Papers

Holographic multiple-input multiple-output arrays provide high spatial resolution and flexible wavefront control, but curved or conformal deployments jointly introduce curvature, mutual coupling (MC), and direction-dependent visibility. This paper studies a two-dimensional (2D) holographic curvaturereconfigurable aperture (HoloCuRA). A geometry-based model characterizes element positions, steering, and spatial correlation, while an impedance-based coupling matrix and a visibility vector capture electromagnetic interactions and shadowing. Relative directivity and spectral efficiency are evaluated versus curvature. Results show that MC mainly reshapes local directivity fluctuations, whereas visibility dominates the global pattern and spectral-efficiency variation under both isotropic scattering and CDL-A channels.

We prove that every finite simple triangle-free graph $G$ on $n\geq1$ vertices satisfies $q_{\min}(G)\leq7n/50$, where $q_{\min}(G)$ is the least eigenvalue of its signless Laplacian matrix $Q_G=D_G+A_G$, with $D_G$ the diagonal matrix of vertex degrees and $A_G$ the adjacency matrix. The constant is sharp, since the Higman–Sims graph and its balanced blow-ups attain equality, as shown by de Lima, Nikiforov and Oliveira (2016). This proves the bound conjectured by Balogh, Clemen, Lidický, Norin and Volec (2023). The proof uses five test column vectors associated with an ordered edge to obtain a matrix Rayleigh inequality. Our use of local Rayleigh inequalities together with flag-algebra semidefinite programming is inspired by Zhou (2026). We use semidefinite programming and rational approximation to find the required positive semidefinite matrices, and complete the proof by exact correction and verification.

We define a $d$-cube in a graph $G$ to be monophobic in $G$ if, whenever it supports a $d$-face of a discrete singular $d+1$-cube, it must also support another face. In many cases, the presence of such cubes implies the nontriviality of discrete cubical homology and homotopy groups associated with $G$. We describe several large classes of examples and characterize the property completely when $d=2$.

We propose an encrypted collision avoidance maneuver (CAM) design protocol using homomorphic encryption (HE) that enables operators to outsource CAM design to an untrusted server without revealing their private covariance. To this end, we reformulate the constrained CAM optimization as a sampling based estimation problem via a variational approach, so that the private covariance dependent terms can be efficiently evaluated under HE. The protocol requires only a single round of communication, and experiments on six real-world conjunction data messages demonstrate improved feasibility and reduced fuel consumption compared with CAMs designed using public covariance.

Many widely used activation functions, including ReLU, GELU, Swish, and Sigmoid, produce predominantly positive outputs, resulting in a nonzero post-activation mean that we term activation bias. Despite its ubiquity, the effect of activation bias on learning dynamics remains poorly understood. Using controlled models, we show that activation bias accelerates initial loss reduction but is accompanied by overshooting, non-smooth trajectories, and a steeper loss landscape. This early advantage also coincides with the model's use of task-relevant information encoded in the activation bias component. In contrast, reducing activation bias slows initial learning but ultimately yields a lower final loss. Experiments on larger vision and language models exhibit a similar pattern: smaller activation bias is associated with better late-stage performance. Thus, this seemingly minor property can substantially redirect learning and warrants explicit consideration in future activation function design.

3D human pose estimation requires accurate modeling of motion-dependent skeletal structures and inter-joint dependencies. Existing methods typically rely on fixed skeletal topologies in Graph Convolutional Networks (GCNs) or implicitly capture joint relations through attention mechanisms, limiting their adaptability to diverse motion patterns. To address this issue,we proposeDSR-Net, a dynamic structural representation learning framework for 3D human pose estimation. Specifically, a Dynamic Skeletal Relation (DSR) module is introduced to learn adaptive inter-joint dependencies beyond predefined skeletal connections. Based on the learned relations, a hierarchical relation-aware fusion strategy is developed to facilitate effective interaction between structural and temporal representations. In addition, a lightweight Residual Skeletal Recalibration Module (RSRM) is designed to enhance the stability of structure-aware features with minimal computational overhead. Extensive experiments on Human3.6M and MPI-INF-3DHP demonstrate that DSR-Net achieves MPJPEs of 37.9 mm and 14.9 mm, respectively, consistently improving over strong baselines. The results validate the effectiveness of dynamic skeletal relation modeling for robust 3D human pose estimation.

The Copula Active Subspaces (CAS) method of Part I estimates a non-Gaussian observation-noise density $\pi_{\boldsymbol{n}}$ from residual samples. Its KL divergence $\mathcal{D}_{\mathrm{KL}}(\pi_{\boldsymbol{n}}\,\|\,\widehat\pi_{\boldsymbol{n}})$ admits an exact decomposition into marginal, projection and reduced-density terms and bounds the posterior KL divergence averaged over the true predictive distribution. We derive a priori bounds for these terms, together with estimates or bounds from residual samples. Weighted nearest-neighbour entropy estimates are used for the marginal and reduced-density terms, with asymptotic standard errors in exact rank coordinates. For the projection error, the score is fitted on one subsample, and its mean squared magnitude outside the retained subspace is estimated on the other. Corrections for score and rank-transform errors then give a theoretical upper bound with probability at least $1-\delta$. Tests on three known noise laws in dimension $d=20$ use the estimated rank transform. In $11$ of $12$ combinations of example and sample size, the sum of the two estimated terms and the trace bound exceeds the reference divergence after averaging over the estimated models. In these tests, score-estimation error is the largest contribution to looseness. The overestimate from the Gaussian log-Sobolev inequality alone is at most a factor of $1.98$.

EEG foundation models (EFMs) have improved EEG representation learning by transferring knowledge from large-scale multi-subject pretraining to downstream tasks. However, their pretrained embeddings often retain substantial subject-related variation, placing an additional burden on the downstream classifier, especially in cross-subject settings. To address this issue, we propose Disentangled EFM Fine-Tuning (DEFT), a plug-and-play adapter for EFM fine-tuning. DEFT encourages branch disentanglement of pretrained tokens into task-oriented content and subject-consistent style, then performs downstream prediction from the content branch. It instantiates this factorize-before-classify design with recursive Gaussian refinement, temporal-variance regularization, subject-aware contrastive learning, and reconstruction. Experiments across multiple EEG datasets and four EFM backbones show that DEFT improves matched fine-tuning baselines. Further ablations and analyses support the intended content/style branch organization.

For a subset $A$ of a finite abelian group $G$, let $\Sigma_k(A)$ denote the set of sums of $k$ distinct elements of $A$, and let $\Sigma(A)=\bigcup_{k\ge1}\Sigma_k(A)$. The problem of determining when $\Sigma(A)=G$ goes back to Erdős and Heilbronn. Nguyen and Vu later formulated the corresponding fixed-length question: for a prescribed $k$, when does $\Sigma_k(A)=G$? For groups of odd order, we prove the sharp large-order threshold $|A|>|G|/3+1$, which forces $\Sigma_k(A)=G$ throughout the entire middle range $4\le k\le |A|-4$. This determines the limiting thresholds posed by Girard, Griffiths and Hamidoune. If $p$ is the least prime divisor of $|G|$, then, for fixed odd $p$ and sufficiently large $|G|$, the exact threshold $|A|\ge |G|/p+p-1$ forces covering throughout the middle range beginning at $2p-2$. Thus, throughout this range, prescribing the number of summands does not change the sharp covering threshold. For $3\le k\le2p-3$, the restricted and ordinary $k$-fold covering problems have the same asymptotic threshold. As an application to MDS elliptic codes on an elliptic curve $E/\F_q$, we sharpen the half-scale conjecture of Li, Wan and Zhang, obtaining the sharp $1/3$-scale bound when $|E(\F_q)|$ is odd and, more generally, the exact bound determined by the least prime divisor of $|E(\F_q)|$.

Explainer videos – a worked geometry problem, why the sky is blue, how a dynasty fell – are usually one more target for pixel video generation. We argue they are a different kind of artifact: a program (script, storyboard and per-frame animation code) that a deterministic renderer turns into frames. A program can be read, diffed, tested and audited; pixels can only be watched. Reels-RSI is an open-source reel director for any agent harness; its self-improvement happens at the program level rather than in weights or prompts. Every revision pass is logged with a code snapshot; a retrospective pairs each check finding with the diff that fixed it and proposes lessons; a lesson that a static check can catch is compiled into a rule with a must-fail and a must-pass example, so the memory is unit-tested; and a rule learned today runs over everything made before it. A bounded, benchmark-gated step lets an agent rewrite the director's playbook, kept only when blind pairwise judgments on held-out topics prefer it; nothing enters memory without a human decision. Across 111 reels – a field study, two playbook evolutions, a 2x2 memory ablation, a weaker director, a 48-reel learning curve and a judge-reliability study – we find that the rule learned in the field caught four latent defects in frames authored before it existed and none after; that rules are an audit for a strong director (eight unguarded reels, no violation) and a guard for a weaker one (thirty violations unguarded versus five); that lessons shorten revision loops but do not compound or move the judge; and that the first rewrite's 11–1 verdict was a position bias in our vote design (judges pick the second-shown reel 62–81% of the time), masking scripts that traded thirteen facts for craft. We document four failure modes of the preference signal, which check caught each, and the fixes. Code, rules, lessons and data are public.

GS 1843+00 (Ginga 1843+009) is a poorly studied Be/X-ray binary system that underwent a major X-ray outburst in 2026 February-May after nearly sixteen years of quiescence. We present an X-ray, optical, and radio study of GS 1843+00 during this outburst, focusing on the evolution of the neutron star emission geometry, Be-disk connection, and possible outflow activity. We analyse three NuSTAR observations obtained at different stages of the outburst, complemented by archival RXTE and Chandra data, optical spectroscopy and imaging with the Nordic Optical Telescope, and a MeerKAT radio observation. NuSTAR observations reveal coherent X-ray pulsations around 29.7 s with a spin-up rate of $(-5.1\pm0.1)\times10^{-8}$ s s$^{-1}$ during the outburst. The pulse profiles show strong evolution with energy and outburst phase, with distinct morphologies even at comparable luminosities. The pulsed fraction varies between 1 and 16% and exhibits a complex reversal in energy dependence above 7 keV, increasing with energy during the first two observations and decreasing during the third. The 3-79 keV broadband spectra are described by thermal Comptonization with a blackbody and an Fe K$\alpha$ emission line, with no evidence for a cyclotron absorption feature. Phase-resolved spectroscopy reveals variations in the Comptonization and blackbody parameters, indicating changes in the accretion-column structure. Optical spectroscopy confirms the presence of a Be-circumstellar disk through H$\alpha$ and Paschen-series emission lines. No radio counterpart is detected, with a $3\sigma$ upper limit of 36 $\mu$Jy beam$^{-1}$ at 1.28 GHz. Weaker H$\alpha$ emission compared with previous epochs indicates that the Be-circumstellar disk was in a different state during the current episode, while the radio non-detection suggests that any associated outflow activity was faint or below the detection limit.

Recent advances in machine learning have improved rainfall-runoff prediction, yet most data-driven models remain difficult to interpret and do not explicitly enforce physical conservation laws. We present the Gated Catchment Hydrology Model (GCHM), a mass-conserving architecture based on the Mass-Conserving Perceptron (MCP). GCHM uses five storage states representing canopy interception, snow, soil moisture, groundwater, and streamflow generation. Constrained gates govern precipitation partitioning, evapotranspiration, infiltration, seepage, runoff, and baseflow while learning context-dependent process relationships. We evaluate GCHM across 50 CAMELS catchments spanning snowmelt-dominated, recent-rainfall, historical-rainfall, and arid regimes, using NOAA AORC meteorological forcing and remotely sensed vegetation and land-surface data. Performance is assessed using the Jawad-Kling-Gupta Efficiency (JKGE) metric, which accounts for stationary and non-stationary hydrologic behavior. We compare the full architecture with ablated variants and recurrent neural network benchmarks. GCHM reproduces streamflow dynamics across diverse hydroclimatic regimes, including arid catchments, while maintaining interpretable states and fluxes. Ablation experiments show that simplified architectures may perform adequately under wet conditions but struggle to represent flow persistence and recession during dry periods, when groundwater storage sustains low flows. Learned gates recover hydrologically plausible relationships governing snowmelt, runoff, seepage, and evapotranspiration. GCHM combines competitive predictive performance with mass conservation and interpretability, providing a foundation for hydrologic modeling and counterfactual analysis.

Microsurgical anastomosis is one of the most challenging and sophisticated procedures for surgical robotics, requiring accurate manipulation of surgical instruments within a limited space to achieve precise operations. Accurate surgical trajectory anticipation is essential for surgical scene understanding, intraoperative planning, robotic manipulation, and reliable behaviour analysis. However, existing approaches mainly focus on features extracted from surgical videos and historical trajectories, while overlooking rich semantic and motion information that can provide more discriminative multi-modal representations for complex surgical instrument-tissue interactions. In this work, we propose a cross-modal feature fusion framework for long-term 2D and 3D microsurgical anastomosis trajectory anticipation. The proposed framework consists of a multi-modal representation extractor and a cross-modal feature fusion network. Semantic recognition of surgical objects and consecutive motion estimation are incorporated as multi-modal visual cues, providing complementary spatial and temporal guidance for instrument-tissue interaction characterization. A coarse-to-fine progressive fusion strategy with a long-short-term memory mechanism is introduced to capture temporal dependencies step by step. The training and evaluation are conducted on our established robotic platform and a comprehensive dataset containing both phantom and ex-vivo surgical scenarios is constructed. Extensive quantitative and qualitative experiments demonstrate that the proposed framework consistently outperforms existing methods and exhibits improved robustness across different surgical scenes, highlighting its potential for both intelligent robot-assisted and autonomous surgical systems.

Let $X$ be a smooth projective complex surface. We obtain explicit formulas for the Chern character operators on the Hilbert schemes of points $X^{[n]}$ in terms of Nakajima operators, determining in particular the coefficients of the terms involving the canonical class $K_X$ and its square. Consequently, the cohomology ring structure of the Hilbert scheme is determined. As an application, we give an alternative proof of the action of the affine Yangian $Y_{t_1,t_2}(\widehat{\mathfrak{gl}}_1)$ on the cohomological Fock space $\mathbb{H}_X = \bigoplus_{n\geq 0} H^*(X^{[n]},\mathbb{Q})$.

We demonstrate a compact, fully stabilized, all-polarization-maintaining (all-PM) Er:fiber optical frequency comb (OFC) implemented on a silica optical-cube platform. The optical comb module, including the oscillator, amplifier, spectral-broadening stage, 780-nm frequency-doubling branch, and f-to-2f interferometer, occupies a volume of 0.88 L. Without active temperature control, the free-running repetition frequency exhibits a peak-to-peak excursion of 75 Hz over 10 h, corresponding to 0.375 ppm. Under the same laboratory conditions, this excursion is 6.2 times smaller than that of a mechanically mounted non-PM reference comb. The amplified pulses are compressed to 94 fs and spectrally broadened in a PM highly nonlinear fiber. A coherent-population-trapping-based chip-scale atomic clock is employed for the first time as the radio-frequency reference for comb locking, thereby reducing the footprint of the microwave-reference subsystem. The resulting in-loop, optical-frequency-normalized tracking instability reaches 6.03E-17 at 1 s and 2.57E-18 at 10,000 s. These results demonstrate a compact comb architecture with reduced sensitivity to environmental perturbations, providing a practical route toward portable optical atomic clocks.

Reusable wearable PPG representations should capture local pulse morphology and broader physiological regularities. Existing pretraining objectives, however, usually apply supervision at the recording or pair level, overlooking that different concepts become reliable over different scopes of waveform evidence. We propose HiPPT, a physiology-grounded framework that matches each supervision source to its appropriate evidence scope. We construct a PPG-Text corpus of over 4 million wrist-worn recordings from 202K users, pairing recordings with waveform-derived morphology and metadata-derived context descriptions. HiPPT combines record-level morphology alignment, which grounds individual recordings in pulse characteristics, with cross-recording context alignment, which aggregates recordings sharing the same metadata tuple to capture context-consistent regularities. Waveform-guided text modeling further recovers masked contextual concepts from PPG patch representations, providing fine-grained waveform-text supervision. Across six downstream tasks spanning blood pressure, vascular stiffness, physiological attributes, and cross-dataset heart-rate estimation, HiPPT achieves the best result on the primary linear-probing metrics. The aligned space also supports zero-shot profiling of age group, sex, and BMI category, individually and jointly, with performance improving as more recordings are aggregated. These results show that evidence-scope matching yields transferable and semantically accessible wearable PPG representations. Code and pretrained weights will be released.

Coding agents read repositories and web pages that anyone can write, and then act with their user's authority. Prompt-injection defenses make it less likely that the model is steered but do not prevent it, so the system around the model must contain an agent that is already compromised. Many exploits against deployed coding agents share one cause: a check approves a name, such as a path, an argument vector, or a directory, and a later step maps that name to a different object: the file behind a symlink, the behavior an option selects, or a repository that git discovers on its own. We present Zero-Trust Agent Cells (ZTAC). A cell runs the agent in a kernel sandbox whose only route to files, programs, and the network is a socket to a broker. The broker resolves each request to the object it will act on, decides on that object, and checks at use that it is acting on the object it authorized. Programs that the broker starts run in a sandbox compiled from the same policy, and approvals are bound to the resolved object and can be used once. Our macOS prototype has 3.2k lines of trusted Rust code, excluding dependencies. We evaluate it against 74 attacks with ground-truth oracles, including replays of 10 publicly disclosed incidents. A tool allowlist loses 66 of them, and an allowlist combined with the kind of command sandbox that deployed agents use loses 56. ZTAC loses 2, both of which encode a secret into traffic to an allowed endpoint, and holds 14 for review. It completes 43 of 46 benign operations, with 8 approval prompts, and 4 of 7 workflows. A no-op brokered request has a median latency of 4.0 ms, and the spawned programs we measured take 8.5 ms to 18.4 ms longer than without ZTAC.

Generative recommendation models based on autoregressive decoding have been adopted as alternatives to classical ranking-based systems. While their modeling flexibility has been widely studied, their resilience to data poisoning remains largely unexplored. In this work, we ask: how susceptible are generative recommenders to conditional targeted poisoning? Under a realistic setting where an adversary can inject limited malicious interactions and query a recommendation API as a black box, our empirical study shows that generative recommenders are vulnerable: injecting fewer than 1% synthetic interactions significantly increases the likelihood of promoting malicious targets under the trigger. To characterize this vulnerability, we analyze the synthesis of malicious interactions from two perspectives: (i) decoding alignment and (ii) prefix popularity. We observe that short, context-aligned prefixes that are both common under the model's induced transition dynamics and strongly linked to a frequent trigger consistently amplify targeted promotion. Motivated by this analysis, we develop AlignPoison, a black-box framework that discovers and samples context-aligned prefixes from API outputs. We conduct experiments on benchmarks in unimodal and multimodal generative recommendation settings. Results show that these prefixes outperform competitive baselines, improving ASR lift by up to 16.3 percentage points and highlighting a decoding-dependent, context-sensitive poisoning vulnerability in generative recommendation models.

An edge labeling of a connected graph $G=(V, E)$ is called a local antimagic labeling if it is a bijection $f: E\rightarrow \{1, 2, \cdots, \left | E \right | \}$ such that for any pair of adjacent vertices $u$ and $v$, $f^{+}(u)\neq f^{+}(v)$ where $f^{+}(x)=\sum f(e)$, with $e$ ranging over all the edges incident to $x$. The local antimagic chromatic number of $G$, denoted $\chi_{la}(G)$, is the minimum number of distinct induced vertex labels across all possible local antimagic labelings of $G$. This paper investigates two families of graphs $G(m^{[n]})$ (a graph class constructed by augmenting a bipartite structure $B(m^{[n]})$ with cyclic edges) and $K_{1,s}\times C_n$ (the direct product of a star $K_{1,s}$ and a cycle $C_n$). Under the conditions odd integer $m\ge 1$ and even integer $n\ge 4$, we establish the exact values of $\chi_{la}(G(m^{[n]}))$ and $\chi_{la}(K_{1,s}\times C_n)$ for various $s$. Furthermore, we characterize several graph families achieving $\chi_{la}(G)=2$ or $3$.

Most of the mass of the visible universe originates from the strong interaction. Quantitatively reproducing the origin of the nucleon mass is one of the important challenges in fundamental science. Since the next-generation Electron-Ion Collider has identified resolving this problem as one of its major goals, further theoretical studies are needed to develop approaches to investigating the origin experimentally. In this paper, we explore the possibility of evaluating the “gluon condensate term” which is particularly difficult to determine among the contributions to the nucleon mass, using high-energy scattering experiments. We first express the gluon condensate term in terms of experimentally accessible twist-4 lightcone matrix elements. We then derive, for the first time, analytic formulas for the cross section including all twist-4 contributions from pure quark, quark-gluon, and pure gluon matrix elements in $ep$ collisions, and show how these contributions can be related to the gluon condensate term.

We derive Bondi four-momentum balance between past and future null infinity in the radiative phase space of the null-surface formulation (NSF). Weyl matching at spatial infinity, together with vanishing endpoint News, gives antipodal continuity of the Bondi mass aspect. Its translation projection is independent of the infrared shear term and transforms as a Lorentz four-vector. Bondi evolution relates the incoming and outgoing News fluxes to the timelike endpoint momenta; in the Dominguez–Kozameh–Ludvigsen radiative sector these momenta vanish and the fluxes are equal. On a common radiative Fock domain, unitary conjugation transports the flux observables. Adopting their classical balance as an additional quantum condition yields exclusive four-momentum conservation. We also calculate the outgoing energy expectation without imposing that condition. For an incoming energy eigenstate, the second-generator commutator vanishes diagonally, leaving a second-order coefficient determined by the cubic generator. Direct canonical contractions isolate the contribution connected to all four external legs. For the equal-energy center-of-mass preparation, axial rotational covariance forces the angularly integrated exchange coefficient to vanish under the finite helicity-projection prescription of Part~I. The connected energy defect therefore vanishes through second order. The full expectation (including spectators) and higher-order balance remain open.

Many large enterprise and cloud deployments rely on metrics observability solutions (e.g., Prometheus) to monitor key performance indicators of system health (e.g., CPU, memory, network, app-level statistics). Monitoring and analyzing these metrics presents a significant challenge in terms of cost-latency tradeoffs at scale. Ideally, system operators would like immediately deployable and backwards compatible solutions to improve the cost-latency tradeoffs without significant impact on the fidelity of their observability applications. Unfortunately, existing solutions including state-of-art timeseries databases and custom approximate querying solutions fail on one or more aspects. In this paper, we present the design and implementation of TurboProm, a backwards-compatible drop-in accelerator for metrics observability deployments. TurboProm recasts metrics analytics tasks using a semantic-preserving summary view rather than classical query-centric view. Instead of estimating query results on raw timeseries metrics data, TurboProm uses semantic-preserving summaries that enable high-fidelity analytics with significant cost-latency improvements. We present a practical implementation of TurboProm that can be retrofitted to any existing observability deployment. We evaluate TurboProm on diverse real and synthetic workloads against state-of-art solutions and show that it can provide significant reduction in query latency and cost. TurboProm is open-source at https://github.com/ProjectASAP/ASAPQuery.

Autonomous Aerial Vehicle (AAV) trajectory planning requires real-time and accurate safety gradient guidance, especially in dynamic environments. Existing methods typically rely on static maps and dynamic obstacle tracking to construct gradient fields, imposing heavy computational burdens and yielding conservative trajectories. To address these limitations, we propose a two-stage Field-to-Planning (F2P) learning framework to provide accurate and efficient safety gradients for trajectory planning in dynamic environments. In the first stage, we introduce an Ego-centric Polar Occupancy (EPO) representation that naturally matches the geometry of the onboard camera frustum, while supporting unstructured scene modeling. Subsequently, we formulate a Temporal Polar Gradient Field (TPGF) via multi-kernel separable Gaussian convolution to provide unified safety gradients, without explicitly distinguishing between static and dynamic obstacles. To mitigate the inherent spatial ambiguity introduced by applying convolutions within the polar grid, we propose an ESDF-guided offline learning strategy that incorporates geometric priors into TPGF. In the second stage, we employ the calibrated TPGF to guide the trajectory planning policy in a self-supervised learning manner. Comparative results with other approaches demonstrate the superiority of our method. Extensive real-world experiments within various cluttered dynamic environments validate the real-time performance and robustness on resource-constrained onboard systems.

Large Language Model-based Multi-Agent Systems (LLM-based MAS) have shown remarkable success in solving complex tasks by coordinating specialized agents through multi-step workflows. To overcome the high cost and specificity of manually designed workflows, recent research has shifted toward learning them automatically. However, a key limitation remains: while agent calls introduce significant latency due to model inference time and service congestion, most automated methods optimize solely for downstream performance, ignoring execution time (makespan). This issue is amplified when workflows are reused across recurring queries, making latency accumulate over time. This limits deployment in recurrent applications such as AI customer support and AI cloud troubleshooting, where users expect fast responses. To bridge this gap, we introduce a new problem, termed latency-aware workflow generation, where the objective is to minimize workflow makespan while aiming to retain downstream performance. We propose LAWA (Latency-Aware Workflow optimizAtion), a novel framework that reduces makespan via theoretically grounded graph edits targeting critical-path bottlenecks. By unifying a graph generator with this graph-editing procedure, LAWA supports both de novo workflow generation and refinement of existing ones. Empirical results across seven benchmarks demonstrate that LAWA significantly reduces makespan while achieving competitive task performance, supporting efficient reuse of multi-agent workflows for recurrent queries.

The observation of superconductivity emerging from a spin- and valley-polarized normal state in rhombohedral graphene raises a basic puzzle: intravalley Cooper pairs involve states that are not generically degenerate because of trigonal warping, and therefore lack the particle–particle nesting underlying the conventional Cooper instability. While recent model studies suggest that nontrivial band geometry can strongly promote chiral pairing, how this geometric enhancement competes with realistic band-structure-induced pair breaking remains unclear. Here we study Kohn–Luttinger superconductivity in electron-doped rhombohedral tetralayer graphene using a DFT-fitted multiorbital tight-binding model, retaining the full Bloch-wave-function form factors in both RPA screening and the projected Cooper interaction. We separate the superconducting instability into three contributions: the symmetry-resolved form-factor-dressed interaction, the Cooper-pair phase space determined by the symmetrized pair dispersion, and the energy mismatch generated by trigonal warping. We find that the Bloch form factors strongly favor intravalley pairing, with a leading predominantly $p_x-ip_y$ instability and a substantially larger maximum $T_c$ than the intervalley reference state. At the same time, trigonal-warping-induced mismatch cuts off the Cooper logarithm and confines intravalley superconductivity to a much narrower region of the phase diagram. The dominant instability occurs near pair-Lifshitz transitions and primarily in reconstructed annular or three-lobed Fermi-surface regimes. Our results show that superconductivity in rhombohedral graphene is controlled by the combined effects of quantum-geometry-dressed interactions, pair fermiology, and trigonal-warping-induced pair breaking.

While language therapy is the primary treatment for post-stroke aphasia, limited clinical resources often restrict patient access to proper care. Addressing this gap requires reliable training simulations for medical professionals. However, current AI efforts focus almost exclusively on speech processing, leaving speech generation largely unexplored. This study introduces AphaVoice, a novel text-to-speech (TTS) framework explicitly engineered to synthesize highly realistic aphasic speech. Evaluated against state-of-the-art TTS systems, AphaVoice demonstrates a superior ability to capture the clinical feature of dysfluency, outperforming existing models in temporal metrics while consistently maintaining speaker identity. By leveraging latent space shifting, the model also supports multi-speaker generation without sacrificing phonetic realism. Ultimately, AphaVoice provides a scalable, high-fidelity tool for medical training and clinical simulation related to aphasia.

Omnidirectional aerial robots offer full 6-DoF independent control over position and orientation, making them popular for aerial manipulation. Although there have been advancements in robotic autonomy, human operation remains essential in complex aerial environments. Previously, we proposed an aerial teleoperation system that brings the rotational flexibility of human hands into the unbounded aerial workspace, which was evaluated through a vertically mounted valve-turning task. However, due to the limited range of human forearm's motion, the robot could not continuously rotate the valve, hindering its application in industry. In this work, we propose an approach to initiate continuous rotation when the forearm angle exceeds a threshold. This switch of control authority exemplifies the concept shared control from human-robot interaction, marking a further step in aerial teleoperation.

Robustness tests for tool-using agents often change the agent's environment, for example by hiding a hostile note in an invoice, and then judge the agent by what it does. The result tells us about the original task only if the changed test case is still a valid test of that task. Many evaluations decide this with weak rules, such as accepting any well-formed edit or any edit that stays close to the original in embedding space. We propose executable task contracts. A contract states which facts define a task, which edits are allowed, and what finishing the task means, and it checks each changed test case before the agent runs. We compare validity rules by scoring the same 14,336 stored agent runs under each rule. Across four encoders, embedding thresholds allowed 75-100% of edits that changed the task, such as swapping the payee, yet rejected 24-45% of valid hostile notes. With a MiniLM cosine threshold, Mistral appeared vulnerable on all 64 tasks, but only 26 had a successful valid attack. Replaying a saved solution and allowing edits to benchmark fields fail in similar ways. Evaluations should state, and check, what counts as the same task.

Comparing large language models can reveal shared representational structure and systematic differences between them. However, aggregate similarity measures can obscure how agreement varies across geometric scales and which linguistic properties support it. We introduce two complementary methods to resolve where models agree and what supports that agreement. Standard representational similarity analysis (RSA) summarizes similarity across all stimulus pairs, potentially obscuring differences in alignment between nearby and distant inputs. Neighborhood-Weighted RSA (NW-RSA) addresses this limitation by weighting stimulus pairs according to their proximity in representational space, placing alignment from local neighborhoods to global geometry on a common scale. Counterfactual RSA (CF-RSA) tests which linguistic properties support alignment by comparing intact representations in one model with representations under controlled linguistic perturbations in another. Across diverse language models, we find strong agreement in coarse representational geometry but substantial divergence in local neighborhoods. Alignment is observed both across broadly sampled natural text and within linguistic equivalence classes that preserve particular aspects of meaning or structure. Counterfactual analyses reveal selective sensitivity: changes to logical scope reduce alignment substantially more than the tested changes to hierarchical syntax, event roles, or discourse structure. Alignment also decreases when sentence structure or lexical meaning is disrupted in word lists and jabberwocky sentences, suggesting that the regularities of natural language support stronger convergence. These findings refine claims of representational universality by showing that strong overall similarity need not imply agreement in how models represent finer linguistic distinctions.

Deploying language models (LMs) on the edge presents unique memory, latency, and power challenges absent in high-performance computing. KV-cache compression emerges as a promising approach to address these challenges. However, the effectiveness of current KV-cache compression methods has not been thoroughly examined on edge devices. Moreover, we show that existing evaluation methods do not always involve the KV-cache under test. We propose EdgeLLM-Nano, a measurement study of KV-cache compression using a nano-scale decoder-only LM (transformer) testbed that has 8.39M parameters, i.e., 33.6 MB, to enable end-to-end examination of each backend, seed, and power sample. Our results give a negative verdict for a polar-guided key-cache quantizer (round-to-nearest (RTN) control causes an order of magnitude less damage), as well as a consistency-anchored progressive weight pruner (one-shot magnitude-based pruning performs as well as, if not better). We also introduce Long-Context Decode Fidelity (LCDF) evaluation, a token-by-token examination of teacher-forced log-losses using the KV-cache under test. Unlike perplexity that gives the same 5.461 score to all methods, LCDF distinguishes different KV-cache compression backends thanks to its paired statistics across seeds and effect sizes. Our companion Raspberry Pi 5 measurement shows all compression backends we tested to be less energy-efficient per token compared to uncompressed FP16 inference. Except for KIVI that is close to lossless in terms of LCDF, our findings indicate KV-cache compression can be improved. LCDF offers a simple and effective way of shedding light on KV-cache compression methods, whose negative impacts have been overlooked by perplexity.