March 2025 arXiv papers — page 57

David Wierichs, Maxwell West, Roy T. Forestano, M. Cerezo

Recursive Cartan decompositions (CDs) provide a way to exactly factorize quantum circuits into smaller components, making them a central tool for unitary synthesis. Here we present a detailed overview of recursive CDs, elucidating their mathematical structure, demonstrating their algorithmic utility, and implementing them numerically at large scales. We adap

Joheen Chakraborty, Erin Kara, Riccardo Arcodia, Johannes Buchner

Quasi-periodic eruptions (QPEs) are recurring soft X-ray transients emerging from the vicinity of supermassive black holes (SMBHs) in nearby, low-mass galaxy nuclei; about ten QPE hosts have been identified thus far. Here we report the \textit{NICER} discovery of QPEs in the optically-selected Tidal Disruption Event (TDE) and Extreme Coronal Line Emitter (EC

Taeksoo Kim, Hanbyul Joo

We present a target-aware video diffusion model that generates videos from an input image, in which an actor interacts with a specified target while performing a desired action. The target is defined by a segmentation mask, and the action is described through a text prompt. Our key motivation is to incorporate target awareness into video generation, enabling

Gelo Noel M. Tabia, Alex Yueh-Ting Shih, Jin-Yuan Zheng, Yeong-Cherng Liang

In quantum information, device-independent protocols offer a new approach to information processing tasks, making minimal assumptions about the devices used. Typically, since these protocols draw conclusions directly from the data collected in a meaningful Bell test, the no-signaling conditions, and often even Born's rule for local measurements, are taken as

Ruixiao Dong, Mengde Xu, Zigang Geng, Li Li

Current generative models, such as autoregressive and diffusion approaches, decompose high-dimensional data distribution learning into a series of simpler subtasks. However, inherent conflicts arise during the joint optimization of these subtasks, and existing solutions fail to resolve such conflicts without sacrificing efficiency or scalability. We propose

Jae Joong Lee, Bedrich Benes, Raymond A. Yeh

Amodal segmentation aims to predict segmentation masks for both the visible and occluded regions of an object. Most existing works formulate this as a supervised learning problem, requiring manually annotated amodal masks or synthetic training data. Consequently, their performance depends on the quality of the datasets, which often lack diversity and scale.

Jihang Zhu, Yang-Zhi Chou, Yi Huang, Sankar Das Sarma

We theoretically predict the in-plane magnetic field-induced orbital Fulde-Ferrell-Larkin-Ovchinnikov (FFLO) superconducting states in twisted WSe$_2$ homobilayers (tWSe$_2$), focusing on its dependence on layer polarization and Fermi surface geometry. For unpolarized layers, finite-momentum pairing emerges only at low temperatures and above a critical field

Aether Team, Haoyi Zhu, Yifan Wang, Jianjun Zhou

The integration of geometric reconstruction and generative modeling remains a critical challenge in developing AI systems capable of human-like spatial reasoning. This paper proposes Aether, a unified framework that enables geometry-aware reasoning in world models by jointly optimizing three core capabilities: (1) 4D dynamic reconstruction, (2) action-condit

Karim Knaebel, Kadir Yilmaz, Daan de Geus, Alexander Hermans

Vision foundation models (VFMs) trained on large-scale image datasets provide high-quality features that have significantly advanced 2D visual recognition. However, their potential in 3D scene segmentation remains largely untapped, despite the common availability of 2D images alongside 3D point cloud datasets. While significant research has been dedicated to

Mingze Xu, Mingfei Gao, Shiyu Li, Jiasen Lu

We introduce SlowFast-LLaVA-1.5 (abbreviated as SF-LLaVA-1.5), a family of video large language models (LLMs) offering a token-efficient solution for long-form video understanding. We incorporate the two-stream SlowFast mechanism into a streamlined training pipeline, and perform joint video-image training on a carefully curated data mixture of only publicly

Fangfu Liu, Hanyang Wang, Yimo Cai, Kaiyan Zhang

With the scale capability of increasing training data, model size, and computational cost, video generation has achieved impressive results in digital creation, enabling users to express creativity across various domains. Recently, researchers in Large Language Models (LLMs) have expanded the scaling to test-time, which can significantly improve LLM performa

Hongru Cai, Yongqi Li, Ruifeng Yuan, Wenjie Wang

Generative retrieval reformulates retrieval as an autoregressive generation task, where large language models (LLMs) generate target documents directly from a query. As a novel paradigm, the mechanisms that underpin its performance and scalability remain largely unexplored. We systematically investigate training and inference scaling laws in generative retri

Ye Tian, Xin Xia, Yuxi Ren, Shanchuan Lin

Diffusion models have demonstrated remarkable capabilities in visual content generation but remain challenging to deploy due to their high computational cost during inference. This computational burden primarily arises from the quadratic complexity of self-attention with respect to image or video resolution. While existing acceleration methods often compromi

Aaron Miller, Joachim Favre, Zoë Holmes, Özlem Salehi

We introduce Majorana Propagation, an algorithmic framework for the classical simulation of Fermionic circuits. Inspired by Pauli Propagation, Majorana Propagation operates by applying successive truncations throughout the Heisenberg evolution of the observable. We identify monomial length as an effective truncation strategy for typical, unstructured circuit

Shenyuan Gao, Siyuan Zhou, Yilun Du, Jun Zhang

World models aim to learn action-controlled future prediction and have proven essential for the development of intelligent agents. However, most existing world models rely heavily on substantial action-labeled data and costly training, making it challenging to adapt to novel environments with heterogeneous actions through limited interactions. This limitatio

Lingyan Ran, Lidong Wang, Guangcong Wang, Peng Wang

The task of translating visible-to-infrared images (V2IR) is inherently challenging due to three main obstacles: 1) achieving semantic-aware translation, 2) managing the diverse wavelength spectrum in infrared imagery, and 3) the scarcity of comprehensive infrared datasets. Current leading methods tend to treat V2IR as a conventional image-to-image synthesis

Bohua Li, Joel Meyers, Paul R. Shapiro

We investigate the kination-amplified inflationary gravitational-wave background (GWB) interpretation of the signal recently reported by various pulsar timing array (PTA) experiments. Kination is a post-inflationary phase in the expansion history dominated by the kinetic energy of some scalar field, characterized by a stiff equation of state $w=1$. Within th

E. Medina-Guerra, I. V. Gornyi, Yuval Gefen

We investigate phase transitions in a non-Hermitian Su-Schrieffer-Heeger (SSH) model with an imaginary chemical potential via Krylov spread complexity and Krylov fidelity. The spread witnesses the $\mathcal{PT}$-transition for the non-Hermitian Bogoliubov vacuum of the SSH Hamiltonian, where the spectrum goes from purely real to complex (oscillatory dynamics

Yifei Feng, Mingxin Yang, Shuhui Yang, Sheng Zhang

Painting textures for existing geometries is a critical yet labor-intensive process in 3D asset generation. Recent advancements in text-to-image (T2I) models have led to significant progress in texture generation. Most existing research approaches this task by first generating images in 2D spaces using image diffusion models, followed by a texture baking pro

Rodrigo Alonso

This letter presents the solution to a counting problem, to the best of our knowledge not known in full generality, which can be mapped to both (i) ways to lace a $N$-holes-per-side shoe with $\ell$ shoestrings and (ii) a sum over the indexes of $2N$-tensors made of symmetric 2-tensors. The coefficients that answer this question are then connected to spin $J

Marcus Haberland, Alessandra Buonanno, Jan Steinhoff

We present SEOBNRv5THM, an accurate and fast gravitational-waveform model for quasi-circular, spinning, non-precessing binary neutron stars (BNS) within the effective-one-body (EOB) formalism. It builds on the binary-black-hole approximant SEOBNRv5HM and, compared to its predecessor SEOBNRv4T, it i) incorporates recent high-order post-Newtonian results in th

Enrico Pallotta, Sina Mokhtarzadeh Azar, Shuai Li, Olga Zatsarynna

Predicting future video frames is essential for decision-making systems, yet RGB frames alone often lack the information needed to fully capture the underlying complexities of the real world. To address this limitation, we propose a multi-modal framework for Synchronous Video Prediction (SyncVP) that incorporates complementary data modalities, enhancing the

Luis A. Anchordoqui, Akitaka Ariga, Tomoko Ariga, Alan J. Barr

The Forward Physics Facility (FPF) is a proposal developed to exploit the unique scientific potential made possible by the intense hadron beams produced in the far-forward direction at the high luminosity LHC (HL-LHC). Housed in a well-shielded cavern 627 m from the LHC interactions, the facility will enable a broad and deep scientific programme which will g

Wontae Kim, Matias Vestberg

This paper concerns elliptic systems of $p$-Laplace type with complex valued coefficient and source term. We extend the real valued theory of the elliptic $p$-Laplace equation to the complex valued case. We establish the existence and uniqueness of solutions to the Dirichlet problem and prove the Schauder estimate in the case of H\"older continuous coefficie

Tobias Spohn, Nicolas Staudenmaier, Philipp J. Vetter, Timo Joas

Nuclear magnetic resonance spectroscopy with solid-state spin sensors is a promising pathway for the detection of nuclear spins at the micro- and nanoscale. Although many nanoscale experiments rely on a single sensor spin for the detection of the signal, leveraging spin ensembles can enhance sensitivity, particularly in cases in which the signal merely origi

Brian Bartoldson, Siddarth Venkatraman, James Diffenderfer, Moksh Jain

Reinforcement learning (RL) is a critical component of large language model (LLM) post-training. However, on-policy algorithms used for post-training are not naturally robust to a diversified content of experience replay buffers, which asynchronous off-policy actors can efficiently populate in parallel to training. We propose efficiently learning on such off

Arun Reddy, Alexander Martin, Eugene Yang, Andrew Yates

In this work, we tackle the problem of text-to-video retrieval (T2VR). Inspired by the success of late interaction techniques in text-document, text-image, and text-video retrieval, our approach, Video-ColBERT, introduces a simple and efficient mechanism for fine-grained similarity assessment between queries and videos. Video-ColBERT is built upon 3 main com

Sabah Shahnoor Anis, Devin M. Kellis, Kris Ford Kaigler, Marlene A. Wilson

Analyzing ultrasonic vocalizations (USVs) is crucial for understanding rodents' affective states and social behaviors, but the manual analysis is time-consuming and prone to errors. Automated USV detection systems have been developed to address these challenges. Yet, these systems often rely on machine learning and fail to generalize effectively to new datas

S. Ramakrishna, S. Fritzsche

Recent studies have highlighted the frequent applications of structured light modes in optically pumped atomic magnetometers. In this work, we theoretically explore how a Poincar\'e beam probes an optically polarized atomic medium. Specifically, we consider atoms polarized by a plane wave with linear polarization, immersed in a constant external magnetic fie

T. Olney, V. Naniyil, V. Boyer, J. Goldwin

Cavity quantum electrodynamics systems using atoms in resonant optical cavities are central elements of many applications such as quantum networks and quantum-enhanced sensing. We present a novel experimental setup that achieves strong spatial mode-matching between a resonant mode of a triangular ring resonator and an ensemble of ultracold atoms trapped in a

Shota Nakagawa, Yuichiro Nakai, Yu-Cheng Qiu, Masaki Yamada

Recent cosmological observations have revealed growing tensions with the standard $\Lambda$CDM model, including indications of isotropic cosmic birefringence and deviations from $w = -1$ in the dark energy equation of state, as suggested by DESI and supernova measurements. In this paper, we point out that such deviations can arise even from a subdominant ene

Meng Cao, Pengfei Hu, Yingyao Wang, Jihao Gu

Recent advancements in Large Video Language Models (LVLMs) have highlighted their potential for multi-modal understanding, yet evaluating their factual grounding in videos remains a critical unsolved challenge. To address this gap, we introduce Video SimpleQA, the first comprehensive benchmark tailored for factuality evaluation in video contexts. Our work di

Ednaldo L. B. Junior, José Tarciso S. S. Junior, Francisco S. N. Lobo, Manuel E. Rodrigues

The framework of General Relativity (GR) has recently been expanded through the introduction of Cotton Gravity (CG), a theoretical extension proposed by J. Harada. This modified approach integrates the Cotton tensor into the gravitational field equations, naturally encompassing all conventional GR solutions while allowing the cosmological constant to emerge

Zhigang Bao, Giorgio Cipolloni, László Erdős, Joscha Henheik

We prove the Paquette-Zeitouni law of fractional logarithm (LFL) for the extreme eigenvalues [arXiv:1505.05627] in full generality, and thereby verify a conjecture from [arXiv:1505.05627]. Our result holds for any Wigner minor process and both symmetry classes, in particular for the GOE minor process, while [arXiv:1505.05627] and the recent full resolution o

Yifan Zhang, Mark Fornace, Michael Lindsey

In this work, we develop deterministic and random sketching-based algorithms for two types of tensor interpolative decompositions (ID): the core interpolative decomposition (CoreID, also known as the structure-preserving HOSVD) and the satellite interpolative decomposition (SatID, also known as the HOID or CURT). We adopt a new adaptive approach that leads t

Viktoriya Bardenova, Neal Bushaw, Brent Cody, Paul Fay

The Wiener index of a vertex coloring of a graph is defined to be the sum of all pairwise geodesic distances between vertices of the same color. We provide characterizations of vertex colorings of paths and cycles whose Wiener index is as large as possible over various natural collections. Along the way we establish a connection between the majorization orde

Fabiano F. Santos, Ankit Anand

In this work, we explore the entanglement entropy equipped with the $\kappa$-algebra. This entanglement entropy is computed through the geometric setup as performed by Hartman-Maldacena, which, in their prescription, finds that the entropy grows linearly in time. In our case, we show that the $\kappa$-algebra embedding provides a richer scenario where the th

Steven V Sam, Keller VandeBogert

We utilize an isomorphism between the character rings of the odd orthogonal group and the orthosymplectic supergroup to understand equivariant positivity properties of the type B quadric hypersurface ring. Our main result establishes a well-behaved functorial construction of Schur modules ``with respect to'' the quadric hypersurface ring, an essential fact u

Pasquale Ambrosio, Giovanni Cupini, Elvira Mascolo

We establish the local boundedness of the local minimizers $u:\Omega\rightarrow\mathbb{R}^{m}$ of non-uniformly elliptic integrals of the form $\int_{\Omega}f(x,Dv)\,dx$, where $\Omega$ is a bounded open subset of $\mathbb{R}^{n}$ ($n\geq2)$ and the integrand satisfies anisotropic growth conditions of the type \[ \sum_{i=1}^{n}\lambda_{i}(x)|\xi_{i}|^{p_{i}}

Audun Myers, Bill Kay, Iliana Alvarez, Michael Hughes

This work presents a novel framework for time series analysis using entropic measures based on the kernel density estimate (KDE) of the time series' Takens' embeddings. Using this framework we introduce two distinct analytical tools: (1) a multi-scale KDE entropy metric, denoted as $\Delta\text{KE}$, which quantifies the evolution of time series complexity a

Adam Samorzewski, Adrian Kliks

In this paper, we conduct an in-depth analysis of radio signal propagation characteristics within the urban environment of Poznan (Poland). The study specifically addresses the deployment of a 5th generation (5G NR - New Radio) Radio Access Network (RAN), which comprises 8 strategically positioned Base Stations (BSs). These base stations are configured with

Emanuel Fallas Hernández, Sergio Martínez Alonso, Alejandro Romero, Jose A. Becerra Permuy

One of the challenges of open-ended learning in robots is the need to autonomously discover goals and learn skills to achieve them. However, when in lifelong learning settings, it is always desirable to generate sub-goals with their associated skills, without relying on explicit reward, as steppingstones to a goal. This allows sub-goals and skills to be reus

Sebastian Hell, Julian Späthe, Morten Førre, Robert Klas

We present coincidence measurements of two-photon double-ionization (TPDI) of argon driven by femtosecond pulses tunable around 26.5 eV photon energy, which are obtained from a high-harmonic generation source. The measured photoelectron spectra are interpreted with regard to three TPDI mechanisms. Theoretical predictions are obtained by an approximate model

Sourish Das, Sudeep Shukla, Alka Yadav, Anirban Chakraborti

Air pollution poses significant health and environmental challenges, particularly in rapidly urbanizing regions. Delhi-National Capital Region experiences air pollution episodes due to complex interactions between anthropogenic emissions and meteorological conditions. Understanding the causal drivers of key pollutants such as $PM_{2.5}$ and ground $O_3$ is c

Zeqing Jin, Zhaocheng Liu, Nagi Elabbasi, Zachary Ulissi

Designing a new varifocal architecture in AR glasses poses significant challenges due to the complex interplay of multiple physics disciplines, including innovated piezo-electric material, solid mechanics, electrostatics, and optics. Traditional design methods, which treat each physics separately, are insufficient for this problem as they fail to establish t

Filippo Girardi, Aadil Oufkir, Bartosz Regula, Marco Tomamichel

The sphere-packing bound quantifies the error exponent for noisy channel coding for rates above a critical value. Here, we study the zero-rate limit of the sphere-packing bound and show that it has an intriguing single-letter form, which we call the umlaut information of the channel, inspired by the lautum information introduced by Palomar and Verd\'u. Unlik

Erick Gordillo Herrerías

In this work, we extend the celebrated result of Avila--Forni~\cite{avila2007weak} on the weak mixing property of interval exchange transformations to the setting of linear involutions, which naturally arise from the study of vertical foliations on half-translation surfaces. Using recent advances on the Kontsevich--Zorich cocycle for quadratic differentials~

Akhiad Bercovich, Mohammad Dabbah, Omri Puny, Ido Galil

We introduce FFN Fusion, an architectural optimization technique that reduces sequential computation in large language models by identifying and exploiting natural opportunities for parallelization. Our key insight is that sequences of Feed-Forward Network (FFN) layers, particularly those remaining after the removal of specific attention layers, can often be

S. Lamperski, L. B. Bhuiyan, C. W. Outhwaite, R. Gorniak

The excess entropy of restricted primitive model electrolytes is calculated using a potential based approach through the symmetric Poisson-Boltzmann and the modified Poisson-Boltzmann theories. The theories are utilized in conjunction with a statistical thermodynamics equation that is shown to be equivalent to thermodynamic integration. Electrolyte systems h

Charles Schwartz

This paper reports improvements and new results for a well developed theory of tachyons (faster-than-light particles) and follows the suggestion that neutrinos - especially the low energy neutrinos in the Cosmic Neutrino Background - may be tachyons. Starting with an application of the Thomas-Fermi method for studying a mass of particles obeying the Pauli ex

Samantha I. Davis, Raju Valivarthi, Andrew Cameron, Cristian Pena

We demonstrate conditional entanglement swapping, i.e. teleportation of entanglement, between time-bin qubits at the telecommunication wavelength of 1536.4 nm with high fidelity of 87\%. Our system is deployable, utilizing modular, off-the-shelf, fiber-coupled, and electrically controlled components such as electro-optic modulators. It leverages the precise

Alessio Cela, Carl Lian

The classical Brill-Noether theorem states that a map from a general curve to a projective space deforms in a family of expected dimension as long as its image does not lie in any hyperplane. In this note, we observe, as a direct consequence of standard results on Severi varieties, an analogous statement for maps from a general curve to any smooth, projectiv

Moussa Kassem Sbeyti, Nadja Klein, Azarm Nowzad, Fikret Sivrikaya

Semi-supervised object detection (SSOD) based on pseudo-labeling significantly reduces dependence on large labeled datasets by effectively leveraging both labeled and unlabeled data. However, real-world applications of SSOD often face critical challenges, including class imbalance, label noise, and labeling errors. We present an in-depth analysis of SSOD und

Teodor Iličin, Rok Žitko

We propose a set of variational wavefunctions for the sub-gap spin-doublet and spin-singlet eigenstates of the particle-hole symmetric superconducting Anderson impurity model. The wavefunctions include up to two Bogoliubov quasiparticles in the continuum which is necessary to correctly capture the weak-coupling asymptotics in all parameter regimes. The eigen

M. Cruz Sanchez, V. Trejos Montoya, O. Pizio

We report the temperature, pressure and composition dependence of some basic properties of model liquid water-methanol mixtures. For this purpose the isobaric-isothermal molecular dynamics computer simulations are employed. Our principal focus is on the united atom non-polarizable UAM-I-EW model for methanol which was recently parametrized the paper by Garci

Danish Nisar, Saif Khan Mohammed, Ronny Hadani, Ananthanarayanan Chockalingam

Linear time-varying (LTV) systems model radar scenes where each reflector/target applies a delay, Doppler shift and complex amplitude scaling to a transmitted waveform. The receiver processes the received signal using the transmitted signal as a reference. The self-ambiguity function of the transmitted signal captures the cross-correlation of delay and Doppl

Michele Dallachiesa, Antonio Pitasi, David Pinger, Josh Goodbody

Many real-world applications are increasingly incorporating automated decision-making, driven by the widespread adoption of ML/AI inference for planning and guidance. This study examines the growing need for verifiable computing in autonomous decision-making. We formalize the problem of verifiable computing and introduce a sampling-based protocol that is sig

Hannes Jakob, Maxwell Levine

We solve two long-standing open problems regarding the combinatorics of $\aleph_{\omega+1}$. We answer a question of Shelah by showing that it is consistent for any $n\geq 1$ that $\mathsf{GCH}$ holds and there is a stationary set of points of cofinality $\aleph_n$ which is not in the approachability ideal. As a corollary, we obtain a model where the notions

Thomas Chabal, Shizhe Chen, Jean Ponce, Cordelia Schmid

This paper addresses the problem of reconstructing a scene online at the level of objects given an RGB-D video sequence. While current object-aware neural implicit representations hold promise, they are limited in online reconstruction efficiency and shape completion. Our main contributions to alleviate the above limitations are twofold. First, we propose a

Łukasz Delong, Selim Gatti, Mario V. Wüthrich

A statistical model is said to be calibrated if the resulting mean estimates perfectly match the true means of the underlying responses. Aiming for calibration is often not achievable in practice as one has to deal with finite samples of noisy observations. A weaker notion of calibration is auto-calibration. An auto-calibrated model satisfies that the expect

M. V. Ushcats, L. A. Bulavin, S. Yu. Ushcats, Zh. Yu. Burunina

Resent achievements in statistical theory, namely, a possibility to reproduce almost unlimited Mayer's activity series based on the information about their convergence radius, on the one hand, and generalization of the lattice statistics by eliminating the simplification of nearest-neighbor interactions, on the other hand, have allowed accurate quantitative

Ivan V. Vlassiouk, Yueh-Chun Wu, Alexander Puretzky, Liangbo Liang

Recently, numerous techniques have been reported for generating optically active defects in exfoliated hexagonal boron nitride (hBN), which hold transformative potential for quantum photonic devices. However, achieving on-demand generation of desirable defect types in scalable hBN films remains a significant challenge. Here, we demonstrate that formation of

Weihao Zeng, Yuzhen Huang, Qian Liu, Wei Liu

DeepSeek-R1 has shown that long chain-of-thought (CoT) reasoning can naturally emerge through a simple reinforcement learning (RL) framework with rule-based rewards, where the training may directly start from the base models-a paradigm referred to as zero RL training. Most recent efforts to reproduce zero RL training have primarily focused on the Qwen2.5 mod

Zhexuan Wang, Yutong Wang, Xuebo Liu, Liang Ding

Multi-agent systems (MAS) based on large language models (LLMs) have demonstrated significant potential in collaborative problem-solving. However, they still face substantial challenges of low communication efficiency and suboptimal task performance, making the careful design of the agents' communication topologies particularly important. Inspired by the man

Jake Doliskani, Morteza Mirzaei, Ali Mousavi

We propose a public-key quantum money scheme based on group actions and the Hartley transform. Our scheme adapts the quantum money scheme of Zhandry (2024), replacing the Fourier transform with the Hartley transform. This substitution ensures the banknotes have real amplitudes rather than complex amplitudes, which could offer both computational and theoretic

Luigi Scurto, Helena Pais, Marco Antonelli, Francesca Gulminelli

The direct Urca (dUrca) process is a key mechanism driving rapid neutrino cooling in neutron stars, with its baryon density activation threshold determined by the microscopic model for nuclear matter. Understanding how nuclear interactions shape the dUrca threshold is essential for interpreting neutron star thermal evolution, particularly in light of recent

Zhengcong Yin, Daniel W. Goldberg, Binbin Lin, Bing Zhou

Geocoding systems are widely used in both scientific research for spatial analysis and everyday life through location-based services. The quality of geocoded data significantly impacts subsequent processes and applications, underscoring the need for next-generation systems. In response to this demand, this review first characterizes the technical requirement

V. Alfradique, C. R. Bom, T. Castro

The advent of multimessenger cosmology, marked by the detection of GW170817, demonstrated that standard sirens are a valuable cosmological probe. In the absence of an electromagnetic counterpart identification, gravitational waves carry valuable information through the dark siren approach, where the source redshift is estimated using galaxy catalogs of poten

Weichen Fan, Amber Yijia Zheng, Raymond A. Yeh, Ziwei Liu

Classifier-Free Guidance (CFG) is a widely adopted technique in diffusion/flow models to improve image fidelity and controllability. In this work, we first analytically study the effect of CFG on flow matching models trained on Gaussian mixtures where the ground-truth flow can be derived. We observe that in the early stages of training, when the flow estimat

Hosho Katsura, Dirk Schuricht

We consider the valence bond solid (VBS) state built of singlet pairs of fundamental representations and projected onto adjoint representations of the exceptional Lie group G$_2$. The two-point correlation function in the VBS state is non-vanishing only for nearest neighbours, but possesses finite string order. We construct a parent Hamiltonian for the VBS s

Joaquín A. Barraza-Jorquera, Matthias R. Schreiber, Diogo Belloni

Angular momentum loss through magnetic braking drives the spin-down of low-mass stars and the orbital evolution of various close binary systems. Current theories for magnetic braking, often calibrated for specific types of systems, predict angular momentum loss rates that differ by several orders of magnitude. A unified prescription would provide valuable co

Savas Ozkan, Andrea Maracani, Hyowon Kim, Sijun Cho

In recent years, vision transformers with text decoder have demonstrated remarkable performance on Scene Text Recognition (STR) due to their ability to capture long-range dependencies and contextual relationships with high learning capacity. However, the computational and memory demands of these models are significant, limiting their deployment in resource-c

Lukas Fuchs, Sabrina Weber, Jialin Men, Niklas Eiermann

Agglomeration is an industrially relevant process for the production of bulk materials in which the product properties depend on the morphology of the agglomerates, e.g., on the distribution of size and shape descriptors. Thus, accurate characterization and control of agglomerate morphologies is essential to ensure high and consistent product quality. This p

K. Haydukivska

The influence of monomer-monomer interactions on the scaling exponents and shape characteristics of a single polymer chain in a selective solvent is investigated using Langevin dynamics simulations. By systematically increasing the temperature of the solution, the effects of interactions between blocks on the conformational properties of the chain are explor

Hyeonggon Ryu, Seongyu Kim, Joon Son Chung, Arda Senocak

We present a unified model capable of simultaneously grounding both spoken language and non-speech sounds within a visual scene, addressing key limitations in current audio-visual grounding models. Existing approaches are typically limited to handling either speech or non-speech sounds independently, or at best, together but sequentially without mixing. This

Christoph Brunken, Sebastien Boyer, Mustafa Omar, Martin Maarand

Coarse-grained (CG) force field methods for molecular systems are a crucial tool to simulate large biological macromolecules and are therefore essential for characterisations of biomolecular systems. While state-of-the-art deep learning (DL)-based models for all-atom force fields have improved immensely over recent years, we observe and analyse significant l

Andrey Galichin, Alexey Dontsov, Polina Druzhinina, Anton Razzhigaev

Recent LLMs like DeepSeek-R1 have demonstrated state-of-the-art performance by integrating deep thinking and complex reasoning during generation. However, the internal mechanisms behind these reasoning processes remain unexplored. We observe reasoning LLMs consistently use vocabulary associated with human reasoning processes. We hypothesize these words corre

Alessio Catanzaro, Subodh Patil, Diego Garlaschelli

Network Models with couplings between link pairs are the simplest models for a class of networks with Higher Order interactions. In this paper we give an analytic, general solution to this family of Random Graph Models extending previous results obtained for specific interaction structures. We use the Hubbard-Stratonovich transform to show that such higher o

Jun Zhan, Congcong Le, Xianxin Wu, Jiangping Hu

The recent discovery of high-temperature superconductivity in pressurized bilayer nickelate La$_3$Ni$_2$O$_7$ and its thin films has generated significant interest in uncovering the underlying pairing mechanisms and correlated electronic states. While earlier theoretical studies have mainly focused on onsite Coulomb interactions, the role of nonlocal Coulomb

Chao Wu

In this paper, we study the singularity formation phenomenon of the 1D model of Electron Magnetohydrodynamics (EMHD). we will construct a solution whose $C^3$-norm blows up in finite time. In the end, we will show that the solution is in $C^{\infty}(\mathbb{R}\backslash \{0\})\cap C^{3,s}(\mathbb{R})\cap H^3(\mathbb{R})$ and is not asymptotically self-simila

Nicholas Steyn, Kris V. Parag, Robin N. Thompson, Christl A. Donnelly

Renewal models are widely used in statistical epidemiology as semi-mechanistic models of disease transmission. While primarily used for estimating the instantaneous reproduction number, they can also be used for generating projections, estimating elimination probabilities, modelling the effect of interventions, and more. We demonstrate how simple sequential

Runze Cheng, Yao Sun, Lan Zhang, Lei Feng

With the significant advances in generative AI (GAI) and the proliferation of mobile devices, providing high-quality AI-generated content (AIGC) services via wireless networks is becoming the future direction. However, the primary challenges of AIGC service delivery in wireless networks lie in unstable channels, limited bandwidth resources, and unevenly dist

Moein Sorkhei, Emir Konuk, Jingyu Guo, Chanjuan Meng

Self-supervised adaptation (SSA) improves foundation model transfer to medical domains but is computationally prohibitive. Although parameter efficient fine-tuning methods such as LoRA have been explored for supervised adaptation, their effectiveness for SSA remains unknown. In this work, we introduce efficient self-supervised adaptation (ESSA), a framework

Yanda Chen, Gongwei Chen, Miao Zhang, Weili Guan

Dataset distillation (DD) excels in synthesizing a small number of images per class (IPC) but struggles to maintain its effectiveness in high-IPC settings. Recent works on dataset distillation demonstrate that combining distilled and real data can mitigate the effectiveness decay. However, our analysis of the combination paradigm reveals that the current one

Yuhang Wang, Hanwei Guo, Sizhe Wang, Long Qian

Model Predictive Control (MPC) has been demonstrated to be effective in continuous control tasks. When a world model and a value function are available, planning a sequence of actions ahead of time leads to a better policy. Existing methods typically obtain the value function and the corresponding policy in a model-free manner. However, we find that such an

Noemi David, Matt Jacobs, Inwon Kim

In this paper we study singular limits of congestion-averse growth models, connecting different models describing the effect of congestion. These models arise in particular in the context of tissue growth. The main ingredient of our analysis is a family of energy evolution equations and their dissipation structures, which are novel and of independent interes

Rui Xie, Asad Ul Haq, Linsen Ma, Yunhua Fang

The efficiency of Large Language Model~(LLM) inference is often constrained by substantial memory bandwidth and capacity demands. Existing techniques, such as pruning, quantization, and mixture of experts/depth, reduce memory capacity and/or bandwidth consumption at the cost of slight degradation in inference quality. This paper introduces a design solution

Konstantinos Rallis, Ioannis Liliopoulos, Evangelos Tsipas, Georgios D. Varsamis

The technology of Quantum Computing (QC) is continuously evolving, as researchers explore new technologies and the public gains access to quantum computers with an increasing number of qubits. In addition, the research community and industry are increasingly interested in the potential use, application, and contribution of QCs to large-scale problems in the

Juan Guillermo Garrido

This paper is devoted to study a characterization of (strong) local maximal monotonicity in terms of a property involving the graphical derivative of a set-valued mapping defined on a Hilbert space. As a consequence, a second-order characterization of variational convexity is provided without the assumption of subdifferential continuity.

Yangjun Ruan, Neil Band, Chris J. Maddison, Tatsunori Hashimoto

Compute scaling for language model (LM) pretraining has outpaced the growth of human-written texts, leading to concerns that data will become the bottleneck to LM scaling. To continue scaling pretraining in this data-constrained regime, we propose that explicitly modeling and inferring the \emph{latent thoughts} that underlie the text generation process can

Junlan Chen, Kexin Zhang, Daifeng Li, Yangyang Feng

The emergence of large language models offers new possibilities for structured exploration of scientific knowledge. Rather than viewing scientific discovery as isolated ideas or content, we propose a structured approach that emphasizes the role of method combinations in shaping disruptive insights. Specifically, we investigate how knowledge unit--especially

Kaïs Ammari, Alessandro Duca, Romain Joly, Kévin Le Balc'h

In this paper, we introduce a novel concept called the Graph Geometric Control Condition (GGCC). It turns out to be a simple, geometric rewriting of many of the frameworks in which the controllability of PDEs on graphs has been studied. We prove that (GGCC) is a necessary and sufficient condition for the exact controllability of the wave equation on metric g

Luisa Beghin, Nikolai Leonenko, Jayme Vaz

We introduce and study renewal processes defined by means of extensions of the standard relaxation equation through ``stretched" non-local operators (of order $\alpha$ and with parameter $\gamma$). In a first case we obtain a generalization of the fractional Poisson process, which displays either infinite or finite expected waiting times between arrivals, de

DeShin Hwa, Tobias Holmes, Klaus Drechsler

While CNNs were long considered state of the art for image processing, the introduction of Transformer architectures has challenged this position. While achieving excellent results in image classification and segmentation, Transformers remain inherently reliant on large training datasets and remain computationally expensive. A newly introduced Transformer de

Suat Icli, Rangana Banerjee Chaudhuri, Elena Jordan, Fatemeh Salahshoori

Electro-optic modulators are fundamental components in atomic physics experiments, including trapped-ion systems used in precision metrology and quantum computing. To enable scalable photonic integration, we design and analyze an integrated photonic electro-optic phase modulator and switch at 411 nm for ytterbium ($Yb^{+}$) ions using aluminum nitride (AlN)

Zunnan Xu, Zhentao Yu, Zixiang Zhou, Jun Zhou

We introduce HunyuanPortrait, a diffusion-based condition control method that employs implicit representations for highly controllable and lifelike portrait animation. Given a single portrait image as an appearance reference and video clips as driving templates, HunyuanPortrait can animate the character in the reference image by the facial expression and hea

Wasim Abbas, Salaki Reynaldo Joshua, Asim Abbas, Je-Hoon Lee

Encryption is crucial for securing sensitive data during transmission over networks. Various encryption techniques exist, such as AES, DES, and RC4, with AES being the most renowned algorithm. We proposed methodology that enables users to encrypt text messages for secure transmission over cellular networks. This approach utilizes the AES algorithm following

Eldar Knar

In this work, a new concept called Vector Dissipation of Randomness (VDR) is developed and formalized. It describes the mechanism by which complex multicomponent systems transition from chaos to order through the filtering of random directions, accumulation of information in the environment, and self-organization of agents. VDR explains how individual random

Sheikh Muhammad Farjad, Sandeep Reddy Patllola, Yonas Kassa, George Grispos

Structural Health Monitoring (SHM) plays a crucial role in maintaining aging and critical infrastructure, supporting applications such as smart cities and digital twinning. These applications demand machine learning models capable of processing large volumes of real-time sensor data at the network edge. However, existing approaches often neglect the challeng

Daniel Lepe-Soltero, Thierry Artières, Anaïs Baudot, Paul Villoutreix

An important objective in computational biology is the efficient integration of multi-omics data. The task of integration comes with challenges: multi-omics data are most often unpaired (requiring diagonal integration), partially labeled with information about biological conditions, and in some situations such as rare diseases, only very small datasets are a