Search

The exploration of ultrafast phenomena is a frontier of condensed matter research, where the interplay of theory, computation, and experiment is unveiling new opportunities for understanding and engineering quantum materials. With the advent of advanced experimental techniques and computational tools, it has become possible to probe and manipulate nonequilibrium processes at unprecedented temporal and spatial resolutions, providing insights into the dynamical behavior of matter under extreme conditions. These capabilities have the potential to revolutionize fields ranging from optoelectronics and quantum information to catalysis and energy storage. This Roadmap captures the collective progress and vision of leading researchers, addressing challenges and opportunities across key areas of ultrafast science. Contributions in this Roadmap span the development of ab initio methods for time-resolved spectroscopy, the dynamics of driven correlated systems, the engineering of materials in optical cavities, and the adoption of FAIR principles for data sharing and analysis. Together, these efforts highlight the interdisciplinary nature of ultrafast research and its reliance on cutting-edge methodologies, including quantum electrodynamical density-functional theory, correlated electronic structure methods, nonequilibrium Green's function approaches, quantum and ab initio simulations.

We present VIRAC2, a catalogue of positions, proper motions, parallaxes and $Z$, $Y$, $J$, $H$, and $K_s$ near-infrared photometric time series of 545 346 537 unique stars. The catalogue is based on a point spread function fitting reduction of nearly a decade of VISTA VVV and VVVX images, which cover $560~{\rm deg}^2$ of the Southern Galactic plane and bulge. The catalogue is complete at the $>90$ per cent level for $11

The Moving-Target Traveling Salesman Problem (MT-TSP) aims to find a shortest path for an agent that starts at a stationary depot, visits a set of moving targets exactly once, each within one of their respective time windows, and then returns to the depot. In this paper, we introduce a new Mixed-Integer Conic Program (MICP) formulation that finds the optimum for the Multi-Agent Moving-Target Traveling Salesman Problem (MA-MT-TSP), a generalization of the MT-TSP involving multiple agents. We obtain our formulation by first restating the current state-of-the-art MICP formulation for MA-MT-TSP as a Mixed-Integer Nonlinear Nonconvex Program, and then reformulating it as a new MICP. We present computational results to demonstrate the performance of our approach. The results show that our formulation significantly outperforms the state-of-the-art, with up to a two-order-of-magnitude reduction in runtime, and up to over 90% tighter optimality gap., Comment: 7 pages, 3 figures

Quantum algorithms can potentially solve a handful of problems more efficiently than their classical counterparts. In that context, it has been discussed that Markov chains problems could be solved significantly faster using quantum computing. Indeed, previous work suggests that quantum computers could accelerate sampling from the stationary distribution of reversible Markov chains. However, in practice, certain physical processes of interest are nonreversible in the probabilistic sense and reversible Markov chains can sometimes be replaced by more efficient nonreversible chains targeting the same stationary distribution. This study uses modern quantum algorithmic techniques and Markov chain theory to sample from the stationary distribution of nonreversible Markov chains with faster worst-case runtime and without requiring the stationary distribution to be computed up to a multiplicative constant. Such an up-to-exponential quantum speedup goes beyond the predicted quadratic quantum acceleration for reversible chains and is likely to have a decisive impact on many applications ranging from statistics and machine learning to computational modeling in physics, chemistry, biology and finance.

Network coding enhances performance in network communications and distributed storage by increasing throughput and robustness while reducing latency. Batched Sparse (BATS) codes are a class of capacity-achieving network codes, but their practical applications are hindered by their structure, computational intensity, and power demands of finite field operations. Most literature focuses on algorithmic-level techniques to improve coding efficiency. Optimization with an algorithm/hardware co-designing approach has long been neglected. Leveraging the unique structure of BATS codes, we first present CS-BATS, a hardware-friendly variant. Next we propose a simple but effective bounded-value generator, to reduce the size of a finite field multiplier by up to 70%. Finally, we report on a scalable and resource-efficient FPGA-based network coding accelerator that achieves a throughput of 27 Gbps, a speedup of more than 300 over software.

Optimizing charged-particle track reconstruction algorithms is crucial for efficient event reconstruction in Large Hadron Collider (LHC) experiments due to their significant computational demands. Existing track reconstruction algorithms have been adapted to run on massively parallel coprocessors, such as graphics processing units (GPUs), to reduce processing time. Nevertheless, challenges remain in fully harnessing the computational capacity of coprocessors in a scalable and non-disruptive manner. This paper proposes an inference-as-a-service approach for particle tracking in high energy physics experiments. To evaluate the efficacy of this approach, two distinct tracking algorithms are tested: Patatrack, a rule-based algorithm, and Exa$.$TrkX, a machine learning-based algorithm. The as-a-service implementations show enhanced GPU utilization and can process requests from multiple CPU cores concurrently without increasing per-request latency. The impact of data transfer is minimal and insignificant compared to running on local coprocessors. This approach greatly improves the computational efficiency of charged particle tracking, providing a solution to the computing challenges anticipated in the High-Luminosity LHC era., Comment: 19 pages, 8 figures, submitted to JINST

Incoherent Diffraction Imaging - IDI - is a diffraction-based imaging technique that has been recently proposed to exploit the partial coherence of incoherently scattered light to retrieve structural information from the scattering centers. Similar to the stellar intensity interferometry of Hanbury Brown and Twiss, the signal builds up on the second-order spatial correlations of the emitted light. The complex spatial distribution of the target is thereby encoded in the spatial intensity fluctuations of the scattered light. The first experimental realisations of this imaging technique have been realised using the fluorescence excited by an ultra-short X-ray pulse at Free Electron Laser (FEL) facilities. Here, we propose an alternative set-up based on a table-top Pseudo-Thermal Light Source. This set-up allows us to explore IDI under a wide range of physically relevant conditions as well as to benchmark numerical and analytical models currently used to determine the imaging capabilities of this technique.

This study investigates the efficacy of machine learning models for predicting house rental prices in Ghana, addressing the need for accurate and accessible housing market information. Utilising a comprehensive dataset of rental listings, we trained and evaluated various models, including CatBoost, XGBoost, and Random Forest. CatBoost emerged as the best-performing model, achieving an $R^2$ of 0.876, demonstrating its ability to effectively capture complex relationships within the housing market. Feature importance analysis revealed that location-based features, number of bedrooms, bathrooms, and furnishing status are key drivers of rental prices. Our findings provide valuable insights for stakeholders, including real estate professionals, investors, and policymakers, while also highlighting opportunities for future research, such as incorporating temporal data and exploring regional variations., Comment: 13 pages, 8 figures, 2 tables

As AI systems become more capable, widely deployed, and increasingly autonomous in critical areas such as cybersecurity, biological research, and healthcare, ensuring their safety and alignment with human values is paramount. Machine unlearning -- the ability to selectively forget or suppress specific types of knowledge -- has shown promise for privacy and data removal tasks, which has been the primary focus of existing research. More recently, its potential application to AI safety has gained attention. In this paper, we identify key limitations that prevent unlearning from serving as a comprehensive solution for AI safety, particularly in managing dual-use knowledge in sensitive domains like cybersecurity and chemical, biological, radiological, and nuclear (CBRN) safety. In these contexts, information can be both beneficial and harmful, and models may combine seemingly harmless information for harmful purposes -- unlearning this information could strongly affect beneficial uses. We provide an overview of inherent constraints and open problems, including the broader side effects of unlearning dangerous knowledge, as well as previously unexplored tensions between unlearning and existing safety mechanisms. Finally, we investigate challenges related to evaluation, robustness, and the preservation of safety features during unlearning. By mapping these limitations and open challenges, we aim to guide future research toward realistic applications of unlearning within a broader AI safety framework, acknowledging its limitations and highlighting areas where alternative approaches may be required.

Current and upcoming generations of visible-shortwave infrared (VSWIR) imaging spectrometers promise unprecedented capacity to quantify Earth System processes across the globe. However, reliable cloud screening remains a fundamental challenge for these instruments, where traditional spatial and temporal approaches are limited by cloud variability and limited temporal coverage. The Spectroscopic Transformer (SpecTf) addresses these challenges with a spectroscopy-specific deep learning architecture that performs cloud detection using only spectral information (no spatial or temporal data are required). By treating spectral measurements as sequences rather than image channels, SpecTf learns fundamental physical relationships without relying on spatial context. Our experiments demonstrate that SpecTf significantly outperforms the current baseline approach implemented for the EMIT instrument, and performs comparably with other machine learning methods with orders of magnitude fewer learned parameters. Critically, we demonstrate SpecTf's inherent interpretability through its attention mechanism, revealing physically meaningful spectral features the model has learned. Finally, we present SpecTf's potential for cross-instrument generalization by applying it to a different instrument on a different platform without modifications, opening the door to instrument agnostic data driven algorithms for future imaging spectroscopy tasks., Comment: 23 pages, 5 figures, in review. Code repository: https://github.com/emit-sds/SpecTf

Technical and environmental noise in ground-based laser interferometers designed for gravitational-wave observations like Advanced LIGO, Advanced Virgo and KAGRA, can manifest as narrow (<1Hz) or broadband ($10'$s or even $100'$s of Hz) spectral lines and features in the instruments' strain amplitude spectral density. When the sources of this noise cannot be identified or removed, in cases where there are witness sensors sensitive to this noise source, denoising of the gravitational-wave strain channel can be performed in software, enabling recovery of instrument sensitivity over affected frequency bands. This noise hunting and removal process can be particularly challenging due to the wealth of auxiliary channels monitoring the interferometry and the environment and the non-linear couplings that may be present. In this work, we present a comprehensive analysis approach and corresponding cyberinfrastructure to promptly identify and remove noise in software using machine learning techniques. The approach builds on earlier work (referred to as DeepClean) in using machine learning methods for linear and non-linear regression of noise. We demonstrate how this procedure can be operated and optimized in a tandem fashion close to online data taking; it starts off with a coherence monitoring analysis that first singles out and prioritizes witness channels that can then be used by DeepClean. The resulting denoised strain by DeepClean reflects a 1.4\% improvement in the binary neutron star range, which can translate into a 4.3\% increase in the sensitive volume. This cyber infrastructure we refer to as Coherence DeepClean, or CDC, is a significant step toward autonomous operations of noise subtraction for ground-based interferometers., Comment: 14 pages, 9 figures

We utilize the cosmological volume simulation, FIREbox, to investigate how a galaxy's environment influences its size and dark matter content. Our study focuses on approximately 1,200 galaxies (886 central and 332 satellite halos) in the low-mass regime, with stellar masses between $10^6$ to $10^9$ $M_{\odot}$. We analyze the size-mass relation ($r_{50} - M_{\star}$), inner dark matter mass-stellar mass ($M^{50}_{\rm DM} - M_{\star}$) relation, and the halo mass-stellar mass ($M_{\rm halo} - M_{\star}$) relation. At fixed stellar mass, we find the galaxies experiencing stronger tidal influences, indicated by higher Perturbation Indices (PI $>$ 1) are generally larger and have lower masses relative to their counterparts with lower Perturbation Indices (PI $<$ 1). Applying a Random Forest regression model, we show that both the environment (PI) and halo mass ($M_{rm halo}$) are significant predictors of a galaxy's relative size and dark matter content. Notably, because $M_{\rm halo}$ is also strongly affected by the environment, our findings indicate that environmental conditions not only influence galactic sizes and relative inner dark matter content directly, but also indirectly through their impact on halo mass. Our results highlight a critical interplay between environmental factors and halo mass in shaping galaxy properties, affirming the environment as a fundamental driver in galaxy formation and evolution., Comment: 17 pages, 9 figures; submitted to ApJ

Motivated by the applications of secure multiparty computation as a privacy-protecting data analysis tool, and identifying oblivious transfer as one of its main practical enablers, we propose a practical realization of randomized quantum oblivious transfer. By using only symmetric cryptography primitives to implement commitments, we construct computationally-secure randomized oblivious transfer without the need for public-key cryptography or assumptions imposing limitations on the adversarial devices. We show that the protocol is secure under an indistinguishability-based notion of security and demonstrate an experimental implementation to test its real-world performance. Its security and performance are then compared to both quantum and classical alternatives, showing potential advantages over existing solutions based on the noisy storage model and public-key cryptography., Comment: 40 pages, 5 images

In contemporary astronomy and astrophysics (A&A), the integration of high-performance computing (HPC), big data analytics, and artificial intelligence/machine learning (AI/ML) has become essential for advancing research across a wide range of scientific domains. These tools are playing an increasingly pivotal role in accelerating discoveries, simulating complex astrophysical phenomena, and analyzing vast amounts of observational data. For India to maintain and enhance its competitive edge in the global landscape of computational astrophysics and data science, it is crucial for the Indian A&A community to fully embrace these transformative technologies. Despite limited resources, the expanding Indian community has already made significant scientific contributions. However, to remain globally competitive in the coming years, it is vital to establish a robust national framework that provides researchers with reliable access to state-of-the-art computational resources. This system should involve the regular solicitation of computational proposals, which can be assessed by domain experts and HPC specialists, ensuring that high-impact research receives the necessary support. By building such a system, India can cultivate the talent, infrastructure, and collaborative environment necessary to foster world-class research in computational astrophysics and data science., Comment: Accepted for publication in The Journal of Astrophysics and Astronomy. This is an expanded version of one of the chapters in the recently released Vision Document of the Astronomical Society of India

The restaurant industry is currently facing a challenging socio-economic situation caused by the rise of delivery services, inflation, and typically low margins. Often, technological opportunities for process optimization or customer retention are not fully utilized. In our design case study, we investigate which technologies are already being used to improve the customer experience in restaurants and explore a novel new approach to this issue. We designed, implemented, and evaluated a platform with customers and restaurateurs to increase visibility and emotional connection to nearby restaurants through their dishes. Some of our key findings include the enormous potential of combining location-based systems and conversational agents, but also the difficulties in creating content for such platforms. We contribute to the field of Human-Food Interaction by (1) identifying promising design spaces as well as customer and restaurateur requirements for technology in this domain, (2) presenting an innovative design case study to improve the user experience, and (3) exploring the broader implications of our design case study findings for approaching a real-world metaverse., Comment: 32 pages, 8 figures

The metallic delafossites host ultra-high mobility carriers in the bulk, while at their polar surfaces, intrinsic electronic reconstructions stabilise markedly distinct electronic phases, from charge-disproportionated insulators, to Rashba-split heavy-hole gases and ferromagnetic metals. The understanding of these phases has been strongly informed by surface spectroscopic measurements, but previous studies have been complicated by the presence of spatially varying terminations of the material surface. Here, we demonstrate the potential of microscopic-area angle-resolved photoemission to overcome these challenges. Our measurements of the model compound PdCoO$_2$ yield extremely high-quality spectra of the electronic structure, which allows us to place new experimental constraints on the weak electron-phonon coupling in the bulk of PdCoO$_2$, while revealing much stronger interactions at its surfaces. While the CoO$_2$-terminated surface exhibits a conventional weak-coupling behavior, our measurements reveal surprising spectroscopic signatures of polaron formation at the Pd-terminated surface, despite its pronounced metallicity. Together, our findings reveal how mode and symmetry-selective couplings can markedly tune the electron-phonon interactions in a single host material, here opening routes to stabilise surprisingly persistent polaronic quasiparticles., Comment: 17 pages including supplementary information

Control of individual spins at the atomic level holds great promise for miniaturized spintronics, quantum sensing, and quantum information processing. Both single atomic and molecular spin centers are prime candidates for these applications and are often individually addressed and manipulated using scanning tunneling microscopy (STM). In this work, we present a hybrid approach and demonstrate a robust method for self-assembly of magnetic organometallic complexes consisting of individual iron (Fe) atoms and molecules on a silver substrate using STM. We employ two types of molecules, bis(dibenzoylmethane) copper(II) [Cu(dbm)2] and iron phthalocyanine (FePc). We show that in both cases the Fe atoms preferentially attach underneath the benzene ring ligand of the molecules, effectively forming an organometallic half-sandwich arene complex, Fe(C6H6), that is akin to the properties of metallocenes. In both situations, a molecule can be combined with up to two Fe atoms. In addition, we observe a change in the magnetic properties of the attached Fe atoms in scanning tunneling spectroscopy, revealing a distinct Kondo signature at the Fe sites. We explain the latter using density functional theory calculations, and find that the bond formation between the Fe 3d-orbitals and the benzene {\pi}-molecular orbitals creates a favorable situation for Kondo screening of the d_xz- and d_yz-like orbitals. Thus, this work establishes a reliable design principle for forming hybrid organometallic complexes and simultaneous tuning of their atomic spin states., Comment: 16 pages, 5 figures

For over three decades, scanning probe microscopy (SPM) has been a key method for exploring material structures and functionalities at nanometer and often atomic scales in ambient, liquid, and vacuum environments. Historically, SPM applications have predominantly been downstream, with images and spectra serving as a qualitative source of data on the microstructure and properties of materials, and in rare cases of fundamental physical knowledge. However, the fast growing developments in accelerated material synthesis via self-driving labs and established applications such as combinatorial spread libraries are poised to change this paradigm. Rapid synthesis demands matching capabilities to probe structure and functionalities of materials on small scales and with high throughput, which are characteristically inherent to SPM. Here, we overview SPM methods applicable to these emerging applications and emphasize their quantitativeness, focusing on piezoresponse force microscopy, electrochemical strain microscopy, conductive, and surface photovoltage measurements. We discuss the challenges and opportunities ahead, asserting that SPM will play a crucial role in closing the loop from material prediction and synthesis to characterization., Comment: 64 pages, 15 figures

Bregman divergences are a class of distance-like comparison functions which play fundamental roles in optimization, statistics, and information theory. One important property of Bregman divergences is that they cause two useful formulations of information content (in the sense of variability or non-uniformity) in a weighted collection of vectors to agree. In this note, we show that this agreement in fact characterizes the class of Bregman divergences; they are the only divergences which generate this agreement for arbitrary collections of weighted vectors.

In this study, we investigate the application of the New Physics Learning Machine (NPLM) algorithm as an alternative to the standard CWoLa method with Boosted Decision Trees (BDTs), particularly for scenarios with rare signal events. NPLM offers an end-to-end approach to anomaly detection and hypothesis testing by utilizing an in-sample evaluation of a binary classifier to estimate a log-density ratio, which can improve detection performance without prior assumptions on the signal model. We examine two approaches: (1) a end-to-end NPLM application in cases with reliable background modelling and (2) an NPLM-based classifier used for signal selection when accurate background modelling is unavailable, with subsequent performance enhancement through a hyper-test on multiple values of the selection threshold. Our findings show that NPLM-based methods outperform BDT-based approaches in detection performance, particularly in low signal injection scenarios, while significantly reducing epistemic variance due to hyperparameter choices. This work highlights the potential of NPLM for robust resonant anomaly detection in particle physics, setting a foundation for future methods that enhance sensitivity and consistency under signal variability.

Cold-start problem is one of the long-standing challenges in recommender systems, focusing on accurately modeling new or interaction-limited users or items to provide better recommendations. Due to the diversification of internet platforms and the exponential growth of users and items, the importance of cold-start recommendation (CSR) is becoming increasingly evident. At the same time, large language models (LLMs) have achieved tremendous success and possess strong capabilities in modeling user and item information, providing new potential for cold-start recommendations. However, the research community on CSR still lacks a comprehensive review and reflection in this field. Based on this, in this paper, we stand in the context of the era of large language models and provide a comprehensive review and discussion on the roadmap, related literature, and future directions of CSR. Specifically, we have conducted an exploration of the development path of how existing CSR utilizes information, from content features, graph relations, and domain information, to the world knowledge possessed by large language models, aiming to provide new insights for both the research and industrial communities on CSR. Related resources of cold-start recommendations are collected and continuously updated for the community in https://github.com/YuanchenBei/Awesome-Cold-Start-Recommendation.

In light of the advancements in transformer technology, extant research posits the construction of stereo transformers as a potential solution to the binocular stereo matching challenge. However, constrained by the low-rank bottleneck and quadratic complexity of attention mechanisms, stereo transformers still fail to demonstrate sufficient nonlinear expressiveness within a reasonable inference time. The lack of focus on key homonymous points renders the representations of such methods vulnerable to challenging conditions, including reflections and weak textures. Furthermore, a slow computing speed is not conducive to the application. To overcome these difficulties, we present the \textbf{H}adamard \textbf{A}ttention \textbf{R}ecurrent Stereo \textbf{T}ransformer (HART) that incorporates the following components: 1) For faster inference, we present a Hadamard product paradigm for the attention mechanism, achieving linear computational complexity. 2) We designed a Dense Attention Kernel (DAK) to amplify the differences between relevant and irrelevant feature responses. This allows HART to focus on important details. DAK also converts zero elements to non-zero elements to mitigate the reduced expressiveness caused by the low-rank bottleneck. 3) To compensate for the spatial and channel interaction missing in the Hadamard product, we propose MKOI to capture both global and local information through the interleaving of large and small kernel convolutions. Experimental results demonstrate the effectiveness of our HART. In reflective area, HART ranked \textbf{1st} on the KITTI 2012 benchmark among all published methods at the time of submission. Code is available at \url{https://github.com/ZYangChen/HART}.

We analyze overlapping multiplicative Schwarz methods as smoothers in the geometric multigrid solution of two-dimensional anisotropic diffusion problems. For diffusion equations, it is well known that the smoothing properties of point-wise smoothers, such as Gauss--Seidel, rapidly deteriorate as the strength of anisotropy increases. On the other hand, global smoothers based on line smoothing are known to generally provide good smoothing for diffusion problems, independent of the anisotropy strength. A natural question is whether global methods are really necessary to achieve good smoothing in such problems, or whether it can be obtained with locally overlapping block smoothers using sufficiently large blocks and overlap. Through local Fourier analysis and careful numerical experimentation, we show that global methods are indeed necessary to achieve anisotropy-robust smoothing. Specifically, for any fixed block size bounded sufficiently far away from the global domain size, we find that the smoothing properties of overlapping multiplicative Schwarz rapidly deteriorate with increasing anisotropy, irrespective of the amount of overlap between blocks. Moreover, our results indicate that anisotropy-robust smoothing requires blocks of diameter ${\cal O}(\epsilon^{-1/2})$ for anisotropy ratio $\epsilon \in (0,1]$.

It is well known that the number of 123-avoiding and 321-avoiding permutations is the same, and these numbers correspond to the Catalan numbers. However, patterns 123 and 321 are not equidistributed. In the context of mesh patterns, patterns formed by the permutations 123 and 321 with identical shadings are sometimes jointly equidistributed. In this paper, we prove 20 joint equidistributions of mesh patterns 123 and 321 with symmetric shadings, and 36 joint equidistributions of the same patterns with antipodal shadings.

Generative Artificial Intelligence (GenAI, generative AI) has rapidly become available as a tool in scientific research. To explore the use of generative AI in science, we conduct an empirical analysis using OpenAlex. Analyzing GenAI publications and other AI publications from 2017 to 2023, we profile growth patterns, the diffusion of GenAI publications across fields of study, and the geographical spread of scientific research on generative AI. We also investigate team size and international collaborations to explore whether GenAI, as an emerging scientific research area, shows different collaboration patterns compared to other AI technologies. The results indicate that generative AI has experienced rapid growth and increasing presence in scientific publications. The use of GenAI now extends beyond computer science to other scientific research domains. Over the study period, U.S. researchers contributed nearly two-fifths of global GenAI publications. The U.S. is followed by China, with several small and medium-sized advanced economies demonstrating relatively high levels of GenAI deployment in their research publications. Although scientific research overall is becoming increasingly specialized and collaborative, our results suggest that GenAI research groups tend to have slightly smaller team sizes than found in other AI fields. Furthermore, notwithstanding recent geopolitical tensions, GenAI research continues to exhibit levels of international collaboration comparable to other AI technologies., Comment: 26 pages, 4 tables, 1 figures, 1 appendix figure

Dynamic 3D scene representation and novel view synthesis from captured videos are crucial for enabling immersive experiences required by AR/VR and metaverse applications. However, this task is challenging due to the complexity of unconstrained real-world scenes and their temporal dynamics. In this paper, we frame dynamic scenes as a spatio-temporal 4D volume learning problem, offering a native explicit reformulation with minimal assumptions about motion, which serves as a versatile dynamic scene learning framework. Specifically, we represent a target dynamic scene using a collection of 4D Gaussian primitives with explicit geometry and appearance features, dubbed as 4D Gaussian splatting (4DGS). This approach can capture relevant information in space and time by fitting the underlying spatio-temporal volume. Modeling the spacetime as a whole with 4D Gaussians parameterized by anisotropic ellipses that can rotate arbitrarily in space and time, our model can naturally learn view-dependent and time-evolved appearance with 4D spherindrical harmonics. Notably, our 4DGS model is the first solution that supports real-time rendering of high-resolution, photorealistic novel views for complex dynamic scenes. To enhance efficiency, we derive several compact variants that effectively reduce memory footprint and mitigate the risk of overfitting. Extensive experiments validate the superiority of 4DGS in terms of visual quality and efficiency across a range of dynamic scene-related tasks (e.g., novel view synthesis, 4D generation, scene understanding) and scenarios (e.g., single object, indoor scenes, driving environments, synthetic and real data)., Comment: Journal extension of ICLR 2024. arXiv admin note: text overlap with arXiv:2310.10642

A unified thermodynamic algorithm (UTA) is presented for constructing thermodynamically consistent dynamical systems, i.e., systems that have Hamiltonian and dissipative parts that conserve energy while producing entropy. The algorithm is based on the metriplectic 4-bracket given in Morrison and Updike [Phys.\ Rev.\ E 109, 045202 (2024)]. A feature of the UTA is the force-flux relation $\mathbf{J}^\alpha = - L^{\alpha\beta}\, \nabla(\delta H / \delta \xi^\beta)$ for phenomenological coefficients $L^{\alpha\beta}$, Hamiltonian $H$ and dynamical variables $\xi^\beta$. The algorithm is applied to the Navier-Stokes-Fourier, the Cahn-Hilliard-Navier-Stokes, and and Brenner-Navier-Stokes-Fourier systems, and significant generalizations of these systems are obtained.

We explore, through a combination of x-ray diffraction and elastoresistivity measurements, the effect of disorder on the strain-tuned charge density wave and associated multicriticality in Pd$_x$ErTe$_3$ (x = 0, 0.01, 0.02 and 0.026). We focus particularly on the behavior near the strain-tuned bicritical point that occurs in pristine ErTe$_3$ (x=0). Our study reveals that while Pd intercalation somewhat broadens the signatures of the CDW phase transitions, the line of first-order transitions at which the CDW reorients as a function of applied strain persists in the presence of disorder and still seemingly terminates at a critical point. The critical point occurs at a lower temperature and a lower strain compared to pristine ErTe$_3$. Similarly, the nematic elastoresistance of Pd$_x$ErTe$_3$, though suppressed in magnitude and broadened relative to that of ErTe$_3$, has a markedly more symmetric response around the critical point. These observations point to disorder driving a reduction in the system's electronic orthorhombicity even while the material remains irrevocably orthorhombic due to the presence of a glide plane in the crystal structure. Disorder, it would appear, reinforces the emergence of a "pseudo-tetragonal" electronic response in this fundamentally orthorhombic material.

This paper presents the results of a Neural Network (NN)-based search for short-duration gravitational-wave transients in data from the third observing run of LIGO, Virgo, and KAGRA. The search targets unmodeled transients with durations of milliseconds to a few seconds in the 30-1500 Hz frequency band, without assumptions about the incoming signal direction, polarization, or morphology. Using the Gravitational Wave Anomalous Knowledge (GWAK) method, three compact binary coalescences (CBCs) identified by existing pipelines are successfully detected, along with a range of detector glitches. The algorithm constructs a low-dimensional embedded space to capture the physical features of signals, enabling the detection of CBCs, detector glitches, and unmodeled transients. This study demonstrates GWAK's ability to enhance gravitational-wave searches beyond the limits of existing pipelines, laying the groundwork for future detection strategies.

We use Matsuki's decomposition for symmetric pairs $(G, H)$ of (not necessarily compact) reductive Lie groups to construct the radial parts for invariant differential operators acting on matrix-spherical functions. As an application, we employ this machinery to formulate an alternative, mathematically rigorous approach to obtaining radial parts of Casimir operators that appear in the theory of conformal blocks, which avoids poorly defined analytical continuations from the compact quotient cases. To exemplify how this works, after reviewing the presentation of conformal 4-point correlation functions via matrix-spherical functions for the corresponding symmetric pair, we for the first time provide a complete analysis of the Casimir radial part decomposition in the case of Lorentzian signature. As another example, we revisit the Casimir reduction in the case of conformal blocks for two scalar defects of equal dimension. We argue that Matsuki's decomposition thus provides a proper mathematical framework for analysing the correspondence between Casimir equations and the Calogero-Sutherland-type models, first discovered by one of the authors and Schomerus., Comment: 72 pages

This paper explores the critical role of data clustering in data science, emphasizing its methodologies, tools, and diverse applications. Traditional techniques, such as partitional and hierarchical clustering, are analyzed alongside advanced approaches such as data stream, density-based, graph-based, and model-based clustering for handling complex structured datasets. The paper highlights key principles underpinning clustering, outlines widely used tools and frameworks, introduces the workflow of clustering in data science, discusses challenges in practical implementation, and examines various applications of clustering. By focusing on these foundations and applications, the discussion underscores clustering's transformative potential. The paper concludes with insights into future research directions, emphasizing clustering's role in driving innovation and enabling data-driven decision-making.

Large language models (LLMs) are widely applied in various natural language processing tasks such as question answering and machine translation. However, due to the lack of labeled data and the difficulty of manual annotation for biochemical properties, the performance for molecule generation tasks is still limited, especially for tasks involving multi-properties constraints. In this work, we present a two-step framework PEIT (Property Enhanced Instruction Tuning) to improve LLMs for molecular-related tasks. In the first step, we use textual descriptions, SMILES, and biochemical properties as multimodal inputs to pre-train a model called PEIT-GEN, by aligning multi-modal representations to synthesize instruction data. In the second step, we fine-tune existing open-source LLMs with the synthesized data, the resulting PEIT-LLM can handle molecule captioning, text-based molecule generation, molecular property prediction, and our newly proposed multi-constraint molecule generation tasks. Experimental results show that our pre-trained PEIT-GEN outperforms MolT5 and BioT5 in molecule captioning, demonstrating modalities align well between textual descriptions, structures, and biochemical properties. Furthermore, PEIT-LLM shows promising improvements in multi-task molecule generation, proving the scalability of the PEIT framework for various molecular tasks. We release the code, constructed instruction data, and model checkpoints in https://github.com/chenlong164/PEIT.

Refusals - instances where large language models (LLMs) decline or fail to fully execute user instructions - are crucial for both AI safety and AI capabilities and the reduction of hallucinations in particular. These behaviors are learned during post-training, especially in instruction fine-tuning (IFT) and reinforcement learning from human feedback (RLHF). However, existing taxonomies and evaluation datasets for refusals are inadequate, often focusing solely on should-not-related (instead of cannot-related) categories, and lacking tools for auditing refusal content in black-box LLM outputs. We present a comprehensive framework for classifying LLM refusals: (a) a taxonomy of 16 refusal categories, (b) a human-annotated dataset of over 8,600 instances from publicly available IFT and RLHF datasets, (c) a synthetic dataset with 8,000 examples for each refusal category, and (d) classifiers trained for refusal classification. Our work enables precise auditing of refusal behaviors in black-box LLMs and automatic analyses of refusal patterns in large IFT and RLHF datasets. This facilitates the strategic adjustment of LLM refusals, contributing to the development of more safe and reliable LLMs., Comment: NeurIPS 2024 Workshop SFLLM

We present a parallelized differentiable traffic simulator based on the Intelligent Driver Model (IDM), a car-following framework that incorporates driver behavior as key variables. Our simulator efficiently models vehicle motion, generating trajectories that can be supervised to fit real-world data. By leveraging its differentiable nature, IDM parameters are optimized using gradient-based methods. With the capability to simulate up to 2 million vehicles in real time, the system is scalable for large-scale trajectory optimization. We show that we can use the simulator to filter noise in the input trajectories (trajectory filtering), reconstruct dense trajectories from sparse ones (trajectory reconstruction), and predict future trajectories (trajectory prediction), with all generated trajectories adhering to physical laws. We validate our simulator and algorithm on several datasets including NGSIM and Waymo Open Dataset., Comment: 8 pages, 6 figures, 2 tables

As deep learning models become larger and more expensive, many practitioners turn to fine-tuning APIs. These web services allow fine-tuning a model between two parties: the client that provides the data, and the server that hosts the model. While convenient, these APIs raise a new concern: the data of the client is at risk of privacy breach during the training procedure. This challenge presents an important practical case of vertical federated learning, where the two parties perform parameter-efficient fine-tuning (PEFT) of a large model. In this study, we systematically search for a way to fine-tune models over an API while keeping the labels private. We analyze the privacy of LoRA, a popular approach for parameter-efficient fine-tuning when training over an API. Using this analysis, we propose P$^3$EFT, a multi-party split learning algorithm that takes advantage of existing PEFT properties to maintain privacy at a lower performance overhead. To validate our algorithm, we fine-tune DeBERTa-v2-XXLarge, Flan-T5 Large and LLaMA-2 7B using LoRA adapters on a range of NLP tasks. We find that P$^3$EFT is competitive with existing privacy-preserving methods in multi-party and two-party setups while having higher accuracy.

While large language models (LLMs) have been applied to automatic speech recognition (ASR), the task of making the model streamable remains a challenge. This paper proposes a novel model architecture, Transducer-Llama, that integrates LLMs into a Factorized Transducer (FT) model, naturally enabling streaming capabilities. Furthermore, given that the large vocabulary of LLMs can cause data sparsity issue and increased training costs for spoken language systems, this paper introduces an efficient vocabulary adaptation technique to align LLMs with speech system vocabularies. The results show that directly optimizing the FT model with a strong pre-trained LLM-based predictor using the RNN-T loss yields some but limited improvements over a smaller pre-trained LM predictor. Therefore, this paper proposes a weak-to-strong LM swap strategy, using a weak LM predictor during RNN-T loss training and then replacing it with a strong LLM. After LM replacement, the minimum word error rate (MWER) loss is employed to finetune the integration of the LLM predictor with the Transducer-Llama model. Experiments on the LibriSpeech and large-scale multi-lingual LibriSpeech corpora show that the proposed streaming Transducer-Llama approach gave a 17% relative WER reduction (WERR) over a strong FT baseline and a 32% WERR over an RNN-T baseline., Comment: Accepted by ICASSP 2025

Remaining Useful Life (RUL) of a component or a system is defined as the length from the current time to the end of the useful life. Accurate RUL estimation plays a crucial role in Predictive Maintenance applications. Traditional regression methods, both linear and non-linear, have struggled to achieve high accuracy in this domain. While Convolutional Neural Networks (CNNs) have shown improved accuracy, they often overlook the sequential nature of the data, relying instead on features derived from sliding windows. Since RUL prediction inherently involves multivariate time series analysis, robust sequence learning is essential. In this work, we propose a hybrid approach combining Convolutional Neural Networks with Long Short-Term Memory (LSTM) networks for RUL estimation. Although CNN-based LSTM models have been applied to sequence prediction tasks in financial forecasting, this is the first attempt to adopt this approach for RUL estimation in prognostics. In this approach, CNN is first employed to efficiently extract features from the data, followed by LSTM, which uses these extracted features to predict RUL. This method effectively leverages sensor sequence information, uncovering hidden patterns within the data, even under multiple operating conditions and fault scenarios. Our results demonstrate that the hybrid CNN-LSTM model achieves the highest accuracy, offering a superior score compared to the other methods., Comment: conference paper

This paper presents a novel deep learning framework for estimating multivariate joint extremes of metocean variables, based on the Semi-Parametric Angular-Radial (SPAR) model. When considered in polar coordinates, the problem of modelling multivariate extremes is transformed to one of modelling an angular density, and the tail of a univariate radial variable conditioned on angle. In the SPAR approach, the tail of the radial variable is modelled using a generalised Pareto (GP) distribution, providing a natural extension of univariate extreme value theory to the multivariate setting. In this work, we show how the method can be applied in higher dimensions, using a case study for five metocean variables: wind speed, wind direction, wave height, wave period and wave direction. The angular variable is modelled empirically, while the parameters of the GP model are approximated using fully-connected deep neural networks. Our data-driven approach provides great flexibility in the dependence structures that can be represented, together with computationally efficient routines for training the model. Furthermore, the application of the method requires fewer assumptions about the underlying distribution(s) compared to existing approaches, and an asymptotically justified means for extrapolating outside the range of observations. Using various diagnostic plots, we show that the fitted models provide a good description of the joint extremes of the metocean variables considered.

In this work, we prove the NP-completeness of two variants of tokenisation, defined as the problem of compressing a dataset to at most $\delta$ symbols by either finding a vocabulary directly (direct tokenisation), or selecting a sequence of merge operations (bottom-up tokenisation).

The rotational states of ultracold polar molecules possess long radiative lifetimes, microwave-domain coupling, and tunable dipolar interactions. Coherent dynamics between pairs of rotational states have been used to demonstrate simple models of quantum magnetism and to manipulate quantum information stored as qubits. The availability of numerous rotational states has led to many proposals to implement more complicated models of quantum magnetism, higher-dimensional qudits, and intricate state networks as synthetic dimensions; however, these are yet to be experimentally realised. The primary issue limiting their implementation is the detrimental effect of the optical trapping environment on coherence, which is not easily mitigated for systems beyond two levels. To address this challenge, we investigate the applicability of magic-wavelength optical tweezer traps to facilitate multitransition coherence between rotational states. We demonstrate simultaneous second-scale coherence between three rotational states. Utilising this extended coherence, we perform multiparameter estimation using a generalised Ramsey sequence and demonstrate coherent spin-1 dynamics encoded in the rotational states. Our work paves the way to implementing proposed quantum simulation, computation, and metrology schemes that exploit the rich rotational structure of ultracold polar molecules., Comment: 15 pages, 8 figures

We introduce the Velocity Jumps approach, denoted as JUMP, a new class of Molecular dynamics integrators, replacing the Langevin dynamics by a hybrid model combining a classical Langevin diffusion and a piecewise deterministic Markov process, where the expensive computation of long-range pairwise interactions is replaced by a resampling of the velocities at random times. This framework allows for an acceleration in the simulation speed while preserving sampling and dynamical properties such as the diffusion constant. It can also be integrated in classical multi-timestep methods, pushing further the computational speedup, while avoiding some of the resonance issues of the latter thanks to the random nature of jumps. The JUMP, JUMP-RESPA and JUMP-RESPA1 integrators have been implemented in the GPU-accelerated version of the Tinker-HP package and are shown to provide significantly enhanced performances compared to their BAOAB, BAOAB-RESPA and BAOAB-RESPA1 counterparts respectively.

Multimodal machine learning has achieved remarkable progress in many scenarios, but its reliability is undermined by varying sample quality. In this paper, we find that current multimodal classification methods lack dynamic networks for sample-specific depth and parameters to achieve reliable inference. To this end, a novel framework for multimodal reliable classification termed Multi-Level Quality-Adaptive Dynamic Multimodal Network (QADM-Net) is proposed. QADM-Net first adopts a novel approach based on noise-free prototypes and a classifier-free design to reliably estimate the quality of each sample at both modality and feature levels. It then achieves sample-specific network depth via the \textbf{\textit{Global Confidence Normalized Depth (GCND)}} mechanism. By normalizing depth across modalities and samples, \textit{\textbf{GCND}} effectively mitigates the impact of challenging modality inputs on dynamic depth reliability. Furthermore, QADM-Net provides sample-adaptive network parameters via the \textbf{\textit{Layer-wise Greedy Parameter (LGP)}} mechanism driven by feature-level quality. The cross-modality layer-wise greedy strategy in \textbf{\textit{LGP}} designs a reliable parameter prediction paradigm for multimodal networks with variable depths for the first time. Experiments conducted on four datasets demonstrate that QADM-Net significantly outperforms state-of-the-art methods in classification performance and reliability, exhibiting strong adaptability to data with diverse quality., Comment: 9 pages, 5 figures

The density-density response in optimally doped Bi$_2$Sr$_2$CaCu$_2$O$_{8+x}$ has recently been shown to exhibit conformal symmetry. Using, the experimentally inferred conformal dynamic susceptibility, we compute the resultant quantum Fisher information (QFI), a witness to multi-partite entanglement. In contrast to a Fermi liquid in which the QFI is approximately temperature independent much below the Fermi energy scale, we find that the QFI increases as a power law at low temperatures but ultimately extrapolates to a constant at $T=0$. The constant is of the form, $\omega_g^{2\Delta}$, where $\Delta$ is the conformal dimension and $\omega_g$ is the UV cutoff which is on the order of the pseudogap. As this constant {depends on both UV and IR properties}, it illustrates that multipartite entanglement in a strange metal exhibits UV-IR mixing, a benchmark feature of doped Mott insulators as exemplified by dynamical spectral weight transfer. We conclude with a discussion of the implication of our results for low-energy reductions of the Hubbard model., Comment: For an issue of Physica C dedicated to Jan Zaanen

Differential privacy is the gold standard for privacy in data analysis. In many data analysis applications, the data is a database of documents. For databases consisting of many documents, one of the most fundamental problems is that of pattern matching and computing (i) how often a pattern appears as a substring in the database (substring counting) and (ii) how many documents in the collection contain the pattern as a substring (document counting). In this paper, we initiate the theoretical study of substring and document counting under differential privacy. We give an $\epsilon$-differentially private data structure solving this problem for all patterns simultaneously with a maximum additive error of $O(\ell \cdot\mathrm{polylog}(n\ell|\Sigma|))$, where $\ell$ is the maximum length of a document in the database, $n$ is the number of documents, and $|\Sigma|$ is the size of the alphabet. We show that this is optimal up to a $O(\mathrm{polylog}(n\ell))$ factor. Further, we show that for $(\epsilon,\delta)$-differential privacy, the bound for document counting can be improved to $O(\sqrt{\ell} \cdot\mathrm{polylog}(n\ell|\Sigma|))$. Additionally, our data structures are efficient. In particular, our data structures use $O(n\ell^2)$ space, $O(n^2\ell^4)$ preprocessing time, and $O(|P|)$ query time where $P$ is the query pattern. Along the way, we develop a new technique for differentially privately computing a general class of counting functions on trees of independent interest. Our data structures immediately lead to improved algorithms for related problems, such as privately mining frequent substrings and $q$-grams. For $q$-grams, we further improve the preprocessing time of the data structure., Comment: 33 pages

SocialED is a comprehensive, open-source Python library designed to support social event detection (SED) tasks, integrating 19 detection algorithms and 14 diverse datasets. It provides a unified API with detailed documentation, offering researchers and practitioners a complete solution for event detection in social media. The library is designed with modularity in mind, allowing users to easily adapt and extend components for various use cases. SocialED supports a wide range of preprocessing techniques, such as graph construction and tokenization, and includes standardized interfaces for training models and making predictions. By integrating popular deep learning frameworks, SocialED ensures high efficiency and scalability across both CPU and GPU environments. The library is built adhering to high code quality standards, including unit testing, continuous integration, and code coverage, ensuring that SocialED delivers robust, maintainable software. SocialED is publicly available at \url{https://github.com/RingBDStack/SocialED} and can be installed via PyPI., Comment: 8 pages, 1 figure, Python library

Fault-tolerance is critically important in highly-distributed modern cloud applications. Solutions such as Temporal, Azure Durable Functions, and Beldi hide fault-tolerance complexity from developers by persisting execution state and resuming seamlessly from persisted state after failure. This pattern, often called durable execution, usually forces frequent and synchronous persistence and results in hefty latency overheads. In this paper, we propose distributed speculative execution (DSE), a technique for implementing the durable execution abstraction without incurring this penalty. With DSE, developers write code assuming synchronous persistence, and a DSE runtime is responsible for transparently bypassing persistence and reactively repairing application state on failure. We present libDSE, the first DSE application framework that achieves this vision. The key tension in designing libDSE is between imposing restrictions on user programs so the framework can safely and transparently change execution behavior, and avoiding assumptions so libDSE can support more use cases. We address this with a novel programming model centered around message-passing, atomic code blocks, and lightweight threads, and show that it allows developers to build a variety of speculative services, including write-ahead logs, key-value stores, event brokers, and fault-tolerant workflows. Our evaluation shows that libDSE reduces end-to-end latency by up to an order of magnitude compared to current generations of durable execution systems with minimal run-time overhead and manageable complexity.

A system of two cubic reaction-diffusion equations for two independent gene frequencies arising in population dynamics is studied. Depending on values of coefficients, all possible Lie and $Q$-conditional (nonclassical) symmetries are identified. A wide range of new exact solutions is constructed, including those expressible in terms of a Lambert function and not obtainable by Lie symmetries. An example of a new real-world application of the system is discussed. A general algorithm for finding Q-conditional symmetries of nonlinear evolution systems of the most general form is presented in a useful form for other researchers., Comment: 23 pages

We employ polarization tomography to characterize the modal properties of a dielectric nanocavity with sub-wavelength mode confinement. Our analysis of reflection spectra shows that the Fano-lineshape depends strongly on the polarization in a confocal configuration, and that the lineshape can be transformed into a Lorentzian-like peak for a certain polarization. For this polarization setting, the background is almost fully suppressed in a finite range of frequencies. This enables us to identify another resonance that has not yet been experimentally reported for these nanocavities. Lastly, we use symmetry-forbidden polarizations and show that, surprisingly, the modal resonance features of the system remain visible.

Verifying factual claims is critical for using large language models (LLMs) in healthcare. Recent work has proposed fact decomposition, which uses LLMs to rewrite source text into concise sentences conveying a single piece of information, as an approach for fine-grained fact verification. Clinical documentation poses unique challenges for fact decomposition due to dense terminology and diverse note types. To explore these challenges, we present FactEHR, a dataset consisting of full document fact decompositions for 2,168 clinical notes spanning four types from three hospital systems. Our evaluation, including review by clinicians, highlights significant variability in the quality of fact decomposition for four commonly used LLMs, with some LLMs generating 2.6x more facts per sentence than others. The results underscore the need for better LLM capabilities to support factual verification in clinical text. To facilitate future research in this direction, we plan to release our code at \url{https://github.com/som-shahlab/factehr}.

The finite monodromy groups of abelian varieties over number fields have been introduced by Grothendieck. They represent the local obstruction to semi-stable reduction. In this paper we prove a criteria for finite groups to be realized as finite monodromy groups in given dimension. An application to the degree of semi-stability gives an effective version of Grothendieck's semi-stable reduction theorem in terms of the degree of the extension with regards to the dimension of the variety., Comment: The text is in French. Comments are welcome