March 2025 arXiv papers — page 80

Daniel Haziza, Timothy Chou, Dhruv Choudhary, Luca Wehrstedt

In this paper, we demonstrate how to leverage 2:4 sparsity, a popular hardware-accelerated GPU sparsity pattern, to activations to accelerate large language model training and inference. Crucially we exploit the intrinsic sparsity found in Squared-ReLU activations to provide this acceleration with no accuracy loss. Our approach achieves up to 1.3x faster Fee

Amir Hossein Baradaran

Predictive maintenance is a key strategy for ensuring the reliability and efficiency of industrial systems. This study investigates the use of supervised learning models to diagnose the condition of electric motors, categorizing them as "Healthy," "Needs Preventive Maintenance (PM)," or "Broken." Key features of motor operation were employed to train various

Brajagopal Das, Cinthia Piamonteze, Lior Kornblum

Magnetic anisotropy in complex oxides often originates from the complex interplay of several factors, including crystal structure, spin-orbit coupling, and electronic interactions. Recent studies on Ru-substituted $La_{0.70}Sr_{0.30}MnO_3$ (Ru-LSMO) films demonstrate emerging magnetic and magneto-transport properties, where magnetic anisotropy plays a crucia

Maisha Binte Rashid, Pablo Rivas

In this paper, we aim to investigate the capabilities of multimodal machine learning models, particularly the OpenFlamingo model, in processing a large-scale dataset of consumer-to-consumer (C2C) online posts related to car parts. We have collected data from two platforms, OfferUp and Craigslist, resulting in a dataset of over 1.2 million posts with their co

Anup Pradhan Sakhya, Mazharul Islam Mondal, Milo Sprague, Resham Babu Regmi

Recently, there has been a growing interest in altermagnetism, a novel form of magnetism, characterized by unique spin-splitting even in the absence of both net magnetic moments and spin-orbit coupling. Despite numerous theoretical predictions, experimental evidence of such spin-splitting in real materials remains limited. In this study, we use angle-resolve

Amir Hossein Baradaran

Advancements in the defense industry are paramount for ensuring the safety and security of nations, providing robust protection against emerging threats. Among these threats, hypersonic missiles pose a significant challenge due to their extreme speeds and maneuverability, making accurate trajectory prediction a critical necessity for effective countermeasure

Yichen Huang, Zachary Novack, Koichi Saito, Jiatong Shi

Despite significant recent advances in generative acoustic text-to-music (TTM) modeling, robust evaluation of these models lags behind, relying in particular on the popular Fr\'echet Audio Distance (FAD). In this work, we rigorously study the design space of reference-based divergence metrics for evaluating TTM models through (1) designing four synthetic met

Niki van Stein, Anna V. Kononova, Lars Kotthoff, Thomas Bäck

Large Language Models (LLMs) have demonstrated great promise in generating code, especially when used inside an evolutionary computation framework to iteratively optimize the generated algorithms. However, in some cases they fail to generate competitive algorithms or the code optimization stalls, and we are left with no recourse because of a lack of understa

Oriol Vendrell-Gallart, Nima Negarandeh, Zahra Zanjani Foumani, Mahsa Amiri

Foundation models are at the forefront of an increasing number of critical applications. In regards to technologies such as additive manufacturing (AM), these models have the potential to dramatically accelerate process optimization and, in turn, design of next generation materials. A major challenge that impedes the construction of foundation process-proper

Qiyu Kang, Xuhao Li, Kai Zhao, Wenjun Cui

Fractional-order differential equations (FDEs) enhance traditional differential equations by extending the order of differential operators from integers to real numbers, offering greater flexibility in modeling complex dynamical systems with nonlocal characteristics. Recent progress at the intersection of FDEs and deep learning has catalyzed a new wave of in

Spyridon Raptis, Paul Kling, Ioannis Kaskampas, Ihsen Alouani

Neuromorphic computing based on spiking neural networks (SNNs) is emerging as a promising alternative to traditional artificial neural networks (ANNs), offering unique advantages in terms of low power consumption. However, the security aspect of SNNs is under-explored compared to their ANN counterparts. As the increasing reliance on AI systems comes with uni

Orhan Donmez

In this paper, we investigate for the first time the quasi-periodic oscillations (QPOs) that arise as a result of halting the Bondi-Hoyle-Lyttleton (BHL) accretion mechanism around a Kerr black hole. Unlike previous studies, which focused on shock cones or perturbed tori, we show that stopping BHL accretion leads to the formation of a new plasma structure. O

Martin Kostelník, Karel Beneš, Michal Hradiš

Logical page segmentation is an important step in document analysis, enabling better semantic representations, information retrieval, and text understanding. Previous approaches define logical segmentation either through text or geometric objects, relying on OCR or precise geometry. To avoid the need for OCR, we define the task purely as segmentation in the

Robert J. Banks, Natasha Feinstein, Roopayan Ghosh, Sougato Bose

In certain scenarios, quantum annealing can be made more efficient by additional $XX$ interactions. It has been shown that the additional interactions can reduce the scaling of perturbative crossings. In traditional annealing devices these couplings do not exist natively. In this work, we develop two gadgets to achieve this: a three-body gadget that requires

L. Konstantinou, S. Akras, J. Garcia-Rojas, K. Bouvis

The physico-chemical properties of the planetary nebula (PN) NGC 3242 are investigated in both 1D and 2D, using Integral Field Unit (IFU) data. This PN has a complex morphology with multiple shells and contains a pair of structures with a lower degree of ionization compared to the main nebular components. These structures are known as low ionization structur

Alejandro Ariza-Casabona, Nikos Kanakaris, Daniele Malitesta

Relational deep learning (RDL) settles among the most exciting advances in machine learning for relational databases, leveraging the representational power of message passing graph neural networks (GNNs) to derive useful knowledge and run predicting tasks on tables connected through primary-to-foreign key links. The RDL paradigm has been successfully applied

Eduard Allakhverdov, Elizaveta Goncharova, Andrey Kuznetsov

Vision encoders typically generate a large number of visual tokens, providing information-rich representations but significantly increasing computational demands. This raises the question of whether all generated tokens are equally valuable or if some of them can be discarded to reduce computational costs without compromising quality. In this paper, we intro

Viet Thanh Duy Nguyen, Truong-Son Hy

Proteins are complex biomolecules that play a central role in various biological processes, making them critical targets for breakthroughs in molecular biology, medical research, and drug discovery. Deciphering their intricate, hierarchical structures, and diverse functions is essential for advancing our understanding of life at the molecular level. Protein

G. Stage, A. Borjigin, J. Ding, M. Davis

Characterization of strip and pixel AC-LGAD devices with both laser TCT and probe station (IV/CV) will be shown on AC-LGADs irradiated with 1 MeV reactor neutrons at JSI/Ljubljana and with 400~MeV protons at FNAL ITA to fluences from 1e13~$n_{eq}/cm^2$ to a few times 1e15~$n_{eq}/cm^2$. This study was conducted within the scope of the ePIC detector time of f

Elizabeth Gasparim

This is a contribution to the Special Volume in Celebration of the 70th Birthday of Edoardo Ballico. First, I describe how some results of Ballico on moduli of vector bundles and categories coherent sheaves were useful for solving problems in a variety of areas: Homological Mirror Symmetry, symplectic geometry, Hodge theory, mathematical physics, noncommutat

E. Aldo Arroyo

We study a real tachyon vacuum solution in cubic superstring field theory that avoids square roots and phantom terms. Using this new solution, we evaluate the vacuum energy and obtain a result consistent with Sen's conjecture. Additionally, we demonstrate that the equation of motion, when contracted with the solution itself, is satisfied.

Maxime Delmas, Magdalena Wysocka, Danilo Gusicuma, André Freitas

The discovery of novel antibiotics is critical to address the growing antimicrobial resistance (AMR). However, pharmaceutical industries face high costs (over $1 billion), long timelines, and a high failure rate, worsened by the rediscovery of known compounds. We propose an LLM-based pipeline that acts as an alarm system, detecting prior evidence of antibiot

José Mário da Silva, Alejandro Pozas-Kerstjens, Fernando Parisio

Nonlocal correlations created in networks with multiple independent sources enable surprising phenomena in quantum information and quantum foundations. The presence of independent sources, however, makes the analysis of network nonlocality challenging, and even in the simplest nontrivial scenarios a complete characterization is lacking. In this work we study

Hanxiao Wang, Biao Zhang, Weize Quan, Dong-Ming Yan

This paper propose iFlame, a novel transformer-based network architecture for mesh generation. While attention-based models have demonstrated remarkable performance in mesh generation, their quadratic computational complexity limits scalability, particularly for high-resolution 3D data. Conversely, linear attention mechanisms offer lower computational costs

Yasuyuki Ihara

The Set Partitioning Problem is a combinatorial optimization problem with wide-ranging applicability, used to model various real-world tasks such as facility location and crew scheduling. However, real-world applications often require solving large-scale instances that involve hundreds of thousands of variables. Although the conventional Column Generation me

Zhicheng Chen

Fundamental logic was introduced by Wesley Holliday (2023) to unify intuitionistic logic and quantum logic from a proof-theoretic perspective, capturing the logic determined solely by the introduction and elimination rules of connectives $\neg$, $\wedge$, $\vee$. This paper incorporates strict implication -- standard in intuitionistic logic and a significant

Dongming Mei, Kunming Dong, Austin Warren, Sanjay Bhattarai

Virtual Majorana neutrinos are indispensable for neutrinoless double-beta (0$\nu\beta\beta$) decay. In this study, we demonstrate that the overlap of the virtual Majorana neutrino wavefunction, predominantly composed of a right-handed antineutrino component with a strongly suppressed left-handed component (with amplitude proportional to the effective Majoran

Sara Gliorio, Emanuele Berti, Andrea Maselli, Nicholas Speeney

The gravitational wave (GW) signals from extreme mass-ratio inspirals (EMRIs), a key target for the Laser Interferometer Space Antenna (LISA), will be affected in the presence of dark matter (DM) halos. In this paper we explore whether the effects of DM are detectable by LISA within a fully relativistic framework. We model the massive EMRI component as a non

Luca Di Luzio, Giacomo Landini, Federico Mescia, Vasja Susič

We explore a class of axion models where an accidental $\mathrm{U}(1)$ Peccei-Quinn (PQ) symmetry automatically emerges from the interplay of vertical (grand-unified) and horizontal (flavor) gauge symmetries. We study a specific Pati-Salam realization in detail, and aim to generalize the conclusions. We show that our specific model offers protection from PQ-

G. F. Paraschos, V. Mpisketzis

Understanding the creation of relativistic jets originating from active galactic nuclei, require a thorough understanding of the accompanying plasma instabilities. Our high sensitivity, high resolution, global very long baseline interferometry observations of the jet in the radio galaxy 3C 84 enable us to study its inner morphology, which resembles a thread-

André T. Cesário, Tiago Debarba

The processing of quantum information is limited by fundamental physical constraints on how information can be encoded, transmitted, and extracted. In particular, the non-orthogonality of quantum states limits their distinguishability, and thermodynamic constraints, including the energetic cost of state preparation and quantum operations, further restrict th

Dhrubajyoti Ghosh, Samhita Pal, Michael Lutz, Sheng Luo

Predicting the risk of clinical progression from cognitively normal (CN) status to mild cognitive impairment (MCI) or Alzheimer's disease (AD) is critical for early intervention in Alzheimer's disease (AD). Traditional survival models often fail to capture complex longitudinal biomarker patterns associated with disease progression. We propose an ensemble sur

Qianye Wu, Chengxuan Xia, Sixuan Tian

The rapid growth of e-commerce has led to an overwhelming volume of customer feedback, from product reviews to service interactions. Extracting meaningful insights from this data is crucial for businesses aiming to improve customer satisfaction and optimize decision-making. This paper presents an AI-driven sentiment analysis system designed specifically for

Wanyi Chen, Mary Cummings

Missing data is prevalent in tabular machine learning (ML) models, and different missing data treatment methods can significantly affect ML model training results. However, little is known about how ML researchers and engineers choose missing data treatment methods and what factors affect their choices. To this end, we conducted a survey of 70 ML researchers

Long Chen, Richard Lai, Shashi Pandey, Dapeng Cui

Antiferromagnets exhibiting the anomalous Hall effect represent a fascinating convergence of magnetism, topology, and electronic structure. Identifying antiferromagnets with large and tunable anomalous Hall effects is crucial for the development of spintronic applications. Here, we report a strain-tunable anomalous Hall plateau in CoNb$_3$S$_6$, which is a p

Qasim Khan, Anthony Suen, Bao Quoc Tang

The effect of multiplicative noise to the Turing instability of the Brusselator system is investigated. We show that when the noise acts on both of the concentrations with the same intensities, then the Turing instability is suppressed provided that the intensities are sufficiently large. This aligns with the stabilizing effect of multiplicative noise in par

Joshua Arroyo, Zachary Hamaker, Graham Hawkes, Jianping Pan

We study Type C $K$-Stanley symmetric functions, which are $K$-theoretic extensions of the Type C Stanley symmetric functions. They are indexed by signed permutations and can be used to enumerate reduced words via their expansion into Schur $Q$-functions, which are indexed by strict partitions. A combinatorial description of the Schur $Q$- coefficients is gi

Thomas Kreutz, Max Mühlhäuser, Alejandro Sanchez Guinea

Realistic crowd simulations are essential for immersive virtual environments, relying on both individual behaviors (microscopic dynamics) and overall crowd patterns (macroscopic characteristics). While recent data-driven methods like deep reinforcement learning improve microscopic realism, they often overlook critical macroscopic features such as crowd densi

Dimitris Boskos, Jorge Cortés, Sonia Martínez

This paper considers non-smooth optimization problems where we seek to minimize the pointwise maximum of a continuously parameterized family of functions. Since the objective function is given as the solution to a maximization problem, neither its values nor its gradients are available in closed form, which calls for approximation. Our approach hinges upon e

S. S. Larsen, A. M. N. Ferguson, J. M. Howell, F. Annibali

We examine the star cluster populations in the three nearby galaxies IC 342, NGC 2403, and Holmberg II, observed as part of the Euclid Early Release Observations programme. Our main focus is on old globular clusters (GCs), for which the wide field-of-view and excellent image quality of Euclid offer substantial advantages over previous work. For IC 342 this i

A. V. Sarantsev, E. Klempt, K. V. Nikonov, T. Seifen

The decays of $N^*$ and $\Delta^*$ resonances into $N\rho$, $\Delta\pi$ and $N\sigma$ final states are studied in a coupled-channel analysis of data on pion- and photo-induced reactions. Improvements in the fit were observed when new resonance contributions were introduced. Frequencies for the intermediate isobars $\Delta(1232)\pi$, $N\rho$, $N\sigma$ are re

Yinchi Zhou, Huidong Xie, Menghua Xia, Qiong Liu

Low-count positron emission tomography (LCPET) imaging can reduce patients' exposure to radiation but often suffers from increased image noise and reduced lesion detectability, necessitating effective denoising techniques. Diffusion models have shown promise in LCPET denoising for recovering degraded image quality. However, training such models requires larg

Shijie Gao, Nicola Bezzo

Despite the remarkable acceleration of robotic development through advanced simulation technology, robotic applications are often subject to performance reductions in real-world deployment due to the inherent discrepancy between simulation and reality, often referred to as the "sim-to-real gap". This gap arises from factors like model inaccuracies, environme

Nikolai Kriukov, Krzysztof Dȩbicki, Michel Mandjes

We consider a queueing network operating under a strictly upper-triangular routing matrix with per column at most one non-negative entry. The root node is fed by a Gaussian process with stationary increments. Our aim is to characterize the distribution of the multivariate stationary workload process under a specific scaling of the queue's service rates. In t

Saugat Pandey, Alvitta Ottley

The increasing integration of Visual Language Models (VLMs) into visualization systems demands a comprehensive understanding of their visual interpretation capabilities and constraints. While existing research has examined individual models, systematic comparisons of VLMs' visualization literacy remain unexplored. We bridge this gap through a rigorous, first

Amar Kumar, Sujeet Chaudhary, Sharat Chandra

Using the plane-wave pseudopotential method within the framework of density functional theory, Co$_2$MnAl (100), (110), and (111) surfaces with different atomic terminations have been studied in the context of some key spintronics properties, viz., surface energy, half-metallicity, magnetization, and magnetic anisotropy. The present study reveals that the Mn

Dana Cohen-Bar, Daniel Cohen-Or, Gal Chechik, Yoni Kasten

As 3D content creation continues to grow, transferring semantic textures between 3D meshes remains a significant challenge in computer graphics. While recent methods leverage text-to-image diffusion models for texturing, they often struggle to preserve the appearance of the source texture during texture transfer. We present \ourmethod, a novel approach that

Julian Ziegler, Patrick Frenzel, Mirco Fuchs

This work concerns itself with the task of reconstructing all edges of an arbitrary 3D wire-frame model projected to an image plane. We explore a bottom-up part-wise procedure undertaken by an RL agent to segment and reconstruct these 2D multipart objects. The environment's state is represented as a four-colour image, where different colours correspond to ba

Moshiur Rahman Tonmoy, Md. Mithun Hossain, Nilanjan Dey, M. F. Mridha

Plant diseases significantly threaten global food security by reducing crop yields and undermining agricultural sustainability. AI-driven automated classification has emerged as a promising solution, with deep learning models demonstrating impressive performance in plant disease identification. However, deploying these models on mobile and edge devices remai

Jing-Ze Ma, Rüdiger Pakmor, Stephen Justham, Selma E. de Mink

Radiation is crucial not only for observing astrophysical objects, but also for transporting energy and momentum. However, accurate on-the-fly radiation transport in astrophysical simulations is challenging and computationally expensive. Here we introduce AREPO-IDORT (Implicit Discrete Ordinates Radiation Transport), a scheme coupled to the explicit magnetoh

Todor Manev

We introduce the concept of F-decomposable systems, well-ordered inverse systems of Hausdorff compacta with fully closed bonding mappings. A continuous mapping between Hausdorff compacta is called fully closed if the intersection of the images of any two closed disjoint subsets is finite. We give a characterization of such systems in terms of a property of t

Senol Gulgonul

Digital system design lectures are mandatory in the electrical and electronics engineering curriculum. Besides HDL simulators and viewers, FPGA boards are necessary for the real implementation of HDL, which were previously costly for students. With the emergence of low-cost FPGA boards, the use of take-home labs is increasing. The COVID-19 pandemic has furth

Deepak A. Suresh, F. Robicheaux

We investigate the photon statistics of light emitted from a system of collectively interacting dipoles in the low-intensity regime, incorporating double-excitation states to capture beyond-single-excitation effects. By analyzing the eigenstates of the double-excitation manifold, we establish their connection to the accessible single-excitation eigenmodes an

Zhongtang Luo

Scientific publications significantly impact academic-related decisions in computer science, where top-tier conferences are particularly influential. However, efforts required to produce a publication differ drastically across various subfields. While existing citation-based studies compare venues within areas, cross-area comparisons remain challenging due t

Michele Fiori, Gabriele Civitarese, Priyankar Choudhary, Claudio Bettini

Explainable Artificial Intelligence (XAI) aims to uncover the inner reasoning of machine learning models. In IoT systems, XAI improves the transparency of models processing sensor data from multiple heterogeneous devices, ensuring end-users understand and trust their outputs. Among the many applications, XAI has also been applied to sensor-based Activities o

Shomik Jain, Margaret Wang, Kathleen Creel, Ashia Wilson

The Rashomon set of equally-good models promises less discriminatory algorithms, reduced outcome homogenization, and fairer decisions through model ensembles or reconciliation. However, we argue from the perspective of allocation multiplicity that these promises may remain unfulfilled. When there are more qualified candidates than resources available, many d

Marco Bonetti, Philipp Rendler, William J. Torres Bobadilla

We compute two-loop electroweak corrections to double Higgs boson production in gluon fusion mediated by light quarks in a fully analytical way. We determine a basis of master integrals satisfying canonical differential equations in $\mathrm{d}\log$ form, enhanced by subsequent rotations to remove unnecessary functions that do not appear in the analytic expr

Dougal Davis, Ruijie Yang

In this paper, we prove a Beilinson-type formula for the V-filtration of Kashiwara and Malgrange on a complex mixed Hodge module, using Hodge filtrations on the localization. Our formula expresses the V-filtration as the filtered D-module underlying a pro-mixed Hodge module. We apply this to the theory of higher multiplier and Hodge ideals. Our first result

Joseph A. Dessi, Evgenios T. A. Kakariadis, Ioannis Apollon Paraskevas

We revisit the results of Kim, and of Katsoulis and Ramsey concerning hyperrigidity for non-degenerate C*-correspondences. We show that the tensor algebra is hyperrigid, if and only if Katsura's ideal acts non-degenerately, if and only if Katsura's ideal acts non-degenerately under any representation. This gives a positive answer to the question of Katsoulis

Malav Patel, Kento Tomita, Koki Ho

Recently, renewed interest in cislunar space spurred by private and public organizations has driven research for future infrastructure in the region. As Earth-Moon traffic increases amidst a growing space economy, monitoring architectures supporting this traffic must also develop. These are likely to be realized as constellations of patrol satellites surveyi

Xiaoran Zhang, Byung-Woo Hong, Hyoungseob Park, Daniel H. Pak

We propose a model-agnostic, progressive test-time energy adaptation approach for medical image segmentation. Maintaining model performance across diverse medical datasets is challenging, as distribution shifts arise from inconsistent imaging protocols and patient variations. Unlike domain adaptation methods that require multiple passes through target data -

Clotilde Djuikem, Josué Tchouanti

Maize production in sub-Saharan Africa faces significant challenges due to the maize stalk borer (Busseola fusca), a major pest that causes substantial yield losses. Chemical control methods have raised concerns about environmental impact and pest resistance, making biological control a promising alternative. In this study, we develop a multi-seasonal mathem

Maria de Lourdes M. Silva, André L. C. Mendonça, Eduardo R. D. Neto, Iago C. Chaves

Computer manufacturers typically offer platforms for users to report faults. However, there remains a significant gap in these platforms' ability to effectively utilize textual reports, which impedes users from describing their issues in their own words. In this context, Natural Language Processing (NLP) offers a promising solution, by enabling the analysis

Ashten Akemoto, Frances Zhu

Many environments, such as unvisited planetary surfaces and oceanic regions, remain unexplored due to a lack of prior knowledge. Autonomous vehicles must sample upon arrival, process data, and either transmit findings to a teleoperator or decide where to explore next. Teleoperation is suboptimal, as human intuition lacks mathematical guarantees for optimalit

Garrett Perkins, Benjamin Macht, Lucas Ritzdorf, Tristan Running Crane

Trusted Execution Environments (TEEs) have emerged at the forefront of edge computing to combat the lack of trust between system components. Field Programmable Gate Arrays (FPGAs) are commonly used as edge computers but were not created with security as a primary consideration. Thus, FPGA-based edge computers are increasingly the target of cyberattacks. We a

Katja Schwarz, Denys Rozumnyi, Samuel Rota Bulò, Lorenzo Porzi

We introduce a recipe for generating immersive 3D worlds from a single image by framing the task as an in-context learning problem for 2D inpainting models. This approach requires minimal training and uses existing generative models. Our process involves two steps: generating coherent panoramas using a pre-trained diffusion model and lifting these into 3D wi

Ning Bao, Grant N. Remmen

We demonstrate that wormholes must be entangled regardless of asymptotic boundary conditions. Assuming black hole complementarity, we argue that traversable wormholes instantiate entanglement-assisted quantum channels and that this entanglement must be present between the stretched horizons as an initial condition prior to traversability. This result demonst

Sofía Rojas-Ruiz, Emmanuel Momjian, Frederick B. Davies, Eduardo Bañados

We present the first study dedicated to measuring the timescales for black hole accretion and jet launch in a quasar at the edge of Reionization, PSO J352.4034-15.3373 at z = 5.832 $\pm$ 0.001. Previous work presented evidence of the strong radio synchrotron emission from the jet affecting the host galaxy dust-dominated continuum emission at $\nu_{\rm rest}=

Luke T. Parker, João M. Mendonça, Hannah Diamond-Lowe, Jayne L. Birkby

The sub-Neptune planets have no solar system analogues, and their low bulk densities suggest thick atmospheres containing degenerate quantities of volatiles and H/He, surrounding cores of unknown sizes. Measurements of their atmospheric composition can help break these degeneracies, but many previous studies at low spectral resolution have largely been hinde

Dilys Ruan, Alyson M. Brooks, Akaxia Cruz, Annika H. G. Peter

The baryonic Tully Fisher relation (bTFR) provides an empirical connection between baryonic mass and dynamical mass (measured by the maximum rotation velocity) for galaxies. Due to the impact of baryonic feedback in the shallower potential wells of dwarf galaxies, the bTFR is predicted to turn down at low masses from the extrapolated power-law relation at hi

Milena Crnogorčević, Carlos Blanco, Tim Linden

The extreme energy of the KM3-230213A event could transform our understanding of the most energetic sources in the Universe. However, it also reveals an inconsistency between the KM3NeT detection and strong IceCube constraints on the ultra-high energy neutrino flux. The most congruous explanation for the KM3NeT and IceCube data requires KM3-230213A to be pro

Charles J. Law, Romane Le Gal, Yoshihide Yamato, Ke Zhang

The abundance and distribution of sulfur-bearing molecules in protoplanetary disks directly influences the composition and potential habitability of nascent planets in addition to providing powerful probes of the physical gas conditions in the disks themselves. Here, we present new and archival ALMA and SMA observations of CS and H$_2$CS, and their C$^{34}$S

Praveen Pai, Fan Zhang

We reveal strong and weak inequalities relating two fundamental macroscopic quantum geometric quantities, the quantum distance and Berry phase, for closed paths in the Hilbert space of wavefunctions. We recount the role of quantum geometry in various quantum problems and show that our findings place new bounds on important physical quantities.

Timothy Cohen, Ipak Fadakar, Andrew Gomes, Alexander Monin

The weak gravity conjecture has been invoked to conjecture that the dimensions of charged operators in a CFT should obey a superadditivity relation (sometimes referred to as convexity). In this paper, we study superadditivity of the operator spectrum in theories expanded about the semi-classical saddle point that dominates correlators of large charge operato

Christian W. Bauer

Most observables at particle colliders involve physics at a wide variety of distance scales. Due to asymptotic freedom of the strong interaction, the physics at short distances can be calculated reliably using perturbative techniques, while long distance physics is non-perturbative in nature. Factorization theorems separate the contributions from different s

Miguel Correia, Mathieu Giroux, Sebastian Mizera

We introduce SOFIA, a Mathematica package that automatizes the computation of singularities of Feynman integrals, based on new theoretical understanding of their analytic structure. Given a Feynman diagram, SOFIA generates a list of potential singularities along with a candidate symbol alphabet. The package also provides a comprehensive set of tools for anal

Anna de Graaff, Hans-Walter Rix, Rohan P. Naidu, Ivo Labbe

The origin of the rest-optical emission of compact, red, high-redshift sources known as `little red dots' (LRDs) poses a major puzzle. If interpreted as starlight, it would imply that LRDs would constitute the densest stellar systems in the Universe. However, alternative models suggest active galactic nuclei (AGN) may instead power the rest-optical continuum

Shi Chen, Damian van de Heisteeg, Cumrun Vafa

Non-geometric flux vacua have recently been revisited, leading to the remarkable discovery of isolated 4D ${\mathcal N}=1$ supersymmetric Minkowski vacua. These constructions rely on the non-renormalization of the superpotential, which is supported by heuristic arguments. Given the significance of verifying the existence of these isolated M-theory-like vacua

Ioannis D. Gialamas, Kyriakos Tamvakis

We combine the ghost-free bimetric theory of gravity with the concept of local Weyl invariance, realized in the framework of Einstein-Cartan gravity. The gravitational sector, characterized by two independent metrics and two independent connections, is coupled to a scalar field that can in principle develop a non-vanishing expectation value through radiative

Fridrik Freyr Gautason, Jesse van Muiden

We study quantum M2-branes in holographic backgrounds and show that their moduli spaces of zero-modes are localised according to an R-symmetry Killing vector. We discuss the relation with recent results in the context of equivariant localisation in gauged supergravity and argue its origin within M-theory path integrals expanded in saddle points over M2-brane

Rohan P. Naidu, Jorryt Matthee, Harley Katz, Anna de Graaff

The physical processes that led to the formation of billion solar mass black holes within the first 700 million years of cosmic time remain a puzzle. Several theoretical scenarios have been proposed to seed and rapidly grow black holes, but direct observations of these mechanisms remain elusive. Here we present a source 660 million years after the Big Bang t

V. Rusakov, D. Watson, G. P. Nikopoulos, G. Brammer

The James Webb Space Telescope (JWST) has uncovered many compact galaxies at high redshift with broad hydrogen and helium lines, including the enigmatic population of little red dots (LRDs). The nature of these galaxies is debated and is attributed to supermassive black holes (SMBHs) or intense star formation. They exhibit unusual properties for SMBHs, such

Yuqing Wang, Zhijie Lin, Yao Teng, Yuanzhi Zhu

Autoregressive visual generation models typically rely on tokenizers to compress images into tokens that can be predicted sequentially. A fundamental dilemma exists in token representation: discrete tokens enable straightforward modeling with standard cross-entropy loss, but suffer from information loss and tokenizer training instability; continuous tokens b

Xiaoyang Wu, Daniel DeTone, Duncan Frost, Tianwei Shen

In this paper, we question whether we have a reliable self-supervised point cloud model that can be used for diverse 3D tasks via simple linear probing, even with limited data and minimal computation. We find that existing 3D self-supervised learning approaches fall short when evaluated on representation quality through linear probing. We hypothesize that th

Ruyi Xu, Guangxuan Xiao, Haofeng Huang, Junxian Guo

Long-Context Transformer Models (LCTMs) are vital for real-world applications but suffer high computational costs due to attention's quadratic complexity. Block-sparse attention mitigates this by focusing computation on critical regions, yet existing methods struggle with balancing accuracy and efficiency due to costly block importance measurements. In this

Jonathan J. Heckman, Max Hübner, Chitraang Murdia

Topological symmetry operators of holographic large $N$ CFT$_D$'s are dual to dynamical branes in the gravity dual AdS$_{D+1}$. We use this correspondence to establish a dictionary between thermal expectation values of symmetry operators in the Euclidean CFT$_D$ and the evaluation of gravitational saddles in the presence of a dynamical brane. Expectation val

Keyan Chen, Chenyang Liu, Bowen Chen, Wenyuan Li

The advancement of RS technology has enabled high-resolution Earth observation; however, interpreting these images using modern VFMs remains a significant challenge. Unlike object-centric natural images, RS imagery is fundamentally characterized by extreme target sparsity and massive spatial redundancy. Key objects of interest (e.g., ships, vehicles) often o

Zigang Geng, Mengde Xu, Han Hu, Shuyang Gu

This paper proposes a fundamentally new paradigm for image generation through set-based tokenization and distribution modeling. Unlike conventional methods that serialize images into fixed-position latent codes with a uniform compression ratio, we introduce an unordered token set representation to dynamically allocate coding capacity based on regional semant

Xi Liu, Chaoyi Zhou, Nanxuan Zhao, Siyu Huang

Differentiable vector graphics (VGs) are widely used in image vectorization and vector synthesis, while existing representations are costly to optimize and struggle to achieve high-quality rendering results for high-resolution images. This work introduces a new differentiable VG representation, dubbed B\'ezier Splatting, that enables fast yet high-fidelity V

Ron Campos, Ashmal Vayani, Parth Parag Kulkarni, Rohit Gupta

Image geolocalization, in which an AI model traditionally predicts the precise GPS coordinates of an image, is a challenging task with many downstream applications. However, the user cannot utilize the model to further their knowledge beyond the GPS coordinates; the model lacks an understanding of the location and the conversational ability to communicate wi

Yuheng Yuan, Qiuhong Shen, Xingyi Yang, Xinchao Wang

4D Gaussian Splatting (4DGS) has recently gained considerable attention as a method for reconstructing dynamic scenes. Despite achieving superior quality, 4DGS typically requires substantial storage and suffers from slow rendering speed. In this work, we delve into these issues and identify two key sources of temporal redundancy. (Q1) \textbf{Short-Lifespan

Quanhao Li, Zhen Xing, Rui Wang, Hui Zhang

Recent advances in video generation have led to remarkable improvements in visual quality and temporal coherence. Upon this, trajectory-controllable video generation has emerged to enable precise object motion control through explicitly defined spatial paths. However, existing methods struggle with complex object movements and multi-object motion control, re

Paul Engstler, Aleksandar Shtedritski, Iro Laina, Christian Rupprecht

We address the challenge of generating 3D worlds from textual descriptions. We propose SynCity, a training- and optimization-free approach, which leverages the geometric precision of pre-trained 3D generative models and the artistic versatility of 2D image generators to create large, high-quality 3D spaces. While most 3D generative models are object-centric

Yang Sui, Yu-Neng Chuang, Guanchu Wang, Jiamu Zhang

Large Language Models (LLMs) have demonstrated remarkable capabilities in complex tasks. Recent advancements in Large Reasoning Models (LRMs), such as OpenAI o1 and DeepSeek-R1, have further improved performance in System-2 reasoning domains like mathematics and programming by harnessing supervised fine-tuning (SFT) and reinforcement learning (RL) techniques

Liming Jiang, Qing Yan, Yumin Jia, Zichuan Liu

Achieving flexible and high-fidelity identity-preserved image generation remains formidable, particularly with advanced Diffusion Transformers (DiTs) like FLUX. We introduce InfiniteYou (InfU), one of the earliest robust frameworks leveraging DiTs for this task. InfU addresses significant issues of existing methods, such as insufficient identity similarity,

Riccardo Alessandro, Matteo Castagnola, Henrik Koch, Enrico Ronca

Multireference systems are usually challenging to investigate using ab initio methods as they require an accurate description of static electron correlation. The urgency of developing similar approaches is even more pressing when molecules strongly interact with light in quantum-electrodynamics (QED) environments. In fact, in this context, multireference eff

Asaf Yehudai, Lilach Eden, Alan Li, Guy Uziel

LLM-based agents represent a paradigm shift in AI, enabling autonomous systems to plan, reason, and use tools while interacting with dynamic environments. This paper provides the first comprehensive survey of evaluation methods for these increasingly capable agents. We analyze the field of agent evaluation across five perspectives: (1) Core LLM capabilities

Justin Khoury, Meng-Xiang Lin, Mark Trodden

We show that a simple coupling between dark energy and dark matter can simultaneously address two distinct hints at new physics coming from cosmological observations. The first is the recent evidence from the DESI project and supernovae observations that the dark energy equation of state~$w$ is evolving over cosmic time from an earlier value that is~$<-1$ to

Christian Kroer, Dominik Peters

Lindahl equilibrium is a solution concept for allocating a fixed budget across several divisible public goods. It always lies in the weak core, meaning that the equilibrium allocation satisfies desirable stability and proportional fairness properties. We consider a model where agents have separable linear utility functions over the public goods, and the outp

Xueyan Zou, Yuchen Song, Ri-Zhao Qiu, Xuanbin Peng

We present 3D Spatial MultiModal Memory (M3), a multimodal memory system designed to retain information about medium-sized static scenes through video sources for visual perception. By integrating 3D Gaussian Splatting techniques with foundation models, M3 builds a multimodal memory capable of rendering feature representations across granularities, encompass