Multiple Instance Learning (ASH-MIL), a framework that introduces parallel anatomy-structured observation branches (cardiac, pulmonary, and agnostic) combined with hierarchical MIL aggregation. Anatomical priors are injected as soft spatial biases into decoder cross-attention, enabling anatomically grounded evidence maps without disease bounding-box supervision. Instance localization is derived directly from MIL-weighted cross-attention maps without bounding box supervision. Experiments on CXR8 and cross-domain MIMIC-CXR held-out sets demonstrate consistent improvements over prior weakly-supervised and anatomy-aware approaches, particularly under stricter localization criteria. Our code is available at https://github.com/jn-kim/ash-mil.

Metric graphs are generalisations of linear networks and provide a natural framework for the definition of continuously-indexed Gaussian processes. We define a new class of distances on these topologies, termed polyharmonic distances, which unify and extend the spectral construction underlying the effective resistance distance and the biharmonic one. We give both a spectral and a variational characterisation. Furthermore, we show an explicit class of stochastic processes whose variograms coincide with the squared polyharmonic distances. Finally, we show how these metrics can be composed with suitable completely monotonic functions to define isotropic processes having any prescribed finite-order mean-square differentiability along the edges and satisfying the Kirchhoff conditions up to order one at the vertices.

Recent 3D large multimodal models (3D-LMMs) rely on a visual bottleneck to compress complex 3D scene evidence into a limited number of visual tokens compatible with large language models (LLMs). Current visual bottlenecks, however, often passively compress heterogeneous 3D evidence into a homogeneous object-centric token sequence, leaving the spatial organization of the scene under-represented. This under-representation forces the LLM to recover spatial relations from a flattened token sequence, leading to unstable reasoning in relation-intensive and spatially ambiguous scenes. To address this issue, we propose SceneScaffold, an active scene-state construction framework for unified 3D scene understanding. SceneScaffold reformulates the visual bottleneck from a passive feature compressor into an active scene organizer, constructing a role-aware spatial scaffold before language reasoning. Specifically, SceneScaffold organizes superpoint-level visual evidence into scene-state components with distinct structural roles: entity states preserve core object semantics, scene-frame states maintain spatial references via boundary and region anchors, relation states encode object-environment interaction cues, and a global summary provides compact context. Through this role-aware construction, SceneScaffold provides the LLM with a spatially organized scene representation before language reasoning. Experiments on unified 3D scene understanding tasks, including 3D visual grounding, question answering, and dense captioning, demonstrate the effectiveness of SceneScaffold, while diagnostic results further show its applicability to relation-intensive and spatially ambiguous cases. Code is available at https://github.com/lixiangqi707/SceneScaffold.

Persistent memory lets language-model agents carry information across long-running collaborations, but leaves a lifecycle question open: what may a returning agent still act on once the shared state has changed? A memory can be correctly retrieved, relevant to the current task, and faithful to its source, and nonetheless be inadmissible for action: an itinerary saved before a pause still names the hotel the team has since replaced. We formalize this as temporal memory admission and present TRACE, a training-free layer that treats re-entry as an eligibility decision rather than a storage or retrieval operation, reconciling a departure checkpoint against absence-period updates, resolving explicit and implicit invalidation, and releasing a bounded Return View only when it covers the returning role's open obligations. We evaluate TRACE under three actor models on Memora, STALE Type II, and a derived ManBench-Return setting, each recast as return episodes: one agent departs, four teammates change the shared state, and the agent rejoins. What separates methods is not overall accuracy but whether one can retain valid memory and reject stale memory at once, and no single-policy baseline can: Restore (reinstate the departure checkpoint in full) admits stale state, Reset (start the return from an empty memory) discards valid state, each bottoming out at 0% on one of the two. TRACE is the only method high on both, reaching 92.6-98.3% valid-information availability with 98.4-99.5% invalid-information rejection on ManBench-Return, within 3.8 points of the best baseline's overall accuracy. On STALE Type II it improves Overall over the strongest comparison policy by 22.3 (Qwen), 18.5 (Gemini), and 27.5 (DeepSeek) points at roughly 2.3 times their tokens, while a write-time consolidation pipeline is more accurate still at 3.99 times TRACE's.

Photoplethysmography (PPG) is widely recorded by clinical monitors and consumer wearables, providing a scalable source of continuous physiological information. These recordings offer an opportunity for physiological assessment at scale, but realizing this potential requires models to learn from both signal-derived physiological supervision and broader clinical context captured in electronic health records (EHRs). This involves aligning information spanning local observations, care events, and entire visits with PPG representations at corresponding temporal scales. However, existing PPG foundation models primarily rely on task-specific prediction heads, while the medical knowledge of large language models does not necessarily translate into waveform understanding. To bridge this gap, we introduce PPG-LM, the first PPG-language model family to learn physiological representations from both signal-derived supervision and broader clinical context captured in EHRs. To construct clinically grounded captions, we develop an automatic captioning pipeline that generates segment-, event-, and visit-level descriptions from signal measurements and structured EHR records. We then learn from these pairs through a two-stage framework that first establishes segment-language correspondence through contrastive learning and waveform-conditioned captioning, then extends alignment to events and visits through time-aware aggregation and temporal statement matching. Pretrained on approximately 73k hours of PPG, PPG-LM supports language-based recognition, cross-modal retrieval, and segment captioning. Experiments on MC-MED, MIMIC-III, and VitalDB show improved retrieval and caption factuality over language-model baselines and gains over PPG and time-series foundation models on multiple clinical prediction tasks.

Many probabilistic predictors must satisfy exact structure in every stochastic realization, yet common hard-constraint approaches form predictions in ambient coordinates and then correct or project them. We introduce SLP-ProbHard, a cross-family, representation-centered framework for probabilistic hard-constrained learning when explicit structural parameterizations are available. Its core object, a Structural Feasible Latent Parameterization (SFLP), combines a structural latent law $Z \sim P^Z_\theta(\cdot\mid x)$ with a feasible map $Y=h_\phi(x,Z)$ that satisfies the constraint for every latent realization. Together these components define the predictive law itself, including its support and boundary probabilities, rather than serving as a final feasibility wrapper. We study how feasible coordinates and maps affect stochastic dimension, dependence, calibration, expressiveness, and computation. Experiments use Gaussian latent laws and fixed geometry-derived maps across affine equalities, ordering and simplex constraints, nonlinear manifolds, and three structural representations of seven-basin hydrological flow-duration-curve (FDC) data. In an official-source affine comparison with ProbHardE2E/DPPL, both methods achieve zero practical constraint violations. SLP-ProbHard uses 8 instead of 11 stochastic coordinates and improves MSE/MAE, while DPPL yields better marginal CRPS and closer-to-nominal coverage; a paired test detects no Energy Score difference across ten seeds. Real-world affine and nonlinear FDC representations reduce 13 to 7 and 14 to 8 ambient versus computational coordinates, respectively. Exact feasibility alone thus does not determine a predictive law, motivating direct structural generation when meaningful feasible coordinates are available.

We investigate mono-$Z$ signature at the future electron-positron collider within the dark matter effective field theory (DMEFT) framework, considering operators up to dimension six for scalar, fermionic, and vector dark matter (DM) candidates. The operators are classified according to their underlying production mechanisms. We identify the phenomenologically viable parameter space by imposing the observed DM relic abundance together with direct and indirect detection constraints, Higgs invisible decay limits, and LHC mono-$Z$ bounds obtained through a recasting analysis. We study the discovery prospects of the surviving DM operators at the future $e^-e^+$ collider including initial and final state radiation, detector simulation, and the reconstruction of both leptonic and hadronic $Z$-boson decay modes. We exploit missing energy, missing transverse momentum, recoil mass, $Z$-boson pseudorapidity, together with beam polarization, which play a significant role to distinguish the signal from the Standard Model background. We find that fermionic dipole, scalar leptophilic, vector Higgs-portal, and gauge-portal DM operators offer robust mono-$Z$ discovery potential, probing multi-TeV effective scales.

Adversarial patches for aerial perception are typically evaluated as digital composites, with printing left as an implementation detail. This study imposes three physical constraints during optimization rather than after it: the color range a particular printer can reproduce, the size of the flat panel a vehicle offers, and the loss of fine detail incurred when the patch is imaged from altitude. The principal comparison isolates the ink set. Two patches share all seventeen recorded optimization settings and differ only in the colors available to them. One is constrained to a uniform color cube; the other to a gamut measured by printing and scanning a 216-patch chart. Each was optimized at three seeds and evaluated against thirteen victim conditions, with every rate reported against a size-matched optimized control. The effect of the measured gamut is victim-dependent rather than uniform. Net attack success rises on three of six closed-set segmentation victims, and for these the seed ranges of the two ink sets are disjoint: $+0.120$ on DeepLabv3-R101 and $+0.041$ on SegFormer-B0. The color-cube patch is consistently stronger on the open-vocabulary segmenter and on two of four detectors, though no detector exceeds a net of $+0.026$ under either ink set. The natural explanation is that a printable palette is simply less chromatic and lower in frequency than a digital one. Eleven further patches test this account and it does not hold. Once cardinality is matched, a palette as chromatic as the cube attacks equally well. Cardinality itself shows no trend from three inks to thirty-two. Palettes matched on cardinality, lightness and chroma, and differing only in hue placement, span $0.035$ to $0.136$. A physical evaluation with printed decals did not detect transfer; it bounds the transferred rate at $0.133$, which does not exclude the simulated value of $0.121$.

We introduce and study Reed–Solomon sunflowers, namely families of Reed–Solomon codes whose pairwise intersections are all equal to the same fixed subspace. This notion lies at the intersection of extremal subspace combinatorics and coding theory: it can be viewed as a structured version of the sunflower problem in the Grassmannian, and it naturally produces constant-dimension subspace codes with prescribed minimum distance. We focus mainly on the case in which the center is the one-dimensional space generated by the all-one vector. We give an algebraic criterion, expressed in terms of generalized $V$-matrices, ensuring that a family of Reed–Solomon codes forms such a sunflower. We then study the size of these families through counting and constructions. In dimension two, we show that all distinct Reed–Solomon codes form a sunflower and determine its size by counting Reed–Solomon codes up to affine equivalence of their evaluation vectors. For fixed dimension $k\geq3$ and length $\ell\geq2k-1$, we give an explicit recursive construction with $\Omega_{k,\ell}(q^{\lfloor\ell/(2k-1)\rfloor})$ petals and a greedy existence argument with $\Omega_{k,\ell}(q^{\ell-2k+2})$ petals as $q\to\infty$. We also apply the greedy argument to obtain families of $[\ell,k]_q$ MDS codes of size $\Omega_{k,\ell}(q^{2(\ell-2k+2)})$, whose pairwise intersections have dimension at most one but need not be equal.

We study linear perturbations of the Maldacena-Milekhin-Popov traversable wormhole in the sector where polar metric perturbations couple to axial gauge perturbations, and reduce it to two coupled master equations of Zerilli-Moncrief type. In the throat the Casimir energy of the lowest Landau level of charged fermions holds the wormhole open, and the linearised Bianchi identities hold only if the fermions respond. Once their state is fixed, their two-dimensional response follows from the conformal anomaly. We lift it to four dimensions on the field lines, assuming that the metric of each line scales with the local field strength. Through the trace anomaly this scaling couples the field strength to the worldsheet curvature, reminiscent of a Wen-Zee term. The response comes with a Hall current that sources the axial gauge mode. At zeroth order in the fermion backreaction $\alpha$ the throat is AdS$_2\times S^2$ and the modes are two Poschl-Teller problems with squared masses $l(l-1)$ and $(l+1)(l+2)$. We give the $O(\alpha)$ corrections in closed form. In the mouths the fermions act through the anomaly at the same order, and mouths and throat match. The higher Landau levels act through the one-loop Euler-Heisenberg action. We keep it to first order in $\eta_{\rm H}=g_c^2/12\pi^2$, which is not suppressed by $1/q$. It enlarges the AdS$_2$ radius and splits the levels shared by the two modes. On the closed-throat levels examined their $O(\alpha)$ mixing is of second order. The $O(\alpha)$ frequency shifts there, measured by the static observer at the throat centre, are real and depend on the assumed scaling, compared here with a lift on the coordinate planes. Region by region, at first order in $\alpha$ and $\eta_{\rm H}$ and on the levels examined, we find no growing mode at $l=2,3,4$, although the $O(\alpha)$ mouth rests on an estimate. Stability of the wormhole as a whole is not shown.

Deep generative models are used to solve partial differential equations and model distributions of physical system states, but ensuring that the generated samples satisfy the governing laws remains challenging. Projection-based flow-matching methods enforce physics by correcting the flow from an unconstrained noise distribution. These corrections shift the generated samples away from the distribution of target solutions, especially in high noise regions. To address this limitation, we propose Source Anchoring for Physical Consistency (SAPC), a Functional Flow Matching method that encodes the physical constraints into the source noise before generation begins. We evaluate SAPC on five systems governed by partial differential equations, covering six tasks with linear and non-linear dynamics, and compare results against five baselines and the unconstrained backbone. Anchoring the source reduces the need for large corrections that drive samples onto admissible but off-distribution states, and SAPC reproduces the target distributions most accurately on every evaluated task, while matching the constraint precision of the best projection-based baselines. Ablation experiments show that this gain arises from pairing source projection with a matched training objective that regresses toward the projected source. These results identify the source distribution as a key design choice for physically consistent generative modelling.

In autonomous deep research, a user provides a task and relevant background, then leaves the agent to conduct an extended investigation without further human intervention. We study how this initial user information is reflected in intermediate actions and how these actions relate to final recommendations. We introduce DRaligned, a counterfactual behavioral evaluation framework built on PDR-Bench. By varying one task-relevant user factor while keeping the remaining context fixed, we compare acquisition requests, working drafts, and final reports. Source-grounded extraction, blinded local judgments, and deterministic aggregation yield coarse directional measurements while leaving ambiguous cases unresolved. Our experiments show that strong user-specific delivery can emerge from a largely shared research process: agents investigate similar broad questions but allocate requests differently, and final recommendations distinguish user conditions more clearly than explicit requests do. Reports can also integrate user factors that were not jointly visible during acquisition. In readable draft-to-report comparisons, recommendations often retain their coarse user-specific direction despite substantial rewriting. Final directional differences recur across tested agent models, execution harnesses, and evaluator models, even as execution paths vary. These findings describe how initial user information shapes autonomous research and clarify the relationship between the process an agent follows and the recommendations it delivers.

In this paper I revisit my 2006 conjecture on correlated errors in entangled physical qubits, originally proposed as a potential obstruction to quantum fault tolerance. The conjecture asserts that, in any physical implementation of a quantum computer, the effective noise channel acting on entangled physical qubits contains a joint fully depolarizing component, with a rate comparable to that of two-qubit gate errors. This hypothesized structural constraint goes beyond standard noise models and, if valid, would pose a significant challenge to scalable quantum fault tolerance. The conjecture remains open, but recent advances in experimental quantum computing bring it within reach of empirical testing on current devices. I also discuss two related directions in my critical study of quantum computation: the role of noise sensitivity and computational complexity in noisy intermediate-scale quantum systems, and the statistical analysis of experimental claims of quantum advantage. Finally, since this paper is written for a volume honoring Yuri Gurevich, I include some reflections on the ways in which my scientific and personal trajectory became intertwined with Yuri's.

We reduce the Alperin-McKay-Navarro conjecture to an inductive condition on the universal covering groups of non-abelian finite simple groups. The reduction yields height-zero character bijections compatible with Brauer correspondence and block relations between H-triples for arbitrary finite ambient groups. We record the resulting arithmetic and structural consequences, including preservation of $p$-rationality levels and character-theoretic criteria for Sylow subgroups. The proof combines semilinear centralization and transfer through quasisimple components with induction on the central index. It uses the Clifford and gluing theorems for block relations and a central-defect Dade-Glauberman-Nagao correspondence established in two related papers. We also verify the inductive condition for several sporadic groups at primes for which the Sylow subgroups have prime order, and in defining characteristic for the Suzuki groups ${}^{2}B_{2}(2^{2m+1})$ with $m \geq 2$ and the small Ree groups ${}^{2}G_{2}(3^{2m+1})$ with $m \geq 1$.