Peer Revenue Past citations are no guarantee of future returns

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Friday 11 September

Analysis of PDEs

Optimal regularity of solutions to a singular cooperative system

In the pioneer work of vectorial free boundary problems \cite[Adv. Math. 280 (2015)]{ASUW15}, Andersson, Shahgholian, Uraltseva, and Weiss investigated the regularity theory of solutions and the free boundary of a singular cooperative system \begin{equation*} \Delta\mathbf{u} = \frac{\mathbf{u}}{|\mathbf{u}|} \chi_{\{|\mathbf{u}|>0\}} \quad\text{in }\Omega\subset\mathbb{R}^n, \qquad \mathbf{u}:\Omega\longrightarrow \mathbb{R}^m, \qquad n,m\ge 2. \end{equation*} At the level of solutions, the boundedness of the right-hand side directly yields $W^{2,p}_{\mathrm{loc}}$ regularity for every $1<p<\infty$ by standard elliptic estimates, while the corresponding $W^{2,\infty}_{\mathrm{loc}}$ regularity question was explicitly left as an open problem on page~753 of that paper. In this paper, we give a positive answer by establishing optimal $W^{2,\infty}_{\mathrm{loc}}$ regularity through a scale-invariant interior Hessian estimate for every weak solution. Unlike the scalar case, the classical Alt–Caffarelli–Friedman monotonicity formula cannot be applied directly to the present system. Our proof instead combines a localized Newtonian potential decomposition with an explicit projection onto quadratic harmonic polynomials on the unit sphere, which isolates the symmetric trace-free quadratic part of the Hessian. A key new observation is that the resulting matrix coefficients satisfy an exact linear ordinary differential equations with constant coefficients on the logarithmic scale. The main novelty of this work lies in a so-called two-regime argument based on the competition between the affine and quadratic parts of the rescaled solution, together with a continuity argument that connects the two regimes and yields a uniform bound for the coefficient. The resulting a priori estimate provides an analytic basis for further study of the fine structure of the free boundary.
Read the full abstract

Signal Processing

Progressively Attenuated Multi-Branch Reception for Inter-HAPS Optical Links

Inter HAPS optical links can experience receiver saturation at short separations and become signal to noise ratio (SNR) limited at longer distances. This work proposes a saturation aware spatial multi branch intensity modulation direct detection receiver using nonoverlapping apertures with progressive attenuation and selection of the nonsaturated branch with the highest instantaneous SNR. A Gaussian beam model with finite aperture pointing loss and common two dimensional pointing jitter yields closed form branch level expressions for saturation, insufficient SNR, usability, and outage probabilities. Results show that additional branches suppress short range saturation outage, shift the minimum outage point toward shorter separations, and enable progressive attenuation to control the saturation SNR tradeoff. Monte Carlo simulations closely match the analytical results. Overall, the proposed architecture provides a low complexity approach to extending the usable dynamic range of inter HAPS optical receivers.
Read the full abstract

All authors Meysam Ghanbari, Betul Beidas, Wasiu O. Popoola, Mazen O. Hasna, Khalid A. Qaraqe, Mohammed Elamassie

Artificial Intelligence

$\tau$-Elicitation: Benchmarking multi-turn entity extraction in voice agents

Voice agents often need to collect names, addresses, identifiers, dates, and times exactly, yet end-to-end benchmarks obscure where capture fails. We introduce $\tau$-Elicitation, a 200-task voice benchmark spanning 10 entity types, controlled difficulty, caller realisms, and three environments. A matched text agent passes all tasks, but four voice configurations achieve robust exact success from 0.14 to 0.41. Agents increase verification for hard and unfamiliar entities and sometimes for incorrect captures, but not for their weakest caller voice; only 24 to 37 percent of verified errors are repaired. A scaffold that prescribes spelling, read-back, correction, and confirmation raises robust Pass$^3$ by 14 to 31 points, at a cost of 21 to 28 seconds per call. Realisms such as spelling variations and restarts do not detectably affect exact success; mispronunciation increases repair effort. These results identify strategy selection and successful recovery as the central bottlenecks in exact spoken entity collection.
Read the full abstract

Cryptography and Security

Mind the Gap: Detecting Description-Execution Mismatch Attacks in DAO Governance

Decentralized autonomous organizations (DAOs) change protocols through a proposal-based process: initiators submit a proposal, members vote on it based on its natural-language description, and if it passes, the project executes the code behind it. This process is inherently vulnerable to deceptive proposals, where the described intent and the actual code execution mismatch. A malicious proposer can submit a benign-looking description to pass voting while the executed code transfers funds or seizes control of the protocol, which we call a Description-Execution Mismatch (DEMI) attack. We present the first systematic framework for DEMI detection in real DAO governance. First, the diversity of DAO deployments makes a unified, scalable analysis difficult; we address this with a DAO-agnostic simulation framework that builds a per-DAO governance profile from historical on-chain transactions and then drives each new proposal through the full governance lifecycle to obtain its execution behavior. Second, free-form descriptions and structured execution traces are hard to compare; we address this with an evidence-mapping paradigm that requires an LLM to locate explicit per-action textual justifications rather than issue a holistic judgment, substantially improving precision and recall over direct querying. On a large-scale dataset of real-world Ethereum DAO governance, our simulation derives execution results for 92.7% of active DAOs and 89.3% of executed proposals, far exceeding existing platforms. Our detector reaches 81.7% mean precision and 98.3% mean recall under stratified cross-validation, and evidence mapping generalizes across LLM vendors rather than depending on one model. Under a systematic red-team/blue-team evaluation, the Robustness Guard defends most adaptive attacks even against an adversary that knows the detector, and this robustness generalizes to held-out proposals.
Read the full abstract

All authors Bowen Cai, Nanzi Yang, Weiheng Bai, Youshui Lu, Yajin Zhou, Kangjie Lu

Quantum Physics

Braid map definition of SLH-networks and the series product

This article investigates a cascade of quantum systems within an SLH-network. The original definition of an SLH-network depends on an open conjecture regarding the essential selfadjointness of the generator of the time evolution of the network. This problem is avoided by using the so-called braid-map, which leads to a rigorous definition of the SLH-network. An expression is derived for the time-evolution of the network with finite delay. The series product is recovered in the instantaneous feed-forward limit as a theorem that no longer depends on the conjecture.
Read the full abstract

Algebraic Geometry

Degrees of Fano Quiver Moduli

This paper provides methods to compute degrees of Fano quiver moduli and gives explicit degree formulas for two classes of them. One such class is a series of toric Fano varieties that parametrize isomorphism classes of representations of the bipartite quiver $K_{(2,q)}^m$ with dimension vector $\underline{1}$, and the other class is the moduli space of q-point configurations on $\mathbb{P}^1$. In the former case, we obtain a formula comprised of sums of binomial coefficients in $q$ and $m$. Its special case $m = 1$ turns out to be the same as the central MacMahon numbers (OEIS entry A177043). In the latter case, we obtained a formula over $\mathbb{Q}$. These formulas are much more efficient than computing directly in Chow rings.
Read the full abstract

Human-Computer Interaction

Reconceptualizing Age Assurance as a Sociotechnical Problem: Connecting Evidence, Evaluation, Claims, and Decisions

Age verification is often treated as a technical problem: can a system determine a child's age accurately? We argue this framing is too narrow. Age assurance becomes consequential when evidence is evaluated, translated into age-related claims, and used to decide whether a person can access, purchase, or belong. We review 85 publications on children's age assurance published from 2020 through February 2026. We find that shared terms such as age verification describe different processes. Age is represented as threshold eligibility or inferred estimation, and the same eligibility claim can arise from different evidence and components. Rights, access, and privacy receive more attention than accuracy, error, and fairness; yet institutional actors are rarely connected to system failures or user remedies, an accountability gap. We introduce the Age-Assurance Process Framework, which treats age assurance as a sociotechnical process connecting evidence, evaluation, claims, and decisions rather than reducing it to a technical problem.
Read the full abstract

All authors Renkai Ma, Prakriti Dumaru, Thomas Synaepa-Addison, Jess Kropczynski, Pamela J. Wisniewski

Computational Finance

Same Book, Different Fills: Partial Identification of FIFO Execution from Aggregate Order Books

Price-level limit order book (L2) data reveal aggregate liquidity but not the ordered queue required by price–time priority. Passive-execution backtests can therefore depend on an unobserved cancellation-allocation rule even when observed prices, quantities, and trades are held fixed. We frame recovery of market-by-order histories from aggregate snapshots as a conditional partial identification problem: multiple histories can reproduce the same aggregate path. Holding that path, reconciled market removals, latent order partitions, and additions fixed, our path-preserving compiler varies only cancellation allocation among front, quantity-weighted-random, and back rules. Within this compiler class, we establish front–back fill ordering for a virtual tagged order during one touch-price spell. We study seven months of synchronized 2025 Tokyo Stock Exchange data for two instruments with different trading activity: RIC 1301.T and RIC 7911.T. Ten-level L2 snapshots provide book states, while L1 trades permit inference of market removals and resting side through reconciliation. Each instrument contributes 1,080 matched five-minute episodes over the same 18 held-out trading days. The aggressive benchmark is invariant across FIFO realizations, but passive execution is sensitive to the cancellation rule. For 1301.T, front rather than back cancellation raises preterminal completion by 8.01 percentage points and reduces implementation shortfall by 1.010 bps. For 7911.T, the corresponding differences are 7.39 percentage points and 0.384 bps. Thus, observationally equivalent aggregate-book paths can imply economically different passive-execution outcomes. Execution policies evaluated from aggregate data should be accompanied by FIFO sensitivity analysis rather than reported as single-point estimates based on an unobservable queue assumption.
Read the full abstract

Symbolic Computation

Certified local rank and uniqueness barriers for a 48-term matrix-multiplication decomposition

We study replacements in fixed bilinear tensor decompositions, counting changes to complete rank-one summands, including output factors. The shortening frontier records the maximum rank defect of a fixed-size subset and determines the minimum length attainable within a change budget. For the rational 48-term Li–Wang–Hu decomposition \(D(2)\) of \(4\times4\) matrix multiplication over \(\mathbb{C}\), we prove rank radius at least 12, strong radius exactly 11, and border radius at least 8. Every shorter complex decomposition therefore changes at least thirteen original summands. An exact rational twelve-term replacement attains the equal-length barrier. The proofs combine exhaustive support reductions with saturated projected kernels and zero-corner completion arguments controlling arbitrary minimal competitors. A reduced-incidence argument transfers kernel certificates to tensor-space neighborhoods. A Laurent normal form gives strong radius exactly 11 for the sixteen-term core at every nonzero complex parameter. On a nonempty Zariski-open subset of the actual parameter curve, the rank radius is at least 12, the strong radius exactly 11, and the border radius at least 8. We also prove incomparability of the full Kothari–Moitra–Wein sufficient criterion and the Sylvester-equipped kernel criterion. These results describe local decomposition structure rather than a new rank bound for full matrix multiplication.
Read the full abstract

Sound

Predictive audio representations for early detection and tracking of hidden dynamic objects

Predicting potential dangers is core to safety. Forecasting the presence of other traffic agents is core to danger prediction. Occluded traffic agents challenge detection systems as they might become visible too late, leaving the autonomous vehicle too little time to identify, plan and act accordingly in a robust and safe way. Previous works proved that auditory perception, being omnidirectional and not constrained by a field-of-view, provides fundamental cues for early spotting of different road users, even when hidden by other vehicles or infrastructures. Yet, those contributions deal with scenarios with only one vehicle present, and they either identify the type of vehicle or estimate its direction of arrival. In this work, we move forward, and propose a multi-task system which, simultaneously, estimates the number of vehicles present, their type, and their direction of arrival. Our methodological contribution is a two-stage pipeline: a self-supervised pre-training stage inspired by the Joint- Embedding Predictive Architecture (JEPA) applied directly to multichannel raw waveforms, followed by supervised multi-task fine-tuning with a bidirectional LSTM and three classification heads. The pre-training stage trains the encoder without labels, pushing it to predict the latent representation of a future audio segment from its past context. To train and test our framework, since no suitable dataset was publicly available, we collected an ad-hoc one covering Non-Line-Of-Sight scenarios with multiple traffic agents simultaneously operating. Experimental evaluation shows that our method outperforms the state of the art; it also proves that our design choice allows the model to learn robust representations, which can be transferred to an unseen driving scenario, maintaining reasonable and stable performance.
Read the full abstract

All authors Katerina Vinciguerra, Moritz Brandes, Danilo Hollosi, Letizia Marchegiani

Statistical Methodology

Distribution-Free Changepoint Inference for Rainfall Patterns

Long records of rainfall contain critical information about the temporal organization of precipitation within a year, including wet and dry spells. Detecting when this temporal pattern changes is central to climate-impact assessment and environmental monitoring.A common data structure features observations from m locations over T years. For location i and year t, the record is a vector of daily measurements $X_{i,t} = (X_{i,t,1}, ..., X_{i,t,D})^\top$, with D = 365. The goal is to identify an unknown year $t_0$ when the temporal rainfall pattern changes across the region. Because changes in the intra-annual pattern affect the frequency-domain behavior of the daily sequence, we represent each yearly curve through a spectral feature capturing the strength of temporal oscillation at a fixed frequency.Classical changepoint methods often rely on parametric likelihoods or Gaussian approximations, which are hard to justify for skewed and heavy-tailed rainfall data. Furthermore, many procedures only return a point estimate. A confidence set for the change year is crucial in environmental applications to quantify temporal uncertainty and distinguish sharp transitions from weak evidence.This paper develops a distribution-free framework for detecting a synchronized structural change in yearly rainfall patterns across independently monitored locations. For each location-year pair, we compute a fixed-frequency nonparametric spectral density estimate. These features form a $T \times m$ data matrix. By exploiting the exchangeability of yearly spectral features before and after the true changepoint, alongside spatial independence, we construct finite-sample valid confidence sets for the unknown change year.
Read the full abstract

High Energy Physics (Experiment)

Density-Guided Conditional Neural Processes for Detector Efficiency Estimation

Selection efficiency connects observed counts to physical rates and rare-event search limits. With scarce calibration data, strong smoothing can erase sharp physical changes, while highly flexible models can overfit the data and mistake statistical fluctuations for physical structure. We introduce a density-guided conditional neural process (CNP) that uses spectral concentrations to control where high-frequency features are available. On MAJORANA data, it reconstructs local structure while matching measured efficiencies within twice their binomial statistical uncertainty in an average of 89% of bins across two continuum windows, versus 37% for Attentive CNP with positional encoding. With 5,000 efficiency-model training events and 500 context events, it exceeds CNP and Attentive CNP trained on twice as many events in both peak-core and continuum summaries, and also exceeds context-only KDE and Bernoulli GP. Our method thus balances the reconstruction of sharp physical features with resistance to statistical overfitting, and achieves this with fewer training events.
Read the full abstract

Distributed Computing

BOOST: Concurrent Access to Host Memory and HBM to Accelerate LLM Inference

GPU memory bandwidth and capacity limit throughput in large language model (LLM) inference. The GPU memory system consists of a primary tier of high-bandwidth memory (HBM) and a secondary tier of host memory connected via CPU-to-GPU interconnect. Current serving systems treat the tiers hierarchically: they serve exclusively from HBM when data fits, and otherwise prefetch data from host memory to HBM before use. In both cases, the host memory bandwidth is never well utilized. Prefetching expands capacity by utilizing host memory, but consumes HBM bandwidth for writes, reducing the bandwidth available for demand loads. We observe that fully utilizing both host and HBM bandwidth requires each wave of GPU threadblocks to access both tiers concurrently and in proportion to their bandwidth ratio. Existing bandwidth-proportional placement strategies fail to provide concurrency because they are not aware of GPU waves, and the large 2MB GPU page size. This paper presents BOOST, the first runtime system that provides concurrent and proportional access to both GPU memory tiers, extracting the combined bandwidth of host memory and HBM for LLM inference without kernel changes. The key insight in BOOST is to use kernel access patterns to make page allocation and runtime data management wave-aware. For static model weights, BOOST applies modulo-based page placement that eliminates access-ratio variance; for dynamically provisioned attention key-value (KV) pairs, it makes the free KV page pool wave-aware. We integrate BOOST into vLLM and evaluate on a Grace Hopper system. At iso-batch size, BOOST improves Time-per-Output-Token (TPOT) by 4.3% over HBM-only serving, whereas prefetching degrades TPOT by 6%. In high-throughput serving, BOOST improves throughput by 31% on average, outperforming prefetching by 15%.
Read the full abstract

All authors Anish Saxena, Jae Hyung Ju, Hritvik Taneja, Po-An Tsai, Aamer Jaleel, Christos Kozyrakis, Moinuddin Qureshi

Systems and Control

Predicted-Flow Control Barrier Functions for Non-Control-Affine Systems

Control barrier functions (CBFs) enforce safety through conditions imposed pointwise in time without consideration of state evolution over a future horizon. Thus, CBF-based controls are typically myopic. Predicted-flow CBFs (P-CBFs) generalize CBFs from a function of the current state to a functional of the predicted flow under a parametrized control. P-CBFs can certify safety over the entire prediction horizon while simultaneously allowing for performance optimization over the horizon. However, prior work with P-CBFs only applies to control-affine systems, and suffers from a limitation where the prediction horizon can shrink or even vanish. This article addresses both of these shortcomings by introducing a planning control that smoothly transitions from a parametric plan to a backup control (if needed) and a time-shift parameter that determines if transition is needed. The evolution of the control-plan and time-shift parameters is determined from a single guaranteed-feasible convex optimization, which reduces to a quadratic program (QP) if the control limits are a convex polyhedron. The real-time executed control is determined from the instantaneous planning control and time shift. This method simultaneously addresses safety certification and performance optimization over the fixed prediction horizon. The approach is compared to nonlinear model predictive control in simulation of an autonomous car navigating a dense obstacle environment.
Read the full abstract

Statistical Methodology

Bayesian Consensus Calibration of Continuously Evolving IRT Item Banks

AI-based item generation and NLP-based prediction of item parameters are producing item banks that are substantially larger, sparser, and more frequently updated than conventional banks. Hierarchical Bayesian item response theory (IRT) is a natural calibration framework for such banks, but the common practice of refitting the entire accumulated response history at each update is costly and can exceed available memory. We describe consensus calibration, a divide-and-conquer procedure that calibrates each time period independently and reconstructs the pooled posterior in two layers. First, the posterior draws of each period are mapped to a common metric by a robust characteristic-curve linking (Haebara) that is solved separately for each draw, which propagates the uncertainty of the linking transformation into the linked posteriors. Second, the linked item posteriors are combined as a product of Gaussian densities from which the population prior contributed by each period is removed and a single prior—obtained by consensus across the per-period population posteriors—is reinstated. The correction targets the posterior dispersion, not only its location. As evidence for consensus calibration, we compare it to a pooled single-run analysis on a large operational assessment in terms of item-parameter recovery, an uncertainty-by-exposure diagnostic, and the ability distributions.
Read the full abstract

All authors Paul A. Jewsbury, Steven W. Nydick, Manqian Liao, Siyuan Chen

Dynamical Systems

Isomorphisms and slow entropy of deterministic $[T,T^{-1}]$ systems

We study skew product measure-preserving systems driven by an irrational rotation and the step function with values $1$ and $-1$ on each half of the circle. These systems can be seen as a deterministic version of Kalikow's $[T,T^{-1}]$ system, and are particular instances of Rokhlin cocycle extensions. Under the assumption of ergodicity, we show that these systems obey an interesting rigidity property for isomorphisms. That is, we show that two skew products are isomorphic if and only if the rotations in the base are isomorphic, and the fiber transformations are flip isomorphic. Under mild extra assumptions, we are able to characterize all isomorphisms. We prove that by choosing the rotation angle suitably, these systems can realize arbitrarily low measure-theoretic complexity in the sense of lower slow entropy. We apply this result to obtain new examples of systems that fail the variational principle for slow entropy. These examples can have a set of invariant measures as rich as desired (any metrizable Choquet simplex, up to affine homeomorphism).
Read the full abstract

Robotics

Differential Realizability of Static Control Allocation in Multirotors: An Impossibility under Nonredundant Full Actuation and a Pseudoinverse Obstruction under Redundant Actuation

Control allocation for multirotors with bidirectional propellers is commonly formulated in signed-thrust variables, where the wrench map is linear. The signed-quadratic map from physical rotor speed to thrust, however, is not a local diffeomorphism at zero speed. This work derives two distinct consequences. Under nonredundant full actuation, a single-propeller reversal removes one instantaneous task direction; hence, no global continuously differentiable exact static allocator exists over the complete task space. Under redundant actuation, the physical task map may remain regular, yet a transverse pseudoinverse zero crossing requires an unbounded rotor-speed derivative. We define differential realizability as regularity of the physical lift of an actuator-output section, derive exact and first-order validity conditions, and distinguish structural rank loss from an allocator- induced rate singularity. A local nullspace deformation repairs isolated pseudoinverse reversals, while a global fixed-orthant construction establishes existence of regular sections at the cost of persistent task-preserving internal actuation.
Read the full abstract

Optimization and Control

The Generalized Excursion Coupling as the Limit of Concave Optimal Transport

The Monge-Kantorovich problem with the Euclidean distance cost is degenerate, typically admitting infinitely many optimal plans. A unique optimal plan is selected by perturbing the distance to a strictly convex or increasing strictly concave cost and passing to the limit: the convex side gives the plan induced by the map which is monotone on each transport ray, while the concave side was understood only on the real line, where Juillet proved that power-cost optimizers converge to the so-called excursion coupling, which is induced by a map when the source is atomless. This paper extends the concave selection to $\mathbb{R}^n$ for mutually singular finite positive Borel measures $\mu$ and $\nu$ with equal total mass and finite first moments, assuming that $\mu$ is absolutely continuous with respect to the Lebesgue measure: for a broad class of increasing strictly concave perturbations of the distance with a well-defined first-order profile, the corresponding optimal maps converge in $\mu$-measure to the same intrinsic limit $t_\#$, independent of the perturbation family and its profile. The map $t_\#$ is obtained by disintegrating the transport along its maximal rays and applying the one-dimensional excursion coupling on each ray. The same limit is established for the power costs $\|x-y\|^{1-\varepsilon}$ under a finite $\|x\|\ln(1+\|x\|)$ moment condition, establishing a conjecture of Juillet.
Read the full abstract

Statistical Methodology

Synthetic Nearest Neighbors: Extending Synthetic Controls for Matrix Completion with Missing Not at Random Data

We develop a causal framework for matrix completion under missing not at random (MNAR) data. Drawing on synthetic controls from the econometric panel data literature, our approach relaxes two assumptions common in MNAR matrix completion: positivity and independence of observation indicators. Unlike traditional panel data models, which often require prescribed block-sparse geometries, our framework accommodates flexible, heterogeneous observation patterns through target-specific local information structures. We propose synthetic nearest neighbors (SNN), a local synthetic-controls-inspired estimator, and establish finite-sample entrywise error bounds and consistency for mean recovery under suitable conditions. We further derive asymptotic normality under heteroskedastic noise and develop feasible entrywise inference. To estimate entry-specific noise variances, we apply the same local principle to squared outcomes, obtaining consistency under bounded noise and asymptotic unbiasedness under general subgaussian noise. Simulation studies corroborate the theoretical findings across a range of missingness designs and observation patterns.
Read the full abstract

All authors Anish Agarwal, Munther Dahleh, Devavrat Shah, Dennis Shen

Distributed Computing

mKernel: Fast Multi-GPU, Multi-Node Fused Kernels

Communication has become a bottleneck in distributed training and inference of large models. Overlapping communication with computation at the granularity of kernels, on separate streams, reduces only part of this communication cost. Fused kernels often have better performance by transmitting each output tile as soon as it is produced, but existing fused kernels are largely confined to a single NVLink domain. We present mKernel, a library of multi-GPU, multi-node fused kernels that overlap computation, intra-node NVLink communication, and inter-node RDMA at tile granularity. mKernel partitions the streaming multiprocessors (SMs) of a persistent kernel into compute and communication roles, and an on-GPU controller tunes the SM partition adaptively at run time, since the best SM partition varies with the kernel and the input shape. It structures data movement hierarchically so that data traversing the inter-node network is minimized. Finally, it drives the network from the GPU through a lightweight command queue and host proxy implemented directly on RDMA verbs, which allows the same kernels to run on any network backend (e.g. InfiniBand and on AWS EFA); we observe, surprisingly, that GPUDirect Async (IBGDA) yields little additional benefit over host-assisted GPU-initiated communication. We implement five kernels spanning tensor, sequence, and expert parallelism. On two 16-GPU H200 clusters, mKernel achieves speedups of up to 1.72x on GEMM+AllReduce and $1.88\times$ on Ring Attention.
Read the full abstract

All authors Ziming Mao, Yihan Zhang, Shawn Wei Chew, Shuang Ma, Costin Raiciu, Yang Zhou, Scott Shenker, Ion Stoica

Quantum Physics

Identifiability in Quantum State, Process, and Network Tomography

Quantum State Tomography (QST), Quantum Process Tomography (QPT), and Quantum Network Tomography (QNT) are related parameter-estimation problems that aim to reconstruct different physical quantities. QST estimates an unknown quantum state, represented by its density matrix, from the measurement outcomes. QPT characterises an unknown quantum channel using known input states and measurements of the corresponding outputs. QNT, in contrast, aims to infer parameters associated with individual links from end-to-end probe measurements collected at accessible monitor nodes. A key distinction among the three tomography problems lies in the conditions required to achieve identifiability, the ability to determine unknown parameters uniquely from the available measurement statistics. In QST and QPT, the experimenter can choose an Informationally Complete (IC) measurement set. QNT limits the reachable measurements to what topology and monitor placement allow, so the admissible probe paths fix the information available about the link parameters. This work studies all three tomography problems through a common Fisher Information Matrix (FIM). We factorise QNT FIM and show that its rank equals the rank of the path-link incidence matrix at every interior parameter value, so local and global identifiability coincide. We then show that QST and QPT attain full rank under IC settings, QNT loses rank when the probe paths leave link parameters indistinguishable, and increasing the number of copies scales the FIM eigenvalues while leaving its rank fixed.
Read the full abstract

All authors Athira Kalavampara Raghunadhan, Matheus Guedes De Andrade, Don Towsley, Indrakshi Dey, Daniel Kilper, Nicola Marchetti

Materials Science

Reducing the Gate-Induced Drain-Leakage Current of Carbon-Nanotube Transistors by Subliming Dispersants with Back-End-of-Line Compatible Temperature

Carbon nanotube field-effect transistors (CNTFETs) are promising for monolithic 3D integration with silicon CMOS circuits in back-end-of-line (BEOL) processes. Further reduction in off-state leakage is essential for practical applications. In this study, we develop a low-temperature cleaning process for CNT-deposited wafers using a sublimable CNT dispersant. The process is carried out below 250 {\deg}C, making it BEOL-compatible, and does not deteriorate CNT morphology or lattice structure, as confirmed by atomic force microscopy and Raman spectroscopy. As a result, gate-induced drain leakage current originating from residual dispersants is reduced by approximately one order of magnitude. In addition, improvements in on-current, subthreshold slope, and threshold voltage variability are observed after the cleaning process. This study demonstrates a material-enabled approach to improve the transfer characteristics of CNTFETs and provides a practical route toward low-power CNT electronics.
Read the full abstract

All authors Takumi Inaba, Yuichi Kato, Yoko Iizumi, Yasuhiko Fujita, Kazufumi Kobashi, Takahiro Morimoto, Toshiya Okazaki

Computation and Language

Same Patient, Different Order: Action-Level Reliability of Clinical LLM Agents Under Repeated Runs

A clinical agent benchmark can report the same verdict on identical inputs while the agent files a materially different order on each run. Such agents order tests, request medications and place referrals, yet benchmarks typically score one run per task and rarely ask whether identical inputs produce identical actions; MedAgentBench, the benchmark we use, scores a single attempt and says so. To measure this gap we introduce "same-input rerun", which replays a task with every input held fixed and compares the orders rather than the score, with six reliability metrics, and apply it to 1000 MedAgentBench runs across 50 tasks from its five write-capable families, two open-weight models below ten billion parameters quantised to four bits, and two temperatures. The study establishes that action-level divergence exists and can pass unrecorded by the score, not that any rate generalises. Under the 8B model at temperature 0.7, all 43 ordering groups emit a different set of orders across five identical runs, 26 emit the order on some runs and not others, and 28 record a different coded value, dose or analyte. In 22 of those 43 the benchmark reports the same failing verdict for materially different behaviour, as it does for all 10 divergent groups of the 4B model at 0.7. Orders also reach different endpoints across runs, one of which the record server rejects while the agent is told it succeeded. These findings motivate repeated-run evaluation, action-level stability reporting and execution-faithful environment feedback in clinical-agent benchmarks.
Read the full abstract

All authors Rohith Reddy Bellibatlu, Manpreet Singh, Zhoutian Han, Wenbin Zhang

Computation and Language

Toward Complete Hospital Discharge Summarization with Abstract Meaning Representation

Discharge summaries are lengthy medical documents that summarize a hospital in-patient visit. Automatically generating them can reduce documentation burden and return clinician time to patient care. Whereas Large Language Model (LLMs) could be used for this task, their Achilles heel is hallucinations, which can have drastic consequences for clinical documentation. We present an evidence-driven alignment framework for discharge summarization at the clinical encounter level, that treats provenance as a first-class constraint, using semantic graphs and deep learning models. Each summary sentence is selected and organized via cross-document semantic alignment and is accompanied by explicit evidence links to its source spans. We show our results on two corpora: a publicly available corpus (MIMIC-III) and clinical notes written by physicians at the University of Illinois Hospital (UIC Health). Additionally, we make source code and trained models available.
Read the full abstract

All authors Paul Landes, Sitara Rao, Aaron Jeremy Chaise, Barbara Di Eugenio

Artificial Intelligence

FLoKD: Adaptive Knowledge Distillation for Federated Low-Rank LLM over Wireless Networks

Large language models (LLMs) have demonstrated strong capabilities across a wide range of natural language processing tasks. However, conventional fine-tuning typically relies on centralized data collection, bringing in privacy concerns. Federated learning (FL) enables collaborative LLM fine-tuning without sharing raw client data, but its deployment over bandwidth-constrained wireless networks is hindered by the communication overhead of model-parameter transmission. Although Low-Rank Adaptation (LoRA) reduces the number of trainable parameters, its communication cost still increases with model scale. Knowledge distillation avoids parameter sharing via output logits, but token-level logits in LLMs incur high communication cost due to sequence length and vocabulary size. Reducing logits lowers the cost but weakens supervision and degrades accuracy. To address these limitations, we propose FLoKD, an adaptive knowledge-distillation framework for federated LoRA fine-tuning of LLMs over wireless networks, which communicates intermediate LoRA activations as the distillation signal rather than logits or full parameters. Since transmitting all blocks over the entire public dataset remains costly, we further propose a transformer block importance scoring framework that selectively transmits the most informative blocks, and two dataset selection strategies that discard public samples deviating from the local data distribution and prioritise those most informative for distillation. Extensive experiments across multiple generative language datasets, including WikiText-103, PTB, and Dialog, demonstrate that our proposed framework reduces communication overhead by 50-65% while achieving rapid convergence to competitive perplexity compared to baselines.
Read the full abstract

All authors Xinlu Zhang, Na Yan, Yang Su, Yansha Deng, Toktam Mahmoodi

Artificial Intelligence

How User-AI Mistreatment Occurs and Matters in Conversational Systems?

Safety research often focuses on model-generated harms, but users may also direct hostility, coercion, and adversarial pressure at models. Understanding how and when that occurs is essential for accurately interpreting model behaviour, alignment drift, and real-world deployment risks. In this paper, we audit 777K English LMSYS-Chat-1M conversations with two independent detectors: an eight-category lexicon for hostility directed at the model, and the dataset's moderation signal; and show that they capture different, weakly overlapping phenomena. The lexicon identifies insults, threats, and jailbreak coercion aimed at the assistant, while moderation flags are dominated by toxic-content solicitation rather than hostility at the model. Together, they mark about 5% of user turns; adjusting the narrower lexicon-harassment union for measured precision puts mistreatment aimed at the assistant at 0.90%. These absolute rates describe arena-style evaluation traffic and should not be read as deployment-wide base rates. We find that user hostility varies 13-fold across models, driven largely by who each model attracts rather than by model behaviour: first-turn hostility spreads far wider than post-response hostility, and more than fifteenfold separates the extremes even after deduplicating opening prompts. Within conversations, assistant apologies are consistently associated with higher odds of next-turn hostility under both detectors; the effect survives restricting to non-refused prior turns and to jailbreak-free conversations, and is positive in 20 of 23 models. Yet across models, more apologetic models receive less hostility overall. Finally, hostility also shows temporal structure, with coercive openings front-loading the first turn while affective hostility accumulates over a session. We release the lexicon, the detector cross-validation pipeline, and all derived tables.
Read the full abstract

Fluid Dynamics

Vertical vortex gust encounters of a delta wing

The present study analyzes encounters of a vertical vortex with a tailless NACA 0012 delta wing with a sweep angle of 60 degrees at an angle of attack of 10 degrees and Reynolds number 1000 using direct numerical simulations. Motivated by the need to fly in complex settings such as urban canyons, mountainous areas, wildfires, or vehicle swarms, we focus on the influence of a vertical vortex gust on the delta wing, complementing previous efforts that predominantly studied spanwise and streamwise vortex encounters. The vertical orientation can impose strong transient loads on all six degrees of freedom, and particular attention is devoted to connecting these loads to the flow response. By varying the gust's lateral position, size, and strength, we find a fundamental process that characterizes vertical vortex gust encounters. The gust consistently imprints a low-pressure core on the wing surface. The evolving core creates an adverse pressure gradient that distorts the baseline skin-friction pattern and induces separation that lifts the boundary layers. These lifted boundary layers are elongated and twisted by the gust's strain into lobed vortical structures, which are shown by force element analysis to make a critical contribution to the lift. As the gust's lateral position is varied from root to tip, the low-pressure core and the induced flow separation are increasingly localized around the leading edge closest to the gust, modifying the lobe evolution and leading to complex dependence of the force and moment response on gust position compared to the largely monotonic dependence on gust size and strength.
Read the full abstract

Software Engineering

Comparative Framework Analysis for Enterprise Generative AI Applications: Chatbot, Automation, and Oracle-to-PostgreSQL Migration

This study compares framework suitability across three classes of enterprise generative AI applications: a documentation-based development assistant, an email and inquiry automation system, and an Oracle-to-PostgreSQL migration tool. The analysis evaluates component boundaries, orchestration, policy retrieval or reasoning, model integration, deterministic validation, persistence, observability, and operational efficiency. Across the three applications, the evidence supports layered architectures in which probabilistic components generate proposals, while deterministic components retain authority over routing, authorization, validation, persistence, idempotency, and final outcomes. The results indicate that framework suitability depends on the application, deployment conditions, and component responsibility: search quality, workflow control, safety behavior, and migration validation cannot be reduced to a single cross-application assessment. Therefore, the study substantiates the need for framework selection at the component level, supported by explicit contracts, application-specific evidence, and independent validation boundaries.
Read the full abstract

All authors Oleg Grynets, Olena Pochernina, Alona Seletska, Daryna Tukalo, Dmytro Kostetskyi, Ivan Fedorchuk, Vasyl Lyashkevych

Number Theory

Galois representations and modularity for twisted moments of Kloosterman sums

Associated with the classical Kloosterman sums, we consider the quadratic twisted symmetric power moments. We investigate the cohomological properties and construct the motives over the field of the rationals underlying such moments. We determine the associated modular forms in the lower degree cases.
Read the full abstract

Instrumentation and Methods for Astrophysics

Sensing ELT petal modes with a discrete beam combiner: an end-to-end theoretical model

The European Southern Observatory's Extremely Large Telescope (ELT), the 39-m telescope under construction in Chile, forms its entrance pupil from 798 hexagonal segments that the six 0.5-m-wide spider arms fragment into six "petals". The low-wind / island effect, adaptive-optics residuals, and mechanical drift imprint an independent piston on each petal. These differential-piston (petal) modes remain largely undetected by the ELT's baseline continuous-pupil wavefront sensors, a major bottleneck for diffraction-limited, high-contrast imaging. This paper presents a complete theoretical model of a single-mode photonic discrete beam combiner (DBC) configured to sense the six petal pistons. The model constructs the coupled-mode description of a two-dimensional evanescently coupled waveguide lattice with six inputs and forty-five outputs and derives its visibility-to-pixel matrix (V2PM); treats the pupil remapping that routes one sub-aperture per petal into the six inputs; develops the five-dimensional petal-mode basis of the fragmented pupil; and propagates the petal modes through the DBC to quantify the sensitivity. Optimising the input-port spacing and interaction length by minimising the V2PM condition number yields a well-conditioned device (CN = 10.4). It recovers petal pistons with unit linear gain over the full +/-lambda/2 (+/-800 nm) unambiguous range and separates the five modes with negligible cross-talk (< 10^-3). For 1% photometric noise, the reconstructed-piston error is 4.7 nm rms at lambda_0 = 1.6 micron. Tested against petal amplitudes from end-to-end ELT simulations (~40-300 nm rms atmospheric residual; ~0.3-1 micron low-wind events), the DBC recovers each piston to a few nm rms at H band, with a two-wavelength extension capturing the larger events. The DBC is therefore a compact, chromatically robust petalometer for the fragmented ELT pupil.
Read the full abstract

All authors Kalaga V. Madhav, Abani Shankar Nayak, Lucas Labadie, Robert R. Thomson