Optics
Silicon nitride nanophotonics for low-redundancy 3D convolution
Shuai Hu, Jingcheng Li, Qichao Ding et al. (+5 more)
Three-dimensional (3D) convolution extracts correlations in high-dimensional data, but overlapping receptive fields introduce substantial redundant data movement. Here, we demonstrate a silicon nitride 3D optical convolution accelerator (3D-OCA) that reconstructs receptive fields through coordinate-aware wavelength-to-group-delay mapping. A deterministically serialized input tensor is broadcast onto multiple wavelength channels, and a chirped waveguide Bragg grating (CWBG) compensates the temporal offsets associated with 3D kernel coordinates. This arrangement continuously forms neighboring receptive fields without repeatedly rearranging and loading their shared input samples. The integrated CWBG provides a differential group delay of 2691 ps and a dispersion of 124.8 ps/nm. At 20 Gbaud, spatial-spectral processing of Indian Pines data yields convolution agreement with a coefficient of determination up to 0.997 and 97.9% classification accuracy, compared with 98.9% digitally. At 10 Gbaud, the optical convolution layer preserves spatiotemporal features and achieves 92.5% accuracy on a four-class KTH
video-recognition task. These results establish low-redundancy streaming 3D convolution across spectral and temporal data dimensions using the same optical delay architecture.
Read the full abstract ▾
All authors Shuai Hu, Jingcheng Li, Qichao Ding, Hongli Wang, Hailong Zhou, Chi Zhang, Jianji Dong, Xinliang Zhang
Representation Theory
Dirac operators for Cartan motion groups
Spyridon Afentoulidis-Almpanis and Eyal Subag
We introduce and study the algebraic Dirac operator for the Cartan motion group associated with a real reductive Lie group. We establish its fundamental properties, including a square formula involving a natural Casimir-type element, define the corresponding notion of Dirac
cohomology, and prove a strong form of Vogan's
conjecture: a module in the Dirac series is completely determined by its Dirac
cohomology. We explicitly determine the Dirac series of a Cartan motion group. We explain how the Dirac theory of a Cartan motion group can be obtained as the zero fiber of the recently introduced Dirac theory for the algebraic deformation family of the corresponding real reductive group.
Read the full abstract ▾
Quantum Physics
L-band single-photon generation from a cavity-coupled silicon C-center
Carolina Crosta, Kyu-Young Kim, Purbita Purkayastha et al. (+7 more)
Monolithic integration of single-
photon sources into silicon photonics offers a compelling solution for realizing scalable
quantum telecommunication architectures. Among the candidate silicon color centers, the C-center features a zero-
phonon emission in the telecom L-band, corresponding to the spectral window of minimum transmission loss in optical fibers. Despite this advantage, isolating single C-centers has remained an unrealized challenge in solid-state
quantum optics. In this work, we report the isolation of a single silicon C-center. We enhance the spontaneous emission rate of the zero-
phonon line by coupling the single emitter to a nanophotonic
crystal cavity, observing a fivefold reduction in the excited-state lifetime compared to bulk conditions, which corresponds to a Purcell factor lower bound of 33. Second-order autocorrelation measurements confirm single-
photon emission from the investigated defect, obtaining an optical purity of $g^{(2)}(0) = 0.23$ without background correction. Our results establish a native telecom L-band
quantum light source in silicon, laying the foundation for ultra-low-loss and long-distance
quantum applications.
Read the full abstract ▾
All authors Carolina Crosta, Kyu-Young Kim, Purbita Purkayastha, Abhijit Biswas, Jasvith Raj Basani, Amirehsan Alizadehherfati, Amirehsan Boreiri, Chang-Min Lee, Fabio Pezzoli, Edo Waks
Computer Vision
Codebook-Guided Cross-Modal Knowledge Distillation for Structurally Heterogeneous Features
Dae Ung Jo, Jongin Lim, YoungJoon Yoo et al. (+1 more)
Cross-modal knowledge
distillation transfers knowledge from a teacher modality to a student modality. Existing feature-level
alignment methods typically assume that teacher and student features reside in structurally alignable representation spaces. However, this assumption does not hold when cross-modal features are structurally heterogeneous and lack clear unit-level correspondence, such as 2D spatial visual grids and 1D temporal audio sequences, thereby limiting the applicability of feature-level
alignment. To address this challenge, we propose a cross-modal
distillation framework that enables effective knowledge transfer across structurally heterogeneous feature spaces via a vector-quantized codebook. Specifically, teacher features are abstracted into a set of vector-form codes regardless of their original feature structure, and the selected codes serve as concept-level anchors for student learning. Code selection is guided by both task relevance and student compatibility, allowing the student to receive transferable teacher knowledge without requiring direct unit-level feature
alignment. Experimental results across diverse cross-modal
distillation scenarios demonstrate the effectiveness of the proposed framework on classification and semantic
segmentation tasks.
Read the full abstract ▾
All authors Dae Ung Jo, Jongin Lim, YoungJoon Yoo, Daeho Um
Algebraic Geometry
Weighted projective spaces admitting $\mathbb Q$-Gorenstein smoothings to $\mathbb P^3$
Jungkai Alfred Chen and Yongnam Lee
We study well-formed weighted projective threefolds that admit $\mathbb Q$-Gorenstein smoothings to $\mathbb P^3$. Two families are known: the $\mathbb P^2$-type and the $Q$-type, and it is conjectured that these are the only possibilities. We derive numerical and local necessary conditions for such a smoothing. In addition to the anticanonical volume equation, constancy of the anticanonical
Hilbert polynomial yields a further identity when all codimension two singularities are of $A$-type. We also obtain a
semigroup condition governing the existence of global smoothing directions along codimension two curves with transverse $A$-type singularities. We apply these conditions to prove the expected classification in several cases. In particular, for every fixed square-free integer $d$, there are only finitely many $\mathbb Q$-Gorenstein smoothable spaces $\mathbb P(1,a,b,c)$ such that $\gcd(a,b)=d$. Our method reduces the possible weights to a finite exact computation; for every
prime $p\le100$, the computation produces only members of the two expected families. Finally, we prove the classification when $\gcd(a,b)=d$ and $a=d^2$.
Read the full abstract ▾
Machine Learning
Trident: Unifying Guarded Dispatch and Host Execution for PyTorch Triton Workloads
Jinjie Liu, Xiaoyan Liu, Shuhan Zhang et al. (+5 more)
User-written Triton kernels enable high-performance GPU computation within PyTorch, but their end-to-end latency can remain dominated by host-side orchestration, especially when device execution is short. Although torch.compile can generate native host wrappers for captured graphs, each invocation still passes through runtime-managed specialization lookup, guard evaluation, and preparation before reaching the wrapper. We present Trident, a compiler backend that removes this recurring overhead from the specialization cache-hit path. Trident introduces the Specialization Cache Module (SCM), which compiles guarded specialization selection, argument and execution-environment preparation, and host execution for multiple specializations into a single executable module. An invocation enters the SCM once, remains in compiled code when a specialization matches, and returns to Python only when a new specialization must be compiled. Built on Torch-MLIR, Trident lowers guards and host-side orchestration to native code while retaining calls to optimized runtime implementations of supported ATen operators. Our evalu- ation on two LLMs shows that Trident achieves up to a 1.47x speedup in model-level end-to-end latency over eager execution and up to 1.68x over torch.compile.
Read the full abstract ▾
All authors Jinjie Liu, Xiaoyan Liu, Shuhan Zhang, Wenjia Sun, Ruilin Yang, Chunlei Men, Yonghua Lin, Shaohua Li
Quantum Physics
Approximate majorization and high-order capacity of quantum depolarizing channels
Zhiwei Song, Xin Wang, Marco Tomamichel et al. (+1 more)
A central challenge in
quantum information theory is to determine whether
entanglement provides an advantage for information processing when coherent operations across multiple channel uses are allowed. For depolarizing channels, King's proof of the additivity of the minimum output
entropy rules out such an advantage for classical communication at leading asymptotic order, while leaving open the possibility of an advantage for higher orders. In this work, we
conjecture a majorization relation for the outputs of depolarizing channels and prove an approximate version of this
conjecture. Our proof combines an exact ordering of shifted log-determinants with an anti-concentration bound for the spectrum of the product-state output, refining King's analysis based on Schatten norms. As a direct application, we obtain a tight third-order expansion for the classical capacity of
quantum depolarizing channels. Consequently, entangled codewords and collective decoding can improve the message size over the product-coding strategy by at most $O(\log \log n)$ bits, sharpening our understanding of the role of
entanglement in the finite-blocklength regime.
Read the full abstract ▾
All authors Zhiwei Song, Xin Wang, Marco Tomamichel, Kun Fang
Machine Learning
Differentiating Bisimulation Metrics: A Framework for Parametric Markov Chain Fitting via Bicausal Optimal Transport
Sergio Calo, Amy Zhang, Javier Segovia-Aguas et al. (+1 more)
Many problems in sequential decision-making, such as imitation learning from observations, state-space compression, world-model learning, and sim-to-real transfer, can be reduced to learning a model such that a notion of distance with respect to the target process is minimized. We consider this general framework and consider the bisimulation metric, equivalently Bicausal Optimal Transport (BOT), as the notion of distance to minimize. We show that BOT, since it can be formulated as a linear program (LP), is differentiable with respect to the model dynamics. We then derive an exact closed-form gradient via the envelope
theorem applied to the LP saddle point. The result is a general
algorithm, Differentiable Bicausal Optimal Transport (D-BOT), that can be applied to each of the problems above. The proposed
algorithm learns the best model by alternating between distance computation and gradient steps. We apply D-BOT for three different settings: state-space compression, parametric model learning, and imitation learning from observations (ILfO). We show empirical results that confirm the viability of all three instantiations.
Read the full abstract ▾
All authors Sergio Calo, Amy Zhang, Javier Segovia-Aguas, Anders Jonsson
High Energy Physics (Phenomenology)
Analytic next-to-leading-order helicity cross sections for vector-boson production at finite transverse momentum
Stefano Camarda, Leandro Cieri, Giancarlo Ferrera et al. (+1 more)
We present a complete analytic calculation of the next-to-leading-order (NLO) QCD cross sections for the production of an electroweak vector
boson ($W^\pm$, $Z/\gamma^*$) at finite transverse momentum, decomposed into the full set of helicity cross sections that determine the angular coefficients $A_0$-$A_7$ of the
lepton-pair angular distribution in the Collins-Soper frame. The calculation includes the unpolarised cross section and all eight helicity projections for both neutral- and charged-current Drell-Yan production, the axial
quark-triangle contributions in the $m_t\to\infty$ limit and with full top-
quark-mass dependence, and closed-form expressions for all coefficient functions. The results are implemented in the DYTurbo program and validated at the level of a few times $10^{-4}$ against its numerical implementation derived from MCFM, as well as through the analytic $q_T\to0$ limit. The analytic implementation achieves a speed-up of more than five orders of magnitude with respect to the numerical code, enabling fast and numerically stable predictions for precision electroweak measurements at the LHC. The calculation was carried out with extensive use of large
language models (Anthropic Claude) as autonomous
agents; we document the methodology and the associated computational and
token budgets.
Read the full abstract ▾
All authors Stefano Camarda, Leandro Cieri, Giancarlo Ferrera, Lorenzo Rossi
Group Theory
Redundant generation in $\operatorname{PSL}_2\left(\mathbb{Q}_p\right)$
Yair Glasner and Amit Levinson-Sela
Given a free group on $n$ generators, $F_n$, and a topological group $G$, $\operatorname{Aut}\left(F_n\right)$ acts on $\operatorname{Epi}\left(F_n,G\right)$, the set of homomorphisms $f:F_n \rightarrow G$ with dense image. $f\in \operatorname{Epi}\left(F_n,G\right)$ is called redundant if there is a proper free factor $A < F_n$ whose image under $f$ is dense in $G$. For $n \ge 3$, $G = \operatorname{PSL}_2 \left(\mathbb{Q}_p\right)$, we prove that every $f\in \operatorname{Epi}\left(F_n,G\right)$ with torsion-free image is redundant, and give sufficient conditions for $f$ to be redundant even when its image is not torsion-free.
Read the full abstract ▾
Artificial Intelligence
Asking for What Was Never Requested: Horizontal and Vertical Proactivity in Agents
Ido Levy, Asaf Yehudai, Segev Shlomov et al. (+2 more)
An
agent that uses tools typically responds to what the user explicitly asks, yet completing the task may require information the user never requested. Work on proactive
agents mainly studies whether and when an
agent should act on its own, not what information it should pursue. We study a distinct axis of proactivity: its content. Horizontal proactivity pursues unstated information that the current context already identifies, and vertical proactivity pursues needs that only earlier evidence reveals. A need graph, recovered from a
benchmark's own decomposition, records which needs depend on which, so both forms, and whether the
agent stops at the right time, can be scored from a transcript without a model judge. To learn this behavior, we propose Q&D (questioner and drafter), which trains a questioner to prefer the question whose continuation retrieves more of the required evidence, with no reward model or judge. On held-out splits of three multi-hop question-answering
benchmarks, at equal
retrieval spend, the trained questioner improves both forms of proactivity over the same model, prompted, and outperforms a prompted model $15\times$ larger in the same role on two of the three, and the gain persists after controlling for question volume and length. Without further training, we place the questioner in an interactive customer-service
agent with a simulated customer, where it completes more tasks while asking fewer questions, and in retail it outperforms the $15\times$ larger model with fewer follow-up turns from the customer. These results show that proactivity depends not only on whether an
agent acts without being asked, but also on what it chooses to pursue and when it stops.
Read the full abstract ▾
All authors Ido Levy, Asaf Yehudai, Segev Shlomov, Asaf Adi, Leshem Choshen
Optics
Photonic sequential logic circuits for stateful intelligence
Jingcheng Li, Wenkai Zhang, Jialong Zhang et al. (+6 more)
By virtue of its high energy efficiency and ultra-low latency, optical computing is regarded as a highly promising computing paradigm in the post-Moore era. However, constructing a Turing-complete optical computing for next-generation AI paradigms, from large
language models (LLMs) to real-time autonomous systems, requires not only stateless function mapping units, but also sequential logic
circuits capable of storing historical states and processing dynamic data streams. Current photonic processors lack this temporal
memory, and the state-of-the-art optical sequential schemes remain impractical due to signal attenuation and the absence of programmable clock control. Here, we propose a photonic sequential logic
circuits (PSLC) to resolve this fundamental challenge. Leveraging a uniquely designed active optoelectronic feedback network, we construct a complete family of photonic sequential logic devices, comprehensively encompassing set-reset latches, D latches, and edge-triggered master-slave D flip-flops. We further construct photonic sequence detectors and asynchronous counters, verifying the capability of the architecture to execute both synchronous and asynchronous sequential tasks, alongside its system-level scalability. Ultimately, by combining the constructed PSLC with stateless function mapping units, we establish a universal hardware paradigm for stateful intelligence. We validate this paradigm by executing highly reliable real-time drone obstacle avoidance in the physical domain, alongside Shakespearean-style text generation in the symbolic domain. This work provides the crucial missing piece of the puzzle for optical computing to achieve stateful computing, establishing a definitive hardware foundation for advancing toward Turing-complete optical computing.
Read the full abstract ▾
All authors Jingcheng Li, Wenkai Zhang, Jialong Zhang, Bo Wu, Qichao Ding, Hongli Wang, Hailong Zhou, Jianji Dong, Xinliang Zhang
General Relativity and Quantum Cosmology
The Irreversible Depletion of Dark Matter Spikes by Extreme-Mass-Ratio Inspirals
Charlie Sharpe, Yonadav Barry Ginat, Thomas F. M. Spieksma et al. (+2 more)
Dense dark matter spikes around massive black holes (MBHs) have been widely predicted to produce gravitational-wave dephasing in extreme-mass-ratio inspirals (EMRIs) that are potentially observable by LISA. We show that EMRIs, the very objects used to probe these spikes, also destroy them. Combining Fokker-Planck models of
nuclear star clusters with post-Newtonian three-body simulations of EMRI-dark matter encounters, we demonstrate that repeated gravitational slingshots irreversibly eject dark matter particles from the cluster's loss cone over gigayear time-scales. For MBHs at $z = 3$ with masses $\lesssim 10^5 \,M_\odot$, even conservative EMRI rates of $\mathcal{O}(1-10) \, \mathrm{Gyr}^{-1}$ suffice to deplete the spike density by orders of magnitude; for more realistic rates of $\mathcal{O}(100-300) \, \mathrm{Gyr}^{-1}$, the mass range extends to $\lesssim 10^6 \,M_\odot$. As a result, the gravitational-wave dephasing expected from such systems is greatly reduced, narrowing the prospects of detecting DM through this channel. Only heavy, high-
redshift systems with sufficiently low EMRI rates – for example, a $\sim 10^6 \,M_\odot$ MBH at $z \gtrsim 5$ with $\lesssim 100$ EMRIs per Gyr – can induce dephasing that exceeds one radian.
Read the full abstract ▾
All authors Charlie Sharpe, Yonadav Barry Ginat, Thomas F. M. Spieksma, Alexander Heger, Bence Kocsis
Programming Languages
DatalogBench: Evaluating Large Language Models on Text-to-Datalog Synthesis
Yuan Li, Hanyun Jiang, Guowei Tian et al. (+2 more)
Datalog underpins
reasoning tasks such as program analysis, but its programs are hard to write. Existing synthesizers automate this task but require users to state their intent as input-output examples. Large
language models (LLMs) suggest a more natural route, text-to-Datalog synthesis from a natural-
language question, yet how well they do so has not been systematically evaluated. We present DatalogBench, a
benchmark of 136 text-to-Datalog synthesis tasks curated from existing Datalog-based artifacts. Synthesized programs are graded by execution on held-out inputs against an oracle validated by mutation analysis. Across six LLMs and four prompting configurations, exact match peaks at 68.4%, and relation descriptions or an input-output example have only modest, model-dependent effects. Under direct prompting, most failures occur at compile time, typically because a model invents auxiliary predicates that it never declares or types consistently. Two coding
agents reach up to 83.8% and eliminate nearly all such failures, leaving mostly semantic errors concentrated in recursive tasks. DatalogBench thus identifies recursive
reasoning and decomposition as open challenges for current LLMs and
agents, and offers a reliable, execution-grounded measure of both.
Read the full abstract ▾
All authors Yuan Li, Hanyun Jiang, Guowei Tian, Chengpeng Wang, Peisen Yao
Analysis of PDEs
Hölder estimates up to the singular time for the Córdoba-Córdoba-Fontelos equation
Ángel Castro, Alberto Enciso and Antonio J. Fernández
We prove that the solution to the Córdoba-Córdoba-Fontelos equation $\theta_t-u\theta_x=0$, $u=H\theta$, on the real line with analytic initial datum $\theta_0(x)=(1+x^2)^{-1}$ develops a finite-time singularity while remaining uniformly bounded in $C^{2/5}$. We also prove that, as the solution approaches the singular time, the $C^{2/3}$ seminorms of both $\theta$ and $u$ diverge. The main ingredient is an integral representation in terms of a probability measure that is preserved by the nonlinear
evolution. From this representation we derive a pointwise quadratic estimate for the key nonlinear term in the equation for $u$, using a comparison of positive kernels. Together with
evolution identities for the moments of the measure, this estimate yields both the uniform Hölder bound and the divergence of the higher seminorms.
Read the full abstract ▾
Quantum Physics
Ultra-high-distance quantum memories from amplified qLDPC codes
Zijian Liang, Boren Gu, Yu-An Chen et al. (+2 more)
A large code distance does not by itself guarantee a well-protected
quantum memory, as faults during syndrome extraction can propagate into correlated data errors. Certifying
circuit distance becomes demanding as
circuits grow. In this work, we introduce distance amplifiers, a modular approach to increasing both code distance and certified
circuit-level protection in
quantum low-density parity-check codes. The construction tensors a base code with an amplifier, adapting their syndrome-extraction
circuits to the amplified code. To certify the resulting
memory, we develop a fault-response certificate that yields rigorous lower bounds on
circuit distance. Suitable gate ordering and flag
qubits control correlated errors, allowing us to establish explicit conditions under which the amplified
circuit distance equals the product of the constituent
circuit distances. We prove this multiplicative relation for three types of amplifiers: the flagged $[[4,2,2]]$, hypergraph-product, and rotated surface codes. The certificate extends recursively, so the
circuit distance multiplies exactly at every amplification step. For example, four successive applications of the flagged $[[4,2,2]]$ amplifier to a $[[18,4,4]]$ seed yield a $[[13320,64,64]]$ code with the
circuit distance $d_{\mathrm{circ}} \geq 48$ and check weights $w \leq 14$. We further show how individual logical
qubits can be tracked explicitly through amplification. By enabling high-distance
quantum memories to be built and certified from smaller building blocks, our framework offers a systematic route toward scalable fault-tolerant
quantum computing.
Read the full abstract ▾
All authors Zijian Liang, Boren Gu, Yu-An Chen, Jens Eisert, Zongyuan Wang
Computer Vision
Seeing Is Not Addressing: Auditing Linguistic Access to Frozen Visual Geometry
Woosang Jeon, Jiwon Yang, Soo Chung et al. (+1 more)
Visual distinctions are often finer than those reflected in linguistic conceptualization.
Vision-
language models exhibit a similar asymmetry: a distinction can remain discriminable in frozen image geometry while being weakly addressable through the native text interface. We study this gap by separating visual discriminability from linguistic addressability in text-to-image
retrieval. Using FactorAtlas, a fully crossed testbed of 23,040 images spanning shape, hue, pattern, and nuisance variation, we compare both readouts on held-out images of the same distinctions. We then derive image-side contrasts that separate each value from its alternatives for matched visual grounding, and test whether this reduces the native-text access gap across factors and models. Direction-specific and visual-absence controls tie these gains to the relevant visual contrast; the gains persist after global
alignment and extend to compositional
retrieval and natural images. Together, these results show that visual discriminability and linguistic addressability need not coincide, and that matched visual grounding can probe and reduce the resulting access gap.
Read the full abstract ▾
All authors Woosang Jeon, Jiwon Yang, Soo Chung, Taehyeong Kim
Computer Vision
AESOP: Asymmetric Human-Camera Generation with Translation-Intensity Control
Jingzhong Lin, Zhanke Wang, Heng Li et al. (+9 more)
Human motion defines an action, while a camera trajectory determines how it is presented. Camera generation for a given human motion and joint human-camera generation are usually treated as separate tasks, although both share an asymmetric dependency: human motion can be generated independently, whereas the camera responds to the realized action. We introduce AESOP, a unified framework with an independent human pathway and a shared human-conditioned camera generator. Its asymmetric architecture serves both tasks while preserving the human output during camera generation. Although human context anchors the shot to the action and camera text describes its movement, translation intensity remains underspecified. We therefore construct trajectory pairs that differ in camera translation magnitude while sharing human motion and camera text, then use these pairs to learn an explicit intensity condition. Experiments on the PulpMotion
dataset demonstrate strong camera distributional and framing quality in both tasks and effective control over camera translation intensity. Project page is available at https://ripemangobox.github.io/AESOP.
Read the full abstract ▾
All authors Jingzhong Lin, Zhanke Wang, Heng Li, Wenxiang Liu, Zhao Zhang, Kecheng Tang, Dongdong Xiang, Changbo Wang, Di Kang, Chunchao Guo, Linchao Bao, Gaoqi He
Plasma Physics
Laser-Polarization Dependence of Betatron X-Ray Production and Angular Collection in Laser-Wakefield Acceleration
Andrei Ciprian Berceanu, Vojtěch Horný, Dan Stutman et al. (+1 more)
Laser polarization affects both the production of betatron x rays and their angular distribution, which together determine the
photon yield collected within a fixed angular acceptance. We investigate these effects using particle-in-
cell simulations of
laser-wakefield acceleration with linearly polarized (LP) and circularly polarized (CP) pulses of equal energy in helium–nitrogen targets of two different lengths. In the longer target, CP produces more
photons early in propagation, whereas LP produces more later. For 20–60-keV
photons transmitted through 2 mm of aluminum, the LP/CP yield ratio before angular selection changes from 0.88 early on to 1.91 in a late stage. Late-emitting LP electrons develop pronounced transverse oscillations along the
laser polarization direction, accompanied by angular broadening of the radiation. This broadening largely removes the late LP production advantage within a central $\pm5$-mrad acceptance, where the two yields are nearly equal. Integrated over the full interaction, however, LP supplies more central
photons, while CP provides more uniform illumination. In the shorter target, the extracted charges above 100 MeV are nearly equal, yet transport through a quadrupole triplet delivers 750 pC for LP and 500 pC for CP in the 350–500-MeV energy band within a radius of 5 mm. Collected
photon and electron yields are thus shaped by propagation length and collection acceptance.
Read the full abstract ▾
All authors Andrei Ciprian Berceanu, Vojtěch Horný, Dan Stutman, Paolo Tomassini
Machine Learning
Interacting particle guidance for sampling reward-tilted generative priors
Adhithyan Kalaivanan, Zheng Zhao, Jens Sjölund et al. (+1 more)
Inference-time steering adapts pretrained
diffusion and flow-based models to new tasks, e.g., to generate samples from a conditional distribution or samples with desired properties, without retraining. This can be formalized as sampling from a reward-tilted generative prior. As exact sampling from this distribution is intractable, guidance-based methods rely on approximations producing biased samples, and sequential Monte Carlo (SMC) methods correct for this bias using importance weights. However, while exact in the large particle limit, SMC suffers from weight degeneracy and particle collapse in practice. We propose interacting particle guidance (IPG), which replaces reweighting with transport. The particles interact through an additional drift, derived from the Feynman–Kac PDE to cancel the reweighting term, and remain unweighted. Choosing the drift in a reproducing kernel
Hilbert space yields a closed-form solution that is cheap to compute, with negligible overhead compared to SMC. We demonstrate the method on Gaussian mixtures with known posteriors, and on high-dimensional image inpainting and
protein structure inference tasks.
Read the full abstract ▾
All authors Adhithyan Kalaivanan, Zheng Zhao, Jens Sjölund, Fredrik Lindsten
Computation and Language
Follow the Entities: A Corpus Map for Agentic Search
Soyeong Jeong, Sujay Kumar Jauhar, Sung Ju Hwang et al. (+1 more)
Answering questions and completing tasks over large document collections often requires connecting evidence spread across multiple documents, such as a project's approval recorded in one, its requirements in another, and its latest status in a third. Recent LLM
agents approach this by iteratively searching the full corpus rather than reading only a fixed set of top-ranked documents. However, when the corpus is exposed only as a flat collection of files, a relevant document gives no indication of how it relates to others, so the
agent must rediscover these relationships for every query, often missing complementary evidence while simultaneously consuming substantial additional
tokens. To address this, we introduce CorpusMap, a
navigation layer that organizes the corpus around its recurring entities, which are identifiable from the documents themselves and can link a single document to many others across sources. Specifically, CorpusMap represents each recurring entity as an Entity Page that aggregates information about it and links to every document that refers to it, forming a graph between entities and documents that the
agent can traverse to gather otherwise disconnected evidence. Moreover, since CorpusMap is constructed offline by resolving mentions of the same entity across documents, its links are shared across queries rather than rediscovered repeatedly at inference time. Using 7 different models with 3
benchmark datasets, we show that CorpusMap improves both evidence discovery and answer quality over raw-corpus agentic search while using fewer
tokens on average, and further outperforms 4 alternative
navigation layers, suggesting that entities serve as effective anchors for navigating large document collections.
Read the full abstract ▾
All authors Soyeong Jeong, Sujay Kumar Jauhar, Sung Ju Hwang, Andrew Joohun Nam
Computer Vision
ResComEmb: Effective and Efficient Multimodal Embedding via Residual Homogeneity Compression
Zijing Cai, Yuzhe Wang, Jingxian Zhu et al. (+2 more)
Multimodal large
language models (MLLMs) have shown strong potential for universal
multimodal representation learning. However, existing methods either compress each input into a single vector, limiting fine-grained expressiveness, or retain long sequences of visual-
token vectors, incurring substantial storage and interaction costs. To resolve this trade-off, we propose ResComEmb, a trainable framework for effective and efficient universal multi-vector
multimodal embedding. ResComEmb first encodes each input at native dynamic resolution into ordered global, intermediate, and fine-grained views. After MLLM contextualization and
embedding projection, a trainable Residual Homogeneity Compression (RHC) module reduces within-granularity redundancy and cross-granularity repetition under explicit visual
token budgets. Then, ResComEmb introduces a length-adaptive Bidirectional Late-Interaction Matching mechanism for robust query-document scoring, which averages the strongest
token-level matches in each direction and combines the two scores using a weight based on how many valid
tokens each side has. Extensive experiments on MMEB, ViDoRe V1, and ViDoRe V2 show that ResComEmb produces higher-quality universal
multimodal embeddings than VLM2Vec-V2, and outperforms ColQwen2.5 in visual document
retrieval using only 37.5% of its full visual
token budget, demonstrating a favorable effectiveness-efficiency trade-off.
Read the full abstract ▾
All authors Zijing Cai, Yuzhe Wang, Jingxian Zhu, Fengbin Zhu, Richang Hong
High Energy Astrophysical Phenomena
Multi-wavelength polarization degree dependence of Mrk 421 for different particle energy distributions
Hritwik Bora, Ranjeev Misra and Rupjyoti Gogoi
We report on the constraints obtained from the multi-wavelength polarization degree and energy spectra measurements for the High Synchrotron Peaked Blazar (HBL) source Mrk 421 using different forms of the electron energy distribution. We find that the optical and X-ray data are consistent with the electron energy distribution being a broken power-law (BPL) or other distributions, with all electrons experiencing the same
magnetic field configuration. This is in contrast to other HBL sources like the Mrk 501, where, for the BPL distribution, the X-ray producing higher energy electrons need to exist in a more ordered
magnetic field environment than the optical producing lower energy ones. This is because the break energy of the spectrum produced by the BPL distribution for Mrk 421 lies below the IXPE energy band. We discuss the implications of the findings.
Read the full abstract ▾
Computation and Language
CredWise: A Controlled Agentic Decision-Intelligence Framework for Explainable and Auditable Credit-Risk Assessment
Aakash Kumar Tiwari
Credit-risk prediction is important in banking, but a prediction alone does not explain why an applicant is risky or how it should be combined with other evidence. This paper presents CredWise, a decision-support framework that integrates credit-risk prediction, probability calibration, explainable artificial intelligence, policy
retrieval, SQL analytics, and controlled
agent-based workflows. An XGBoost model is trained on Lending Club data (1,345,310 loans, 18 features) using a temporal split: 2007–2016 for training, 2017 for validation, and 2018 for testing. On the 2018 test set, the calibrated model achieved a ROC-AUC of 0.7109, PR-AUC of 0.2993, F1-score of 0.3714, and accuracy of 65.44%. Calibration reduced the Brier score from 0.2157 to 0.1273 and the expected calibration error from 0.2862 to 0.0585. SHAP explanations were temporally stable, with a Spearman correlation of 0.9959 between 2017 and 2018 feature rankings. On 28 labeled queries covering nine policy sections, FAISS achieved the best Hit@1 (0.929) and MRR (0.964), while all three
retrieval methods reached Hit@5 = 1.0.
Agent routing achieved 95.6% accuracy (43 of 45 cases), and the SQL
benchmark scored 1.0 on exact-match, execution-success, and result-match across six cases. These results show that CredWise can combine predictions, explanations, policy evidence, and structured analytics in one controlled workflow. It is an academic research prototype, and final decisions remain with a human reviewer.
Read the full abstract ▾
Plasma Physics
Cold atmospheric plasma decontaminates Arabidopsis thaliana seeds and remodels seedling fungal microbiota
Lena Taras, Nicole Chaumont, Caroline Kunz et al. (+2 more)
Seed-associated microorganisms influence seed quality, seedling establishment, and plant health and also constitute a major source of seed-borne pathogens. Cold atmospheric
plasma (CAP) has emerged as a promising alternative to chemical seed treatments because of its antimicrobial activity, although its effects on fungal communities associated with developing seedlings remain poorly understood. Here, Arabidopsis thaliana seeds from two ecotypes (Columbia and Landsberg erecta) were exposed to CAP for 5 or 15 min. Seed decontamination efficiency was assessed by culturing on malt extract agar and nephelometric analyses, while fungal communities associated with seedlings derived from treated and untreated seeds were characterized by ITS1 amplicon sequencing. CAP efficiently reduced fungal contamination without affecting seed germination. CAP altered the composition of fungal communities associated with developing seedlings, but the magnitude of these changes depended on seed batch and ecotype. Dominant taxa markedly declined after treatment, whereas several low-abundance taxa increased in relative abundance. These findings demonstrate that CAP is an effective pesticide-free technology for seed decontamination and can also reshape fungal communities associated with developing seedlings, highlighting broader ecological consequences of
plasma-based seed treatments.
Read the full abstract ▾
All authors Lena Taras, Nicole Chaumont, Caroline Kunz, Thierry Dufour, Christophe Bailly
Artificial Intelligence
OptiCom : A Unified Framework for State-Conditioned Composition in LLM-Driven Optimization
Chenxing Wei, Sichen Liu, Lizhao Liu et al. (+6 more)
Large
language models (LLMs) are increasingly deployed to solve complex scientific and practical problems via iterative optimization. However, dynamically coordinating diverse search mechanisms as candidate quality, failure modes, and resource budgets evolve remains a critical open challenge. Targeted empirical diagnostics reveal that mechanism effectiveness is highly state-dependent. Motivated by this, we analyze how individual decisions drive final outcomes, decomposing the expected terminal improvement under a shared budget into cumulative decision opportunities minus cumulative selection losses. Guided by this opportunity-loss theoretical foundation, we propose OptiCom, a unified framework that represents LLM-driven optimizers within a shared configuration space: C=(A,Q,O,E,M,S), corresponding to artifact, query, operator, evaluation,
memory, and strategy. Operating within this space, a fast LLM-based Optimization Controller dynamically composes immediate mechanisms through structured Action Packages, while a slower Strategy Adapter refines long-term selection preferences, operator weights, and templates based on accumulated trajectory feedback. Comprehensive evaluations across 32
benchmark groups demonstrate the superiority of framework: OptiCom achieves an average Max-score rank of 1.72 among 14 evaluated configurations, securing the top score in 23 groups. Ultimately, these results highlight the broad applicability and high extensibility of OptiCom as a general-purpose paradigm for robust LLM test-time scaling.
Read the full abstract ▾
All authors Chenxing Wei, Sichen Liu, Lizhao Liu, Ningyuan Sun, Chen Bingzhou, Ying He, Bo Jiang, Fei Yu, Yao Shu
Artificial Intelligence
Information Bottleneck-Guided Adaptive Hypergraph Transformer for Brain Disease Diagnosis
Jingxi Feng, Xudong Chen, Yifan Zhang et al. (+5 more)
Exploring high-order correlations and long-range dependencies in
brain networks holds significant value for both neuroscience research and clinical diagnosis. However, previous studies have lacked a unified integration of high-order and long-range dependency information in
brain networks, and there is substantial redundancy behind various types of information. These issues limit their effectiveness in the diagnosis of
brain diseases. To address this, we propose an Information Bottleneck-Guided Adaptive HyperGraph
Transformer (IBAHGT). By incorporating the information bottleneck (IB) principle, this approach enables adaptive learning of high-order correlations and both short- and long-range dependencies within a unified framework for
brain network analysis, achieving high-precision
brain disease diagnosis. IBAHGT consists of three key components: an information bottleneck-guided adaptive hypergraph convolution, which introduces a novel hypergraph information bottleneck (HIB) principle to adaptively learn hypergraph message-passing weights between nodes and hyperedges, optimizes information flow and captures high-order information in
brain networks that is maximally informative and minimally redundant (MIMR). The
Transformer encoder captures global information within
brain networks through the
attention mechanism, specifically modeling short- and long-range dependencies. An information bottleneck-guided node-level adaptive fusion employs the IB principle to learn independent weights for each node, facilitating the fine-grained integration of high-order information and global information to obtain an efficient representation for downstream tasks. Extensive experiments demonstrate that the proposed method outperforms current state-of-the-art methods and can identify biomarkers for clinical applications.
Read the full abstract ▾
All authors Jingxi Feng, Xudong Chen, Yifan Zhang, Heming Xu, Hongcheng Han, Xijing Wang, Dong Zhang, Shaoyi Du
Commutative Algebra
Powers of Alexander Duals of Stanley-Reisner Ideals of Chessboard Complexes
Arka Ghosh and S. Selvaraja
Let $\Delta_{m,n}$ be the chessboard complex, and let $J_{m,n}$ be the Stanley-Reisner ideal of its Alexander dual. We study the symbolic and ordinary powers of $J_{m,n}$, focusing on their homological and algebraic invariants. First, we characterize when $J_{m,n}^{(q)}$ has linear quotients. For the ordinary powers, we give sufficient conditions on $m$ and $n$ for $J_{m,n}^q$ to have linear quotients. We determine the maximal degree of a minimal generator and the Castelnuovo-Mumford regularity of $J_{m,n}^{(q)}$, obtaining explicit formulas for all $q\ge2$ and showing that both invariants have the same asymptotic slope. We also compute the initial degrees and Waldschmidt constant, proving that $\alpha(J_{m,n}^{(q)})$ is a degree-one quasi-
polynomial of period two and $\widehat{\alpha}(J_{m,n})=mn/2$. Explicit formulas for the $\mathrm{v}$-numbers of symbolic and ordinary powers are obtained. Finally, for every $m\ge3$, we show that $J_{m,2m-1}$ is not weakly polymatroidal, disproving a
conjecture of Lu and Wang in this setting.
Read the full abstract ▾
Cryptography and Security
Beyond Semantic Narrowing: Robust and Efficient LLM Watermarking with Hamming Neighborhoods
Zewen Sun, Tongyang Zhao, Liyao Xiang et al. (+3 more)
Semantic watermarking improves robustness against watermark removal attacks by
embedding detectable signals into sentence-level representations. However, existing watermarking methods typically impose watermark-specific semantic preferences on generated sentences without explicitly accounting for the highly non-uniform and context-dependent semantic preference of LLM generation. When these two preferences are poorly aligned, many natural continuations become incompatible with the watermark, causing semantic narrowing: reduced semantic freedom, increased resampling cost, and potential degradation on tasks with strict semantic requirements. To alleviate this problem, we propose HammingMark, which uses the semantic hash of the preceding sentence as a dynamic center and accepts candidates whose hashes fall within its Hamming neighborhood. Defining watermark validity over a Hamming neighborhood in compact hash space retains a larger fraction of naturally likely semantic continuations. The coarse many-to-one hash mapping further allows diverse semantic realizations to remain watermark-valid. Experiments on C4 and BookSum show that HammingMark achieves strong robustness, high detectability, and near-unwatermarked generation quality, requiring only 2.2 sampled candidates per accepted sentence,a 72.8% reduction compared with the most sampling-efficient existing method. On more complex tasks with strict semantic constraints, HammingMark achieves the highest detection rates with the highest or tied-highest ROUGE-L scores, demonstrating its effectiveness in balancing watermark detectability and generation quality under constrained generation settings.
Read the full abstract ▾
All authors Zewen Sun, Tongyang Zhao, Liyao Xiang, Mingxuan Ma, Lingzhe Wang, Zhiyuan Li
Cryptography and Security
When Cyber Scoring Systems Diverge: An Empirical Comparison
Kirsi Hellsten, Joni Herttuainen, Ambrose Kam et al. (+3 more)
Vulnerability scoring systems underpin cyber patch prioritization and risk management, but their comparative behavior is almost always assessed in the abstract, through correlation studies in IT vulnerability databases, rather than by the operational consequences they produce when embedded in a system-level risk model. Here we present an empirical comparison of four vulnerability scoring systems, namely CVSS (Common Vulnerability Scoring System), EPSS (Exploit Prediction Scoring System), SSVC (Stakeholder-Specific-Vulnerability Categorization), and IronMiner (operationally calibrated proprietary scoring system). As a substrate for comparison, we use a reconstruction of the 2015 Ukraine Power Grid operational-technology (OT) network that provides a documented incident
topology. The results show a high degree of disagreement between the scoring systems. This suggests that the choice of the scoring system could significantly influence mitigation strategies and vulnerability prioritization, implying that a composite or hybrid scoring approach could offer a more suitable solution.
Read the full abstract ▾
All authors Kirsi Hellsten, Joni Herttuainen, Ambrose Kam, Arlanda Johnson, David Welsh, Kimmo Kaski