Cryptography and Security
Revisiting Certified Defense with Differential Privacy on Vision Transformers
Jun Yan, Weiquan Huang, Qixian Zhang et al. (+2 more)
Certified defenses that incorporate differential
privacy have proven effective on Convolutional Neural Networks (CNNs), furnishing rigorous robustness guarantees against norm-bounded adversaries. However, the certified robustness behavior of Pixel Differential
Privacy (PixelDP) remains largely unexplored with the self-
attention architecture now dominating the deep-learning landscape. Given that the
Transformer has a profound impact on our daily applications from the digital world to the physical world, it is crucial to study certified robustness through differential-
privacy-style stability. To fill this research gap, we revisit this construction in
Vision Transformers and identify a failure mode that is largely hidden in the convolutional setting. When noise is injected after the patch
embedding, the Laplace mechanism with the inherited grouped $\ell_1$ sensitivity bound collapses to chance-level accuracy across noise scales, whereas the Gaussian mechanism remains trainable. This contrast isolates the source of failure: not the injected noise itself, but the geometry of the sensitivity constraint. We show that the attenuation induced by the inherited $\Delta_{1,1}$ projection increases with layer width and kernel size according to a random-
matrix scale $C/(\sqrt{M}+\sqrt{N})$. Replacing the $\ell_1$-type constraint with a spectral-norm constraint eliminates the collapse across
datasets and architectures, but creates a fundamental obstacle: the repaired models no longer satisfy the sensitivity condition required by the standard Laplace certificate. We resolve this mismatch by deriving a dimension-free $(\varepsilon,\ \delta)$-
privacy guarantee for the Laplace mechanism under $\ell_2$ sensitivity through concentration of the
privacy loss.
Read the full abstract ▾
All authors Jun Yan, Weiquan Huang, Qixian Zhang, Yan Bai, Shutai Zhang
High Energy Physics (Theory)
Fate of Bouncing Singularities at Finite $G_N$
Zohar Komargodski
The
gravity approximation to a thermal two-point function can develop a complex-time singularity associated with a trajectory that reaches the black-hole singularity. We show that the corresponding singularity is absent at finite $G_N$. This is done by extending Conformal Field Theory analyticity properties to the second sheet (and to the universal cover). We also put two different upper bounds on the amplitude of the correlator at the would-be singularity, by certain finite partial sums, and by a certain partition function on a
manifold with nontrivial
topology, where two thermal geometries are glued. We discuss the possible implications for the black hole singularity.
Read the full abstract ▾
Astrophysics of Galaxies
SQuIGG$\vec{L}$E: Ubiquitous Cool Gas Outflows with Large Mass-Loading Factors in $z\sim0.7$ Post-starburst Galaxies
Yuanze Luo, Justin S. Spilker, David J. Setton et al. (+13 more)
While recent James Webb Space
Telescope studies have highlighted cool gas
outflows as an important mechanism for rapid quenching at $z\gtrsim2$, such space-based spectroscopic samples remain limited in size. In this work, we search for and characterize cool gas
outflows traced by Fe II $\lambda\lambda$2586,2600 and Mg II $\lambda\lambda$2796,2803 absorption in post-starburst
galaxies (PSBs) at $0.5<z<0.9$, which are accessible to ground-based spectroscopy. Utilizing the SQuIGG$\vec{L}$E survey containing over 1300 PSBs, we stack spectra in bins of
stellar mass, (specific)
star formation rate, burst mass, burst mass fraction, and time since quenching. We observe blueshifted line centroids indicative of
outflows in all stacks except the one containing
galaxies that quenched more than 260 Myr ago.
Outflow velocities decline with increasing time since quenching, while showing little correlation with other
galaxy properties. The inferred
outflows have large mass loading factors of 1 – 120, maximum velocities above the interstellar medium escape velocities of their host
galaxies, and energy and momentum fluxes exceeding those expected from the current
star formation. Our results suggest that cool gas
outflows are ubiquitous in intermediate-
redshift PSBs and are likely relic
outflows launched by previous starburst or AGN activity. These
outflows have the potential to quench
star formation by transporting gas into the
galactic halos, consistent with
outflows playing an important role in rapid quenching from intermediate to high
redshift.
Read the full abstract ▾
All authors Yuanze Luo, Justin S. Spilker, David J. Setton, Timothy Heckman, Rachel Bezanson, Vincenzo R. D'Onofrio, Alex Geiger, Andy D. Goulding, Jenny E. Greene, Kate Rowlands, Katherine A. Suess, Elizabeth Taylor, Margaret E. Verrico, Andrew Weldon, Vivienne Wild, Pengpei Zhu
Astrophysics of Galaxies
SQuIGG$\vec{L}$E: Spatially resolved NIRSpec Pa$\alpha$ reveals that the most CO-luminous, $z\sim0.7$ post-starburst galaxies exhibit falling—but still active—star formation behind optically thick dust
David J. Setton, Jenny E. Greene, Rachel Bezanson et al. (+13 more)
One of the fundamental uncertainties in the study of
galaxies at the interface between
star formation and quiescence–post-starburst
galaxies–is their instantaneous
star formation rate. While these
galaxies appear quiescent in the rest UV-optical, many of the youngest systems are highly CO luminous, implying dense molecular gas, and potentially
star formation. However, their instantaneous
star formation rates have proved elusive, as measurements with rest-optical emission lines or UV-to-IR SED fitting yield order-of-magnitude uncertainty. Here, we present JWST/NIRSpec G395M/170LP integral field spectroscopy of the three most CO-luminous massive
galaxies in the SQuIGG$\vec{L}$E survey, spatially resolving Pa$\alpha$/Br$\gamma$/Br$\beta$ to map the
dust-corrected
star formation rate. We find that these
galaxies host
star formation rates of $27-80$ $M_\odot \ \mathrm{yr^{-1}}$, though $\sim30-40\%$ of the emission may arise from non-
star-forming ionization (AGN or shocks), implied by high [FeII]/Pa$\alpha$ in central spaxels. We also find effective $A_{\mathrm{Pa}\alpha}\sim1$ and individual $\sim700$ pc spaxels with $A_{\mathrm{Pa\alpha}}\sim2$, implying significant
dust obscuration of nebular regions that contribute little to the rest-optical lines. These
galaxies are not consistent with total quiescence; the measured
star formation rates place these
galaxies on or above the main sequence. However, their
star formation rates are significantly lower ($\sim0.5-1$ dex) than the peak
star formation rates inferred from spectrophotometric fitting. We propose that the CO luminous post-starburst phase represents the last gasp of
star formation in
galaxies on the rapid quenching pathway, and that
dust-obscured
star formation,
outflows, and (speculatively) an uncertain CO-to-H2 conversion reconcile their gas exhaustion with observed timescales of fading.
Read the full abstract ▾
All authors David J. Setton, Jenny E. Greene, Rachel Bezanson, Katherine A. Suess, Justin S. Spilker, Vincenzo R. D'Onofrio, Elia Cenci, Robert Feldmann, Alex J. Geiger, Andy D. Goulding, Mariska Kriek, Anika Kumar, Yuanze Luo, Desika Narayanan, Margaret E. Verrico, Pengpei Zhu
Machine Learning
Benchmarking Attention for Tabular Foundation Models
Maximilian Schambach, Clemens Biehl and Sam Thelin
Tabular in-context learners such as TabPFN, Mitra, or ConTextTab rely on alternating row and column
attention over 2D sequences of latent
embeddings. These
attention patterns differ markedly from the one-dimensional case in
language models: row
attention involves longer sequences while column
attention operates on much shorter ones, and the strided
memory layout of tabular data makes producing contiguous tensors costly. Moreover, the hidden dimensions used in current models are small compared to recent
language models. Yet efficient
attention has been studied mostly for one-dimensional sequences, leaving the two-dimensional tabular setting unexplored. To this end, we create a reproducible benchmarking setup and study the unique characteristics of tabular
attention across several backends – Torch SDPA (efficient and cuDNN), FlashAttention-2/3/4, and the inference-only backends vLLM and SageAttention – measuring forward and backward throughput across realistic tabular shapes on three GPU generations (A100, H100, B200). We find that the optimal backend choice differs between column and row
attention and varies across hardware as well as model specifics: While the FlashAttention implementations tailored for each GPU generation perform overall best, they are at times outperformed by CuDNN in the case of column
attention at longer sequences with cross-over points depending on the head dimension. Among inference-only backends, SageAttention performs well for row
attention and large sequences beyond 16\,k rows. Our reproducible
benchmark lays the foundation for future improvements to table-native
attention. The self-contained benchmarking and evaluation code is openly available at: https://github.com/SAP-samples/tabular-
attention-
benchmark
Read the full abstract ▾
Astrophysics of Galaxies
A solution to the paradox of youth in the Galactic centre
Rodrigo P. Silva, Alexandre C. M. Correia, Tjarda C. N. Boekholt et al. (+1 more)
The supermassive black hole at the
Galactic centre, Sgr A*, is surrounded by an S cluster of young
stars. These
stars are predominantly B-type
stars, but their origin remains uncertain. Unlike massive
stars in the
Galactic field, which exhibit a binary fraction of about $70\%$, those in the S cluster show a significantly lower fraction. It is key to understand this discrepancy for constraining their formation. We show that the binary deficit arises naturally from dynamical interactions between
stellar binaries and Sgr A* in an in situ formation scenario. Using K-amplitude data, we derived an observational binary fraction of $f_{\mathrm{obs}} = 0.43 \pm 0.09$ for the S-cluster
stars. Our dynamical modelling predicts that interactions with Sgr A* cause $18\%$ of the binaries to merge, $20\%$ to be disrupted, and $62\%$ to survive, yielding a theoretical binary fraction of $f_{\mathrm{theo}} = 0.38 \pm 0.10$. The predicted binary fraction with distance from Sgr A* is also consistent with independent observational constraints. These results show that the decline in binarity towards the
Galactic centre is driven by stronger
tidal forces, and they support an in situ origin of the S cluster in the same event that produced the
Fermi bubbles.
Read the full abstract ▾
All authors Rodrigo P. Silva, Alexandre C. M. Correia, Tjarda C. N. Boekholt, Paulo J. V. Garcia
Differential Geometry
Topology and Dirichlet spectrum of free boundary minimal and CMC surfaces
Alcides de Carvalho and Roney Santos
We find bounds for the genus and for the number of boundary components of a surface that is either index one as a free boundary minimal surface or stable as a free boundary constant mean
curvature surface in a Riemannian three-
manifold with convex boundary and non-negative
Ricci curvature, provided the first Dirichlet
eigenvalue of its Jacobi operator is non-negative. As an application, we establish strong topological restrictions on such surfaces in compact strictly convex domains of three-dimensional normal homogeneous spaces.
Read the full abstract ▾
Machine Learning (Statistics)
Geometric Moment Contraction for Stochastic Nesterov Acceleration
Wei Biao Wu
We study
geometric moment contraction (GMC) of the constant-parameter
stochastic Nesterov recursion \[ Y_k=\Theta_k+\beta(\Theta_k-\Theta_{k-1}),\qquad \Theta_{k+1}=Y_k-\gamma G(Y_k,X_{k+1}). \] Under mean strong monotonicity and
stochastic $L^p$ Lipschitz continuity, an explicit Perron comparison proves synchronous $L^p$ contraction when $\beta\gamma L_p<(1-\beta)(1-q_{\gamma,p})$. This direct criterion includes infinite-variance gradients for $1<p<2$, but its small-step regime requires $\beta<\mu/(\mu+L_p)$. A complementary power-Lyapunov argument establishes a positive, generally much smaller, step-size interval for every fixed $\beta<1$ and every $p>1$, using only a finite $p$th gradient moment. At $p=2$, a simpler explicit certificate gives \[ 0<\gamma<\frac{2\mu(1-\beta)^2}{L_2^2(1-\beta+2\beta^2)}. \] Its quadratic high-momentum scaling is a limitation of the chosen metric, not a sharp stability boundary. We quantify this loss, provide a general mean-only quadratic $S$-procedure, and exploit endpoint Lyapunov inequalities under stronger samplewise sector information. Verified endpoint certificates can be orders of magnitude less conservative than the explicit metric.
Read the full abstract ▾
Distributed Computing
KCensus: Synthesizing Latency-Optimal Consensus Fast Paths (Extended Version)
Clément Burgelin, Antoine Murat, Gal Sela et al. (+2 more)
Strongly consistent geo-replication often relies on fast paths to reduce latency in the common case of no failures or contention. Existing fast-path schemes, however, are ad hoc and restrictive: each corresponds to a point in a broad design space shaped by network
topology, workload, and latency objective, so no single scheme works best across settings. This paper looks at fast-path schemes from a new perspective, as mechanisms that spread knowledge about proposals. With this view, we identify a fundamental condition on the spread of knowledge for a fast-path scheme to work. We then introduce KCensus, a framework that turns this condition into an optimization problem, synthesizing new fast-path schemes that are optimal for a given setting. We use KCensus to build a geo-replicated key-value store and evaluate it across AWS regions. Our system outperforms competing protocols, with up to 16% lower average latency.
Read the full abstract ▾
All authors Clément Burgelin, Antoine Murat, Gal Sela, Marcos K. Aguilera, Rachid Guerraoui
Software Engineering
Beyond Approved Actions: Runtime Validation of Persistent Outcomes in Agent Workflows
Haoran Zhang, Hengtong Zhang, Zhiyu Liang et al. (+3 more)
Large
language model
agents increasingly act on software systems, no longer merely generating text but also changing databases and online services. However, an approved database update may succeed yet leave an unapproved notification because execution can produce persistent effects beyond the requested change. Current safeguards can approve an action or record its aftermath, but without checking the persistent result before continuation, an unapproved outcome can be accepted as success and propagated to later steps. We present EffectMatch, a runtime that collects persistent changes within a controlled execution boundary and compares them with what the application approved for the current state and execution. The comparison governs commit and dependent execution. In comparative evaluation on 206 public business tasks, EffectMatch preserved all clean executions and prevented all tested incorrect commits. Six 20-run ablations exposed the failure caused by each removed mechanism, while 80 task-
topology cases preserved truthful handoffs and blocked invalid continuation. Together, these results show that EffectMatch blocks the silent acceptance and downstream propagation of persistent outcomes inconsistent with application approval.
Read the full abstract ▾
All authors Haoran Zhang, Hengtong Zhang, Zhiyu Liang, Yu Yan, Decheng Zuo, Hongzhi Wang
Information Theory
New distance bounds for one-generator quasi-cyclic codes with applications to Hermitian LCD codes and entanglement-assisted quantum codes
Kanat Abdukhalikov, Duy Ho and Rasha M. Shat
We introduce lower bounds on the minimum distance of one-generator quasi-cyclic codes of arbitrary index, obtained by puncturing and shortening the code at the level of its blocks. For balanced codes, these bounds form a nondecreasing chain whose first term is the Jensen bound. We apply these bounds to conduct a computer search and obtain many Hermitian LCD codes with good parameters. Among them there are six codes of length at most 30 which improve the corresponding lower bounds of Araya and Harada. These codes further give maximal-
entanglement entanglement-assisted
quantum codes, which we compare with the parameters currently recorded in the literature.
Read the full abstract ▾
Information Theory
A Mathematical Theory of Near-Field Super-Resolution
Sajad Daei, Gabor Fodor and Mikael Skoglund
Finite-aperture near-field sensing leads to a super-resolution geometry fundamentally different from the translation-invariant
Fourier setting. In the Fresnel regime, wavefront
curvature introduces a range-dependent quadratic aperture phase, so distinguishability is governed by incomplete quadratic exponential sums of the form \( \sum_{n=0}^{N_r-1} a_n e^{i(\omega_1 n+\omega_2 n^2)} \), rather than by angular separation alone. We develop a deterministic recovery theory for sparse measures with ranges on a finite grid and continuous angles, and introduce a support-uniform quadratic-phase aperture criterion replacing classical minimum separation. Under this criterion, total-variation minimization exactly recovers every sparse measure in the admissible support class, uniformly over all nonzero complex amplitudes. The proof develops nonasymptotic support-uniform bounds for finite quadratic sums and combines them with a gauged Hermite dual certificate controlling interpolation, local
curvature, and off-support leakage. We further construct a finite-harmonic Bessel-Vandermonde lift with explicit truncation error. In the far-field limit, the quadratic phase disappears and the theory reduces to
Fourier-type angular super-resolution.
Read the full abstract ▾
Computer Vision
UniAR: A Unified Framework for Autism Recognition Enhanced by Multi-View Prompt Learning
Lei Xin, Zeheng Wang, Jiayin Zhu et al. (+6 more)
Autism Spectrum Disorder (ASD) is a complex neurodevelopmental disorder for which early and accurate diagnosis is critical to improving long-term developmental outcomes. However, existing ASD recognition methods are often constrained by the scarcity of diagnostic text data, forcing them to rely mainly on visual analysis and limiting their ability to model clinically meaningful semantic
reasoning. To address this challenge, we propose UniAR, a unified framework enhanced by multi-granularity
prompt learning for robust ASD recognition under heterogeneous data variations. Specifically, UniAR leverages a large
multimodal model to generate hierarchical diagnostic descriptions at the word, phrase, and sentence levels, compensating for the lack of paired clinical reports. To align the generated semantics with visual evidence, we further design a Mixture-of-Experts-based Multi-Scale
Alignment Module, which dynamically matches vector-quantized visual prototypes with semantic representations at corresponding granularities. Extensive experiments on four
benchmarks covering
brain MRI and facial expression scenarios show that UniAR consistently outperforms existing state-of-the-art methods, achieving average accuracies of 75.9% on MRI
benchmarks and 91.6% on facial
benchmarks, while improving average Accuracy on MRI
benchmarks by 1.5 percentage points and average Accuracy on facial
benchmarks by 1.2 percentage points over baselines. These results demonstrate that UniAR offers a robust and interpretable framework for ASD screening under semantic scarcity.
Read the full abstract ▾
All authors Lei Xin, Zeheng Wang, Jiayin Zhu, Shihong Huang, Fanhu Zeng, Changjiang Jiang, Dengbo He, Yutao Yue, Zhenglun Kong
Social and Information Networks
Who Belongs Together? Topical and Social Structure in Bluesky Starter Packs
Sima Adleyba and Onur Varol
Bluesky starter packs are human curated collections of accounts and feeds aimed at helping users discover new communities, especially during onboarding. Prior research has examined their effects on platform growth and account visibility. However, whether these packs capture meaningful topical and social structure remains unanswered. We study 2718 active starter packs containing approximately 144K accounts, combining pack metadata, follow graphs, user post histories and inferred demographic attributes to answer this question. We assign starter packs to 17 topical categories, and examine semantic coherence, curator–member resemblance, shared audiences, and follow reciprocity. We found that pack members are closely aligned with both their fellow pack members and their pack's description compared to accounts included in other packs of the same topic, indicating topical specificity beyond the broader topical categories. We also identify semantic resemblance between curators and their pack members, while demographic resemblance varies by attribute and topic. Network analyses show higher follow reciprocity among members of the same pack and between members and curators than between a pack member and either another member from the same broader topic or a nonmember. Members also share substantial portions of their follower networks with fellow pack members. Despite this coherence, a typical member's follower follows only a small fraction of other members in the pack, indicating limited audience coverage. These results suggest that starter packs capture coherent topical and social groups while leaving opportunities for further account discovery. More broadly, results show human-curation can be useful to capture relevant communities and connections, and platforms could integrate human-curation into recommendation systems as a complement to algorithmic systems to expand opportunities for discovery.
Read the full abstract ▾
High Energy Physics (Phenomenology)
Improving predictions for the $pp \to t\bar{t}W^+$ process at the LHC with the MINLO method
Nikolaos Dimitrakopoulos
We compare the standard NLO and $\texttt{MiNLO}$ approaches for the full off-shell $pp \to t\bar{t}W^+ + X$ process at the LHC, at both the integrated and differential cross-section levels. Contrary to NLO, in the $\texttt{MiNLO}$ approach, which is now also available within the HELAC-NLO framework, renormalization and factorization scales are dynamically extracted and Sudakov form factors are incorporated. Results are particularly presented for the $pp \to t\bar{t}W^+j$ process at NLO accuracy in perturbative QCD using two different scale choices. Merged predictions up to two jets to improve the overall modeling of the full off-shell $pp \to t\bar{t}W^+ + X$ process are also discussed.
Read the full abstract ▾
Combinatorics
On harmonic centers of graphs
Tomáš Madaras and Matúš Paralič
The harmonic centrality of a vertex $v$ in a graph $G = (V,E)$ is the sum of reciprocals of distances of vertices of $G$ from $v$. The vertices of $G$ which have the maximum (minimum) harmonic centrality form the harmonic center (or periphery, resp.) of $G$. We study harmonic centers of graphs and their localization in graph blocks, presenting sufficient conditions for graphs (in terms of diameter or number of edges) to have those centers contained in a single block; in addition, we show that each connected graph is the harmonic center as well as harmonic periphery of some graphs.
Read the full abstract ▾
Analysis of PDEs
On Shnirelman's inequality in 2D: Irreversible behavior and Euler-Lagrange variational formulation
Stefan Schiffer and Martina Zizza
In this paper we prove a version of Shnirelman's inequality in the two-dimensional square $[0,1]^2$. More precisely, given a target fluid configuration $f$ (a volume-preserving diffeomorphism) we construct a divergence-free velocity field $v\in L^q_tL^p_x$ whose flow connects $f$ to the identity and whose $L^q_{t} L^p_x$-norm can effectively be bounded by the $L^p$-difference of $f$ and $id$. However, higher regularity of the vector field might be affected. The main difference with the higher dimensional case is that, because of the topological obstructions of dimension $\nu=2$, we have to allow that different fluid trajectories 'intersect' in space. We also observe that this choice leads to the emergence of irreversible behaviors in the Eulerian dynamics.
Read the full abstract ▾
Audio and Speech Processing
Why Alzheimer's Speech Screening Fails to Generalize: Bridging the Deployment Gap via Cross-Corpus Evidence Anchoring
Zijian Lu, Sizhe Liu, Yin Zhang et al. (+6 more)
Speech-based screening is a promising, non-invasive approach for detecting Alzheimer's disease and related cognitive risks. However, models trained on a single domain often generalize poorly to unseen
languages, tasks, or recording protocols. This paper investigates this deployment gap using a leave-one-corpus-out evaluation across four distinct
datasets. Among 70 interpretable
speech and
language features, 59 exhibit direction conflicts between healthy control and cognitive risk groups across corpora, with pause, silence, and
speech rate showing high protocol sensitivity. Furthermore, while the XLM-R text baseline achieves strong average performance, its Area Under the ROC Curve (AUC) drops to 0.520 on the weakest held-out domain. A standard GroupDRO baseline reaches a 0.766 mean speaker AUC and a 0.504 worst-domain AUC under the same protocol. To address this, we propose a fusion method that integrates XLM-R text baseline scores with evidence anchors selected during training. Balanced fusion achieves a 0.785 mean speaker AUC, while anchor-heavy fusion raises the worst-case speaker AUC to 0.615. This work highlights the need to audit feature transferability and report worst-case domain robustness in cognitive
speech screening.
Read the full abstract ▾
All authors Zijian Lu, Sizhe Liu, Yin Zhang, Jixuan Deng, Xinrong Lin, Xinchen Yuan, Chicheng Jin, Yiping Zuo, Yuanchao Li
Networking
Dynamic Sampling for Telemetry in Microservices: A Reinforcement Learning and Entropy-Based Approach
Renan Martins Alves, Jéferson Campos Nobre and Juliano Araujo Wickboldt
Microservices architectures are increasingly deployed in cloud-based distributed environments, making application development and maintenance more dynamic, but also increasing the complexity of troubleshooting and observability. Distributed tracing tools are therefore essential for request analysis and debugging, despite introducing additional overhead that can be amplified by excessive and inefficient data collection. This article proposes RADAR (Reinforcement Learning
Agent for Dynamic And Relevant trace sampling), an
agent that combines reinforcement learning with a data
entropy assessment to achieve more efficient capture of traces relevant to system monitoring, based on the OpenTelemetry standard. RADAR tests different sampling rules to discover which combination is most efficient. A test environment simulating a minimalist online store with several microservices distributed across a Kubernetes cluster served as the basis for the experiments, which evaluated the
agent's convergence and the system's performance in terms of resource consumption and collected data quality. Results showed that RADAR reduced network bandwidth consumption by 97.4% and CPU usage by 99.0% compared to full data collection, also outperforming a fixed-rate sampling baseline. Beyond these resource savings, the approach preserved observability of critical scenarios, retaining approximately 85.6% of rare trace patterns and increasing the average
entropy of the stored information by approximately 25%, validating the feasibility of using
entropy to orchestrate telemetry autonomously and efficiently.
Read the full abstract ▾
Machine Learning
Softmax Reparameterization for Output-Head Quantization
Asim Kadav, Christian Flores, Chirag Arora et al. (+5 more)
Large vocabularies make output heads a substantial inference cost in small
language models. We introduce softmax reparameterization, a post-training method that searches over functionally equivalent output heads before quantization. The method subtracts a scalar multiple of the vocabulary-row mean from every output row and selects the coefficient by validation KL. For linear-softmax heads, these shifts preserve full-precision predictions exactly and require no decoder retraining; a rank-one correction extends the construction to nonlinear logit paths. Across seven output heads and three quantizers, W4 gains are largest where baseline quantization substantially distorts predictions: test KL falls by 93% on XGLM under RTN and by 73–77% on Phi, BLOOM, and BLOOMZ under activation-weighted MSE. Heads with low baseline error change little; at W2, used as a compression stress test, benefits extend more broadly. On Phi, the gains persist under stronger GPTQ calibration; a separate untouched holdout reproduces the improvements on Phi and BLOOM. Frozen WikiText-selected coefficients also transfer without retuning to C4 and OpenWebMath. Residual analysis on Phi shows how fidelity can improve despite greater total logit error: the selected representative reduces error on likely outputs and lowers its Fisher-weighted cost. For shift-compatible heads, the shift adds no inference operation. With the decoder held in BF16, a packed W4 Phi output head reduces batch-one generation latency by 10.8%, and reparameterization preserves this speedup.
Read the full abstract ▾
All authors Asim Kadav, Christian Flores, Chirag Arora, Varun Kotte, Hongbo Zheng, Lan Yan, Priya Shanmugasundaram, Tracy Holloway King
Data Structures and Algorithms
An Optimal Structure for All-Pairs Nearest Mincuts and Sensitivity Oracles for Edge Insertions
Koustav Bhanja, Yotam Kenneth-Mordoch and Asaf Petruschka
Given an undirected weighted graph $G=(V,E)$ on $n$ vertices, the classical Gomory-Hu tree of $G$ is a structure that encodes an arbitrary minimum $s,t$-cut for every $s,t\in V$ using just $O(n)$ space. In this work, we ask whether the same compactness is achievable for the natural and structured family of all-pairs nearest minimum cuts. The nearest minimum $(s,t)$-cut is the unique inclusion-wise minimal one among all minimum $s,t$-cuts containing $s$. This family has proven useful in a wide range of applications, including fault-tolerant reachability, minimum cut sensitivity oracles, cactus representations, and fast Gomory-Hu tree constructions. Despite its fundamental role, no subquadratic space representation is known for them to date. The $O(n)$ space representations are known only in single-source settings, where given a source $s$, one can report the nearest minimum $(t,s)$-cut for any $t\in V\setminus \{s\}$. We close this gap by presenting the first optimal space representation of all-pairs nearest minimum cuts, providing a natural analogue of the Gomory-Hu tree. Our main result is an $O(n)$ space structure that encodes the nearest minimum cut between every pair of vertices. Furthermore, given any pair $s,t\in V$, it can report the nearest minimum $s,t$-cut in $O(n)$ time. Both bounds match those of the Gomory-Hu tree and are worst-case optimal. As an application, we design an all-pairs minimum cut sensitivity oracle for edge insertion: a data structure that occupies $O(n)$ space and, given any edge $e$, can determine for all pairs $s,t\in V$ whether the minimum $s,t$-cut value increases upon insertion of $e$ in $O(n^2)$ total time. Existing insertion sensitivity oracles were either limited to the single-source setting or used $O(n^2)$ space for all-pairs [Baswana, Gupta, and Knollmann, Algorithmica'22; Baswana and Pandey, SODA'22].
Read the full abstract ▾
Group Theory
The word problem in verbal products of groups
Javier Brude and Román Sasyk
We prove that nilpotent and metabelian products of groups with decidable word problem have decidable word problem. In contrast, we exhibit families of groups with decidable word problem for which their solvable product have undecidable word problem. In doing so, we exhibit
hyperbolic groups with undecidable membership problem in its derived subgroups.
Read the full abstract ▾
Software Engineering
When the Model Retires: An Empirical Study of LLM Migration in Open-Source Applications
Hyungjin Lukas Kim
Applications built on commercial large
language model (LLM) APIs depend on model versions that providers retire on their own schedule, with notice periods ranging from one year to two weeks. We ask what actually happens to applications when a model is retired. We mine GitHub for commits that migrate away from officially deprecated models and endpoints of OpenAI, Anthropic, and Google, matching each commit to the provider's published announcement and shutdown dates. From 22,555 commits in 17,703 non-fork repositories (2024-2026), 5,139 are matched to an official event; two independent coders validated a stratified sample of 300 (kappa = 0.89-0.95), and we reweight all estimates by their labels. We find that an estimated 82% (95% CI 79-84) of migrations away from retired models were committed after the shutdown date - after the application had started failing - regardless of repository popularity, prior retirement experience, or the presence of a provider-abstraction layer. The share tracks the provider's notice policy: 89% for Anthropic's 60-114-day notices versus 13% for OpenAI's one-year Assistants API notice, and each e-fold increase in notice length reduces the odds of post-shutdown migration by about three quarters. Model identifiers are hard-coded in 94% of migrating applications, migration effort scales from a median of 6 added lines for
prompt-only applications to nearly 700 for fine-tuned ones, and only 8% of migrations switch provider. We release the
dataset and pipeline and discuss implications for deprecation policy, dependency-risk assessment of LLM products, and tooling.
Read the full abstract ▾
Astrophysics of Galaxies
H2 emission in the Orion Bar, NGC7023 and the Horsehead Nebula observed with the James Webb Space Telescope
M. Zannese, A. Sidhu, A. Guerras et al. (+25 more)
Molecular hydrogen (H2), the most abundant molecule in the interstellar medium, is a handy tool to study Photodissociation Regions. The JWST, with its high spatial resolution, sensitivity, and wavelength coverage, provides unique access to the detection and spatial morphology of the H2 rotational and rovibrational lines. Our goal is to analyze H2 line emission detected with JWST in several PDRs (the Orion Bar, NGC7023, and the Horsehead Nebula) to constrain the physical structure of dissociation front, and compare the impact of the ultraviolet field and the gas density across different regions. The analysis of H2 spatial morphology reveals similar filamentary structures across all regions. Spatial shifts between H2 lines (0.5") are observed and linked to differences in excitation mechanisms (thermalized vs FUV-pumped lines) and the temperature gradient. Despite differences in incident irradiation conditions, analysis of H2 rotational excitation yields high gas temperatures that are similar across the three PDRs. This is expected for a certain regime of excited PDRs (such as Orion and NGC7023), where the G0 at the DFs and the density are similar. However, it is surprising for less excited PDRs like the Horsehead Nebula. OPRs in the rotational and rovibrational levels differ across the three PDRs. We also used H2 rovibrational lines and the 0-0 S(3) line to estimate the visual extinction across the field of view, yielding markedly different results that constrain both the geometry and the grain composition. Due to the complex 3D geometry, H2 emission is observed throughout the FOV and is dominated by the thin UV illuminated layer, making H2 an unreliable tracer of gas temperature in deeper and coolers molecular layers of the PDR. Therefore, H2 emission is not sufficient to trace the possible temperature and density gradient across PDRs and must therefore be associated with other tracers.
Read the full abstract ▾
All authors M. Zannese, A. Sidhu, A. Guerras, M. Pound, M. Wolfire, A. G. G. M. Tielens, E. Peeters, E. Habart, A. Abergel, M. Baes, O. Berné, C. Boersma, E. Bron, J. Cami, R. Chown, E. Dartois, P. Dell'Ova, J. R. Goicoechea, K. D. Gordon, P. Guillard, O. Kannavou, K. Misselt, N. Monnier, A. Noriega-Crespo, T. Onaka, D. Van de Putte, L. Verstraete, A. N. Witt
Artificial Intelligence
G2MAF: Test-Time Gradient Guidance for Multi-Agent Flow Policies
Guowei Zou, Haitao Wang, Guoxin Wang et al. (+4 more)
Offline multi-
agent reinforcement learning (MARL) learns cooperative policies from fixed
datasets without further environment interaction and a learned policy is frozen at deployment. Such a frozen policy typically proposes a single joint action and executes it directly at deployment time. However, this one-shot deployment often commits to a suboptimal proposal, even when better nearby alternatives remain consistent with the behavior data. To address this issue, we propose Gradient Guided Multi
Agent Flow (G2MAF), a refinement framework for optimizing joint policies at test-time. G2MAF applies one globally normalized, projected critic gradient to guide and coordinate all
agents' corrections while keeping the action both feasible and close to the frozen policy proposal. Across 24 MPE and SMAC settings, its canonical variant improves 20 frozen settings, with mean relative gains of 9.2% on MPE and 8.9% on SMAC, with model inference latency increased by about 6% only.
Read the full abstract ▾
All authors Guowei Zou, Haitao Wang, Guoxin Wang, Zhiquan Chen, Beiwen Zhang, Guojie Wang, Hejun Wu
Computer Vision
MoTop: Motion-Topological Model For Micro AU Detection
Huai-Qian Khor, Mengting Wei, Yante Li et al. (+2 more)
Facial micro-expressions are spontaneous, brief, and subtle facial movements that reveal suppressed
emotions in high-stakes environments. In contrast to classic expression analysis, detecting action unit (AU) yields a finer representation of facial movements, serving as a preliminary step before defining expression classes and other downstream tasks. Therefore, it represents a crucial upstream task in facial analysis, and improving an AU detection module increases the precision of facial analysis. Despite that, detecting AU is challenging because of the constrictive nature of the AU activation regions, leading to confusion among different AUs known as AU ambiguity. To model the fine-scale changes, we propose MoTop, a motion-topological model that is augmented with a learnable motion context, yielding regional soft guidance for facial activity, followed by facial landmarks that capture the fine-scale topological changes of micro AUs. To increase the micro facial landmark representations, we amplify the encoded facial landmark transitions via linear extrapolation, thereby increasing the spatial proximity of landmarks and enhancing the low-intensity landmark dynamics. In addition, we design anatomical facial clusters that enhance the hierarchical representation, facilitating multi-scale modelling of facial geometry and improving micro-topological representations. With these contributions, we have achieved state-of-the-art performance on the CD6ME protocol for the micro AU detection task.
Read the full abstract ▾
All authors Huai-Qian Khor, Mengting Wei, Yante Li, Chu Kiong Loo, Guoying Zhao
Statistical Methodology
Rate-Preserving Shrinking-Support Gaussian Process Prediction
Xiaopeng Xiang, Wenlin Dai, Marc G. Genton et al. (+1 more)
Gaussian process prediction is a central tool in spatial statistics, but standard implementations require dense
matrix operations that become prohibitive for large
datasets. We propose a scale-adjusted compactly supported working correlation for Gaussian process prediction, using the generalized Wendland family with a support radius $\phi_n$ that is allowed to decrease with the sample size $n$. Under fixed-domain asymptotics with quasi-uniform designs, a fixed support radius does not yield asymptotic sparsity, whereas a shrinking support radius can make the covariance
matrix sparse. We show that $\phi_n$ and the regularization parameter can be jointly chosen so that the resulting predictor preserves the optimal integrated mean squared prediction error rate while reducing the number of nonzero covariance entries and the associated sparse
matrix-vector multiplication cost. We also establish an analogous rate-preserving sparsification result for kernel ridge regression. Simulations and an ERA5 temperature application show that the proposed method achieves competitive prediction accuracy while retaining the computational efficiency provided by sparse linear
algebra.
Read the full abstract ▾
All authors Xiaopeng Xiang, Wenlin Dai, Marc G. Genton, Wenjia Wang
Instrumentation and Methods for Astrophysics
MATISSE-py: a high-level Python interface for efficient, automated VLTI/MATISSE data reduction
A. Soulain, A. Matter, P. Berio et al. (+4 more)
The Multi AperTure mid-Infrared SpectroScopic Experiment (MATISSE) at the Very Large
Telescope Interferometer (VLTI) delivers rich mid-infrared interferometric
datasets that are reduced through a robust, multi-stage ESO pipeline. In its native form, however, the reduction relies on command-line calls to esorex, hand-edited set-of-files (SOF) and configuration files, and manual iterations that are difficult to automate, reproduce, or integrate into modern Python workflows. We present MATISSE-py, a modular Python interface that wraps the official ESO esorex recipes behind a single, user-friendly matisse command. It automates discovery, reduction, calibration, and inspection of the data, and ships as a tested, open-source package built on industry-standard practices. Beyond convenience, MATISSE-py introduces science-grade improvements developed over seven years of MATISSE operations: visibility- and photometry-factor (V-factor and P-factor) corrections derived from the
GRAVITY fringe tracker to mitigate residual visibility losses, per-baseline data filtering, beam-commuting-device (BCD) "magic-number" corrections, and an absolute spectrophotometric flux calibration using state-of-the-art reference spectra. On faint targets observed under degraded conditions, the V-factor correction recovers up to a few percent of lost visibility and reduces the statistical uncertainties by nearly 20%, while yielding calibrator diameters that are consistent across BCD modes. MATISSE-py thus consolidates the consortium's expertise into a maintainable package and shortens the reduction-to-science loop for the community.
Read the full abstract ▾
All authors A. Soulain, A. Matter, P. Berio, F. Millour, B. Lopez, R. Petrov, S. Lagarde
Cryptography and Security
Resource-Optimized and Energy-Aware Agentic AI Framework Anchored on Blockchain for Secure Software Supply Chains
Toqeer Ali Syed and Asadullah Abdullah Khan
This paper proposes a blockchain-backed agentic security framework designed to safeguard the complete software development lifecycle (SDLC) while also securing the agentic AI components responsible for monitoring it. The framework coordinates a set of specialised security
agents, covering source integrity, dependency and SBOM analysis, CI configura tion
auditing, artifact verification, and runtime policy evaluation, each supported by a large
language model (LLM) that interprets artefacts, reasons over tool outputs, and produces structured security reports. To ensure
agent trustworthiness, every
agent generates a cryptographically signed attestation that is recorded in a permissioned blockchain via smart contracts, including an
agent registry, an immutable attestation log, and an enforceable release-policy module. Communication among
agents and with blockchain nodes is secured using a consortium-operated certificate authority, ensuring authenticated and tamper-resistant interactions. A detailed use-case and sequence flow demonstrate how a source code security
agent performs analysis, anchors its attestation on-chain, and triggers a verifiable allow/block deployment decision. The proposed framework of fers decentralised integrity transparent provenance, uninterrupted security assurance and a generalisable architecture to incorporate the agentic AI into the modern software supply chain security.
Read the full abstract ▾
Artificial Intelligence
MA-WAM: Multi-Agent World-Action Model for Test-Time Planning
Guowei Zou, Haitao Wang, Guoxin Wang et al. (+4 more)
Multi-
agent cooperative tasks require different
agents to execute a joint action simultaneously, and each
agent's action affects both the observations and responses of the other
agents. Hence, a world model is needed to predict the team return resulting from the joint actions of all
agents. A naive extension directly applies a single-
agent world model to each
agent's action when predicting the team return step by step. However, such an extension fails to capture the dependencies among the simultaneous actions of multiple
agents. We propose Multi-
Agent World-Action Model (MA-WAM), a test-time planning framework that enables a frozen multi-
agent flow policy to evaluate futures of candidate joint actions. To our knowledge, MA-WAM is the first test-time world-model planner for multi-
agent flow policies. MA-WAM predicts the consequences of each joint action according to cross-
agent dependencies and enables efficient candidate scoring. Across 30 offline multi-
agent reinforcement learning (MARL) settings on MAMuJoCo, SMAC, and MPE, MA-WAM achieves mean relative gains of 22.0% over direct execution and 25.6% over uniform action selection. Under the standard evaluation protocol on an A100 GPU, MA-WAM adds 12.1 ms, accounting for 2.5% of the measured generation-and-scoring time.
Read the full abstract ▾
All authors Guowei Zou, Haitao Wang, Guoxin Wang, Beiwen Zhang, Zhiquan Chen, Guojie Wang, Hejun Wu