Search

Suspensions of self-propelled objects represent a novel paradigm in colloidal science. In such active baths traditional concepts, such as Brownian motion, fluctuation-dissipation relations, and work extraction from heat reservoirs, must be extended beyond the conventional framework of thermal baths. Unlike thermal baths, which are characterized by a single parameter, the temperature, the fundamental descriptors of an active bath remain elusive, especially in confined environments. In this study, buoyant, passive tracers are employed as generalized probes to investigate an active bath comprising motile bacteria confined within a droplet. We demonstrate that momentum transfer from the bath to the tracer can be effectively described as colored noise, characterized by temporal memory and an enhanced effective diffusivity significantly larger compared to thermal Brownian motion values. Using a stochastic analytical framework, we extract the temporal memory and diffusivity parameters that define such an active bath. Notably, the diffusivity scales linearly with bacterial concentration, modulated by a factor representing the role of confinement, expressed as the ratio of the confining radius to the probe radius. This finding, while still awaiting a complete theoretical explanation, offers new insights into the transport properties of confined active baths and paves the way for a deeper understanding of active emulsions driven by confined active matter.

Flows of particles through bottlenecks are ubiquitous in nature and industry, involving both dry granular materials and suspensions. However, practical limitations of conventional experimental setups hinder the full understanding of these flows in confined geometries. Here, we present a microfluidic setup to investigate experimentally the flow of dense suspensions in a two-dimensional hopper channel. Particles with controlled properties are in-situ fabricated with a photolithographic projection method and compacted at the channel constriction using a Quake valve. The setup is characterized by examining the flow of a dense suspension of hard, monodisperse disks through constrictions of varying widths. We demonstrate that the microfluidic hopper discharges particles at constant rate, resulting from the channel resistance being dominated by the presence of densely packed particles within the tapered section of the hopper. Under imposed flow rate the discharge remains independent of particle and orifice sizes, whereas it exhibits a Beverloo-like scaling under pressure-imposed conditions. Additionally, we show that the statistics of clog formation in our microfluidic hopper follow the same stochastic laws as reported in other systems. Finally, we show how the versatility of our microfluidic model system can be used to investigate the outflow and clogging of suspensions of more complex particles.

Deep model merging represents an emerging research direction that combines multiple fine-tuned models to harness their specialized capabilities across different tasks and domains. Current model merging techniques focus on merging all available models simultaneously, with weight interpolation-based methods being the predominant approaches. However, these conventional approaches are not well-suited for scenarios where models become available sequentially, and they often suffer from high memory requirements and potential interference between tasks. In this study, we propose a training-free projection-based continual merging method that processes models sequentially through orthogonal projections of weight matrices and adaptive scaling mechanisms. Our method operates by projecting new parameter updates onto subspaces orthogonal to existing merged parameter updates while using an adaptive scaling mechanism to maintain stable parameter distances, enabling efficient sequential integration of task-specific knowledge. Our approach maintains constant memory complexity to the number of models, minimizes interference between tasks through orthogonal projections, and retains the performance of previously merged models through adaptive task vector scaling. Extensive experiments on CLIP-ViT models demonstrate that our method achieves a 5-8% average accuracy improvement while maintaining robust performance in different task orderings.

The anomalous Hall effect (AHE) occurs in magnetic systems and also unexpectedly in non-magnetic materials adjacent to magnetic insulators via the heterointerface interactions. However, the AHE in heterostructures induced by magnetic proximity effect remains quite weak, restricting their practical device applications. Here, we report a large intrinsic AHE with a resistivity of 114 n{\Omega} cm at 5 K in noncoplanar magnetic heterostructures of Cr5Te6/Pt. This is the record-high AHE value among all the magnetic insulators/heavy metal heterostructures. A reversal of the AHE signal occurs due to the reconstruction of Berry curvature at the Fermi level, which is verified by the first-principles calculations. Topological spin textures at the interface are directly visualized via high-magnetic-field magnetic force microscopy, which accounts for the large AHE, as confirmed by the atomic simulations. These findings open a new avenue for exploring the large AHE in heterointerfaces and facilitate the potential applications in topological spintronic devices., Comment: 23 pages, 15 figures

We realize laser cooling and trapping of titanium (Ti) atoms in a mangeto-optical trap (MOT). While Ti does not possess a transition suitable for laser cooling out of its $\mathrm{3d^24s^2}$ $\mathrm{a^3F}$ ground term, there is such a transition, at an optical wavelength of $\lambda=498\mathrm{nm}$, from the long-lived $\mathrm{3d^3(^4F)4s}$ $\mathrm{a^5F_5}$ metastable state to the $\mathrm{3d^3(^4F)4p}$ $\mathrm{y^5G^o_6}$ excited state. Without the addition of any repumping light, we observe MOTs of metastable $\mathrm{^{46}Ti}$, $\mathrm{^{48}Ti}$, and $\mathrm{^{50}Ti}$, the three stable nuclear-spin-zero bosonic isotopes of Ti. While MOTs can be observed when loaded directly from our Ti sublimation source, optical pumping of ground term atoms to the $\mathrm{a^5F_5}$ state increases the loading rate by a factor of 120, and the steady-state MOT atom number by a factor of 30. At steady state, the MOT of $\mathrm{^{48}Ti}$ holds up to $8.30(26)\times10^5$ atoms at a maximum density of $1.3(4)\times10^{11}\mathrm{cm}^{-3}$ and at a temperature of $90(15)\mathrm{\mu K}$. By measuring the decay of the MOT upon suddenly reducing the loading rate, we place upper bounds on the leakage branching ratio of the cooling transition $(\leq2.5\times 10^{-6})$ and the two-body loss coefficient $(\leq2\times10^{-10}\mathrm{cm}^3\mathrm{s}^{-1})$. Our approach to laser cooling Ti can be applied to other transition metals, enabling a significant expansion of the elements that can be laser cooled., Comment: 6 pages, 4 figures, 1 table

Merging multiple expert models offers a promising approach for performing multi-task learning without accessing their original data. Existing methods attempt to alleviate task conflicts by sparsifying task vectors or promoting orthogonality among them. However, they overlook the fundamental requirement of model merging: ensuring the merged model performs comparably to task-specific models on respective tasks. We find these methods inevitably discard task-specific information that, while causing conflicts, is crucial for performance. Based on our findings, we frame model merging as a constrained optimization problem ($\textit{i.e.}$, minimizing the gap between the merged model and individual models, subject to the constraint of retaining shared knowledge) and solve it via adaptive projective gradient descent. Specifically, we align the merged model with individual models by decomposing and reconstituting the loss function, alleviating conflicts through $\textit{data-free}$ optimization of task vectors. To retain shared knowledge, we optimize this objective by projecting gradients within a $\textit{shared subspace}$ spanning all tasks. Moreover, we view merging coefficients as adaptive learning rates and propose a task-aware, training-free strategy. Experiments show that our plug-and-play approach consistently outperforms previous methods, achieving state-of-the-art results across diverse architectures and tasks in both vision and NLP domains.

In this paper we derive a new direct inversion method to simulate squared Bessel processes. Since the transition probability of these processes can be represented by a non-central chi-square distribution, we construct an efficient and accurate algorithm to simulate non-central chi-square variables. In this method, the dimension of the squared Bessel process, equivalently the degrees of freedom of the chi-square distribution, is treated as a variable. We therefore use a two-dimensional Chebyshev expansion to approximate the inverse function of the central chi-square distribution with one variable being the degrees of freedom. The method is accurate and efficient for any value of degrees of freedom including the computationally challenging case of small values. One advantage of the method is that noncentral chi-square samples can be generated for a whole range of values of degrees of freedom using the same Chebyshev coefficients. The squared Bessel process is a building block for the well-known Cox-Ingersoll-Ross (CIR) processes, which can be generated from squared Bessel processes through time change and linear transformation. Our direct inversion method thus allows the efficient and accurate simulation of these processes, which are used as models in a wide variety of applications., Comment: 23 pages, 3 figures

Extracting metaphors and analogies from free text requires high-level reasoning abilities such as abstraction and language understanding. Our study focuses on the extraction of the concepts that form metaphoric analogies in literary texts. To this end, we construct a novel dataset in this domain with the help of domain experts. We compare the out-of-the-box ability of recent large language models (LLMs) to structure metaphoric mappings from fragments of texts containing proportional analogies. The models are further evaluated on the generation of implicit elements of the analogy, which are indirectly suggested in the texts and inferred by human readers. The competitive results obtained by LLMs in our experiments are encouraging and open up new avenues such as automatically extracting analogies and metaphors from text instead of investing resources in domain experts to manually label data., Comment: Accepted to COLING 2025, long paper

Global analyses in the Standard Model Effective Field Theory (SMEFT) framework serve as a tool to probe potential directions of new physics. To break degeneracies between the Wilson coefficients of the SMEFT, it is essential to combine observables from various experiments. Since different observables entering global fits may be measured at different energy scales, it becomes increasingly important to account for this fact through the renormalisation group evolution (RGE) of the Wilson coefficients. In this work, we investigate the effects of the RGE on a global SMEFT fit under the assumption of a $U(3)^5$ symmetry within the minimal flavour violation framework. We comment on the role of next-to-leading order SMEFT predictions for breaking potential degeneracies between Wilson coefficients arising as a result of RGE effects., Comment: 14 pages, 8 figures

Reverse Dictionary (RD) is the task of obtaining the most relevant word or set of words given a textual description or dictionary definition. Effective RD methods have applications in accessibility, translation or writing support systems. Moreover, in NLP research we find RD to be used to benchmark text encoders at various granularities, as it often requires word, definition and sentence embeddings. In this paper, we propose a simple approach to RD that leverages LLMs in combination with embedding models. Despite its simplicity, this approach outperforms supervised baselines in well studied RD datasets, while also showing less over-fitting. We also conduct a number of experiments on different dictionaries and analyze how different styles, registers and target audiences impact the quality of RD systems. We conclude that, on average, untuned embeddings alone fare way below an LLM-only baseline (although they are competitive in highly technical dictionaries), but are crucial for boosting performance in combined methods., Comment: 9 pages, Accepted at COLING 2025

The Sagittarius dwarf spheroidal galaxy (Sgr dSph) is a satellite orbiting the Milky Way that has experienced multiple stripping events due to tidal interactions with our Galaxy. Its accretion history has led to a distinct stellar overdensity, which is the remnant of the core of the progenitor. We present a complete chemical analysis of 111 giant stars in the core of Sgr dSph to investigate the chemical evolution and enrichment history of this satellite. Employing the metallicity-sensitive Ca H&K photometry from the Pristine Inner Galaxy Survey, we selected stars spanning a wide metallicity range and obtained high-resolution spectra with the ESO FLAMES/GIRAFFE multi-object spectrograph. For the stellar sample covering $-2.13 < \rm{[Fe/H] < -0.35}$, we derived abundances for up to 14 chemical elements with average uncertainties of $\sim 0.09$ dex and a set of stellar ages which allowed us to build an age-metallicity relation (AMR) for the entire sample. With the most comprehensive set of chemical species measured for the core of Sgr, we studied several [X/Fe] ratios. Most trends align closely with Galactic chemical trends, but notable differences emerge in the heavy $n$-capture elements, which offer independent insights into the star formation history of a stellar population. The deficiency in the $\alpha$-elements with respect the Milky Way suggests a slower, less efficient early star formation history, similar to other massive satellites. $S$-process element patterns indicate significant enrichment from AGB stars over time. The AMR and chemical ratios point to an extended star formation history, with a rapid early phase in the first Gyr, followed by declining activity and later star-forming episodes. These findings are consistent with Sgr hosting multiple stellar populations, from young ($\sim 4$ Gyr) to old, metal-poor stars ($\sim 10$ Gyr), Comment: Submitted to A&A. 23 pages and 13 figures. Comments welcome. The abstract has been modified for the ArXiv preview

Activity and autonomous motion are fundamental aspects of many living and engineering systems. Here, the scale of biological agents covers a wide range, from nanomotors, cytoskeleton, and cells, to insects, fish, birds, and people. Inspired by biological active systems, various types of autonomous synthetic nano- and micromachines have been designed, which provide the basis for multifunctional, highly responsive, intelligent active materials. A major challenge for understanding and designing active matter is their inherent non-equilibrium nature due to persistent energy consumption, which invalidates equilibrium concepts such as free energy, detailed balance, and time-reversal symmetry. Furthermore, interactions in ensembles of active agents are often non-additive and non-reciprocal. An important aspect of biological agents is their ability to sense the environment, process this information, and adjust their motion accordingly. It is an important goal for the engineering of micro-robotic systems to achieve similar functionality. With many fundamental properties of motile active matter now reasonably well understood and under control, the ground is prepared for the study of physical aspects and mechanisms of motion in complex environments, of the behavior of systems with new physical features like chirality, of the development of novel micromachines and microbots, of the emergent collective behavior and swarming of intelligent self-propelled particles, and of particular features of microbial systems. The vast complexity of phenomena and mechanisms involved in the self-organization and dynamics of motile active matter poses major challenges, which can only be addressed by a truly interdisciplinary effort involving scientists from biology, chemistry, ecology, engineering, mathematics, and physics.

In this paper, we derive a Chen-Strichartz formula for stochastic differential equations driven by Levy processes, that is, we derive a series expansion of the logarithm of the flowmap of the stochastic differential equation in terms of commutators of vector fields with stochastic coefficients, and we provide an explicit formula for the components in this series. The stochastic components are generated by the Levy processes that drive the stochastic differential equation and their quadratic variation and power jumps; the vector fields are given as linear combinations of commutators of elements in the pre-Lie Magnus expansion generated by the original vector fields governing our stochastic differential equation. In particular, we show the logarithm of the flowmap is a Lie series. These results extend previous results for deterministic differential equations and continuous stochastic differential equations. For these, the Chen-Strichartz series has shown to play a pivotal role in the design of numerical integration schemes that preserve qualitative properties of the solution such as the construction of geometric numerical schemes and in the context of efficient numerical schemes., Comment: 25 pages

As a promising technology, physical layer security (PLS) enhances security by leveraging the physical characteristics of communication channels. However, the conventional PLS approach leads to a considerable disparity in the effort legitimate users need to secure data compared to eavesdroppers. To address this issue, we propose a physical layer deception (PLD) framework, which applies random deceptive ciphering and orthogonal frequency-division multiplexing (OFDM) to defend against eavesdropping proactively. While ensuring the same level of confidentiality as traditional PLS methods, the PLD approach additionally introduces a deception mechanism, even when the eavesdropper possesses the same knowledge about the transmitter end as the legitimate receiver. Through thorough theoretical analyses and numerical simulations, we prove the superiority of our method over the conventional PLS approach., Comment: Submitted to ICC 2025 (appendices are excluded from the submitted version due to length limit)

Multi-task learning (MTL) leverages a shared model to accomplish multiple tasks and facilitate knowledge transfer. Recent research on task arithmetic-based MTL demonstrates that merging the parameters of independently fine-tuned models can effectively achieve MTL. However, existing merging methods primarily seek a static optimal solution within the original model parameter space, which often results in performance degradation due to the inherent diversity among tasks and potential interferences. To address this challenge, in this paper, we propose a Weight-Ensembling Mixture of Experts (WEMoE) method for multi-task model merging. Specifically, we first identify critical (or sensitive) modules by analyzing parameter variations in core modules of Transformer-based models before and after finetuning. Then, our WEMoE statically merges non-critical modules while transforming critical modules into a mixture-of-experts (MoE) structure. During inference, expert modules in the MoE are dynamically merged based on input samples, enabling a more flexible and adaptive merging approach. Building on WEMoE, we further introduce an efficient-and-effective WEMoE (E-WEMoE) method, whose core mechanism involves eliminating non-essential elements in the critical modules of WEMoE and implementing shared routing across multiple MoE modules, thereby significantly reducing both the trainable parameters, the overall parameter count, and computational overhead of the merged model by WEMoE. Experimental results across various architectures and tasks demonstrate that both WEMoE and E-WEMoE outperform state-of-the-art (SOTA) model merging methods in terms of MTL performance, generalization, and robustness.

Model merging has gained significant attention as a cost-effective approach to integrate multiple single-task fine-tuned models into a unified one that can perform well on multiple tasks. However, existing model merging techniques primarily focus on resolving conflicts between task-specific models, they often overlook potential security threats, particularly the risk of backdoor attacks in the open-source model ecosystem. In this paper, we first investigate the vulnerabilities of existing model merging methods to backdoor attacks, identifying two critical challenges: backdoor succession and backdoor transfer. To address these issues, we propose a novel Defense-Aware Merging (DAM) approach that simultaneously mitigates task interference and backdoor vulnerabilities. Specifically, DAM employs a meta-learning-based optimization method with dual masks to identify a shared and safety-aware subspace for model merging. These masks are alternately optimized: the Task-Shared mask identifies common beneficial parameters across tasks, aiming to preserve task-specific knowledge while reducing interference, while the Backdoor-Detection mask isolates potentially harmful parameters to neutralize security threats. This dual-mask design allows us to carefully balance the preservation of useful knowledge and the removal of potential vulnerabilities. Compared to existing merging methods, DAM achieves a more favorable balance between performance and security, reducing the attack success rate by 2-10 percentage points while sacrificing only about 1% in accuracy. Furthermore, DAM exhibits robust performance and broad applicability across various types of backdoor attacks and the number of compromised models involved in the merging process. We will release the codes and models soon., Comment: 21 pages,8 figures

Carbon-rich (C-rich) stars can be found at all metallicities and evolutionary stages. They are often the result of mass-transfer from a companion, but some of the most metal-poor C-rich objects are likely carrying the imprint of the metal-free First Stars from birth. In this work, we employ a neural network to predict metallicities and carbon abundances for over 10 million stars with Gaia low-resolution XP spectra, down to [Fe/H] = -3.0 and up to [C/Fe] $\approx$ +2. We identify ~2000 high-confidence bright (G<16) carbon-enhanced metal-poor (CEMP) stars with [Fe/H] < -2.0 and [C/Fe] > +0.7. The majority of our C-rich candidates have [Fe/H] > -2.0 and are expected to be binary mass-transfer products, supported by high barium abundances in GALAH and/or their Gaia RUWE and radial velocity variations. We confirm previous findings of an increase in C-rich stars with decreasing metallicity, adopting a definition of $3\sigma$ outliers from the [C/Fe] distribution, although our frequency appears to flatten for -3.0 < [Fe/H] < -2.0 at a level of 6-7%. We also find that the fraction of C-rich stars is low among globular cluster stars (connected to their lower binary fraction), and that it decreases for field stars more tightly bound to the Milky Way. We interpret these last results as evidence that disrupted globular clusters contribute more in the inner Galaxy, supporting previous work. Homogeneous samples like these are key to understanding the full population properties of C-rich stars, and this is just the beginning., Comment: accepted for publication in MNRAS, small changes only (new figures 5 and 11, table 2, appendix B)

Physical layer deception (PLD) is a novel security mechanism that combines physical layer security (PLS) with deception technologies to actively defend against eavesdroppers. In this paper, we establish a novel semantic model for PLD that evaluates its performance in terms of semantic distortion. By analyzing semantic distortion at varying levels of knowledge on the receiver's part regarding the key, we derive the receiver's optimal decryption strategy, and consequently, the transmitter's optimal deception strategy. The proposed semantic model provides a more generic understanding of the PLD approach independent from coding or multiplexing schemes, and allows for efficient real-time adaptation to fading channels., Comment: Submitted to ICC 2025

We report on XMM-Newton, NuSTAR, and NICER X-ray observations of CTCV J2056-3014, a cataclysmic variable (CV) with one of the fastest-spinning white dwarfs (WDs) at P = 29.6 s. While previously classified as an intermediate polar (IP), CJ2056 also exhibits the properties of WZ-Sge-type CVs, such as dwarf novae and superoutbursts. With XMM-Newton and NICER, we detected the spin period up to approximately 2 keV with 7-$\sigma$ significance. We constrained its derivative to |$\dot{P}$| < 1.8e-12 s/s after correcting for binary orbital motion. The pulsed profile is characterized by a single broad peak with approximately 25% modulation. NuSTAR detected a four-fold increase in unabsorbed X-ray flux coincident with an optical flare in November 2022. The XMM-Newton and NICER X-ray spectra in 0.3-10 keV are best characterized by an absorbed optically-thin three-temperature thermal plasma model (kT = 0.3, 1.0, and 4.9 keV), while the NuSTAR spectra in 3-30 keV are best fit by a single-temperature thermal plasma model (kT = 8.4 keV), both with Fe abundance $Z_{Fe}/Z_\odot$ = 0.3. CJ2056 exhibits similarities to other fast-spinning CVs, such as low plasma temperatures, and no significant X-ray absorption at low energies. As the WD's magnetic field strength is unknown, we applied both non-magnetic and magnetic CV spectral models (MKCFLOW and MCVSPEC) to determine the WD mass. The derived WD mass range (M = 0.7-1.0 $M_\odot$) is above the centrifugal break-up mass limit of 0.56 $M_\odot$ and consistent with the mean WD mass of local CVs (M $\approx$ 0.8-0.9 $M_\odot$)., Comment: 22 pages, 12 figures, accepted to ApJ

The climate crisis is a salient issue in online discussions, and hypocrisy accusations are a central rhetorical element in these debates. However, for large-scale text analysis, hypocrisy accusation detection is an understudied tool, most often defined as a smaller subtask of fallacious argument detection. In this paper, we define hypocrisy accusation detection as an independent task in NLP, and identify different relevant subtypes of hypocrisy accusations. Our Climate Hypocrisy Accusation Corpus (CHAC) consists of 420 Reddit climate debate comments, expert-annotated into two different types of hypocrisy accusations: personal versus political hypocrisy. We evaluate few-shot in-context learning with 6 shots and 3 instruction-tuned Large Language Models (LLMs) for detecting hypocrisy accusations in this dataset. Results indicate that the GPT-4o and Llama-3 models in particular show promise in detecting hypocrisy accusations (F1 reaching 0.68, while previous work shows F1 of 0.44). However, context matters for a complex semantic concept such as hypocrisy accusations, and we find models struggle especially at identifying political hypocrisy accusations compared to personal moral hypocrisy. Our study contributes new insights in hypocrisy detection and climate change discourse, and is a stepping stone for large-scale analysis of hypocrisy accusation in online climate debates., Comment: cite the public version, published at CPSS 2024 @ KONVENS

The commissioning and operation of future large-scale scientific experiments will challenge current tuning and control methods. Reinforcement learning (RL) algorithms are a promising solution thanks to their capability of autonomously tackling a control problem based on a task parameterized by a reward function. The conventionally utilized machine learning (ML) libraries are not intended for microsecond latency applications, as they mostly optimize for throughput performance. On the other hand, most of the programmable logic implementations are meant for computation acceleration, not being intended to work in a real-time environment. To overcome these limitations of current implementations, RL needs to be deployed on-the-edge, i.e. on to the device gathering the training data. In this paper we present the design and deployment of an experience accumulator system in a particle accelerator. In this system deep-RL algorithms run using hardware acceleration and act within a few microseconds, enabling the use of RL for control of ultra-fast phenomena. The training is performed offline to reduce the number of operations carried out on the acceleration hardware. The proposed architecture was tested in real experimental conditions at the Karlsruhe research accelerator (KARA), serving also as a synchrotron light source, where the system was used to control induced horizontal betatron oscillations in real-time. The results showed a performance comparable to the commercial feedback system available at the accelerator, proving the viability and potential of this approach. Due to the self-learning and reconfiguration capability of this implementation, its seamless application to other control problems is possible. Applications range from particle accelerators to large-scale research and industrial facilities.

We analyse high-resolution ESPaDOnS/CFHT spectra of 20 very metal-poor stars ([Fe/H]~$<-2.0$) in the solar neighbourhood (within $\sim2$ kpc) selected to be on planar orbits (with a maximum height of $\lesssim4$ kpc). Targets include 11 prograde and 9 retrograde stars, spanning a wide range of eccentricities ($0.20-0.95$). Their chemical abundances are consistent with those observed in the Galactic halo but show a smaller spread, with no notable difference between progrades and retrogrades. This suggests a common chemical evolution and likely a shared formation site (except for one star). In this case, chemical evolution models indicate that the formation site would have had a baryonic mass of $\sim1.4\times10^9\msun$, similar to classical dwarf galaxies. High-energy supernovae and hypernovae are needed to reproduce the [X/Fe] up to the Fe-peak, while fast-rotating massive stars and neutron star merger events explain the [X/Fe] of the neutron-capture elements. The absence of Type Ia supernova signatures suggests a star formation duration of $\lesssim1$~Gyr. Cosmological zoom-in simulations support the scenario that an in-plane infall of a single system could disperse stars over a wide range of angular momenta during the early Galactic assembly. We propose that these stars originated in a proto-Galactic building block, which we name Loki. Less likely, if progrades and retrogrades formed in two different systems, their chemical evolution must have been very similar, with a combined baryonic mass twice that of a single system. Forthcoming surveys will provide a large and homogeneous dataset to investigate whether Loki is associated with any of the known detected structures. A comparison (primarily [$\alpha$/Fe]) with other VMPs moving in planar orbits suggests multiple systems contributed to the Galactic planar population, presenting some differences in their kinematical parameters., Comment: REference list updated; discussion on GSE updated

This paper presents LiteVSR2, an enhanced version of our previously introduced efficient approach to Visual Speech Recognition (VSR). Building upon our knowledge distillation framework from a pre-trained Automatic Speech Recognition (ASR) model, we introduce two key improvements: a stabilized video preprocessing technique and feature normalization in the distillation process. These improvements yield substantial performance gains on the LRS2 and LRS3 benchmarks, positioning LiteVSR2 as the current best CTC-based VSR model without increasing the volume of training data or computational resources utilized. Furthermore, we explore the scalability of our approach by examining performance metrics across varying model complexities and training data volumes. LiteVSR2 maintains the efficiency of its predecessor while significantly enhancing accuracy, thereby demonstrating the potential for resource-efficient advancements in VSR technology.

Thermography is especially valuable for the military and other users of surveillance cameras. Some recent methods based on Neural Radiance Fields (NeRF) are proposed to reconstruct the thermal scenes in 3D from a set of thermal and RGB images. However, unlike NeRF, 3D Gaussian splatting (3DGS) prevails due to its rapid training and real-time rendering. In this work, we propose ThermalGaussian, the first thermal 3DGS approach capable of rendering high-quality images in RGB and thermal modalities. We first calibrate the RGB camera and the thermal camera to ensure that both modalities are accurately aligned. Subsequently, we use the registered images to learn the multimodal 3D Gaussians. To prevent the overfitting of any single modality, we introduce several multimodal regularization constraints. We also develop smoothing constraints tailored to the physical characteristics of the thermal modality. Besides, we contribute a real-world dataset named RGBT-Scenes, captured by a hand-hold thermal-infrared camera, facilitating future research on thermal scene reconstruction. We conduct comprehensive experiments to show that ThermalGaussian achieves photorealistic rendering of thermal images and improves the rendering quality of RGB images. With the proposed multimodal regularization constraints, we also reduced the model's storage cost by 90\%. The code and dataset will be released., Comment: 10 pages, 7 figures

Globular clusters (GCs) are sites of extremely efficient star formation, and recent studies suggest they significantly contributed to the early Milky Way's stellar mass build-up. Although their role has since diminished, GCs' impact on the Galaxy's initial evolution can be traced today by identifying their most chemically unique stars--those with anomalous nitrogen and aluminum overabundances and oxygen depletion. While they are a perfect tracer of clusters, be it intact or fully dissolved, these high-[N/O], high-[Al/Fe] GC-origin stars are extremely rare within the current Galaxy. To address the scarcity of these unusual, precious former GC members, we train a neural network (NN) to identify high-[N/O], high-[Al/Fe] stars using low-resolution Gaia BP/RP spectra. Our NN achieves a classification accuracy of approximately $\approx99\%$ and a false positive rate of around $\approx7\%$, identifying 878 new candidates in the Galactic field. We validate our results with several physically-motivated sanity checks, showing, for example, that the incidence of selected stars in Galactic GCs is significantly higher than in the field. Moreover, we find that most of our GC-origin candidates reside in the inner Galaxy, having likely formed in the proto-Milky Way, consistent with previous research. The fraction of GC candidates in the field drops at a metallicity of [Fe/H]$\approx-1$, approximately coinciding with the completion of spin-up, i.e. the formation of the Galactic stellar disk., Comment: 18 pages, 12 figures + 4 in the appendix. Accepted to MNRAS. Alt text included for figures. Comments welcome!

Context: In the context of Galactic archaeology, the outer halo remains relatively unexplored with respect to its metallicity distribution, merger debris, and the abundance of known very/extremely metal-poor ([Fe/H]<-2.5) stars. Aim: We utilize the Pristine survey's publicly available, Pristine data release 1 (PDR1) and Pristine-Gaia synthetic (PGS) catalogues of photometric metallicities, to select Red Giant Branch (RGB) stars in the outer halo. Methods: The RGB selection pipeline selects giants based on the absence of a well-measured parallax in the brightness range where dwarfs have reasonable parallax estimate from Gaia DR3 data. The photometric distances are calculated using a BaSTI-isochrone fitting code and the photometric metallicities. Results: Photometric distances derived from PDR1- and PGS-giants show typical uncertainties of 12% and a scatter of up to 20% and 40% respectively, when validated against inverted-parallax and Starhorse-code distances. The PDR1-giants catalogue provides a low-to-no bias view of the metallicity structure versus distance compared to the PGS-giants catalogue (with a distance-metallicity selection bias), while the PGS-giants catalogue provides an all-sky view of the outer halo, especially in the VMP end. The PDR1-giants catalogue is used to study the metallicity distribution function of the halo out to ~100 kpc, using Gaussian Mixture Model decomposition. Additionally, we use the PDR1-giants with radial velocities from spectroscopic surveys to map the metallicity view of the integrals-of-motion space where accreted dwarf galaxy debris conserve their orbital parameters for a long time. Using the PGS-giants catalogue, we associate 41 stars tentatively to the stellar counterpart of the Magellanic stream in the VMP end, out to 70 kpc. Conclusions: We publish two RGB catalogues between -4.0<[Fe/H]<+0.1 with reliable photometric distances and metallicities., Comment: Submitted to A&A, 25 pages (19 Figures) + 4 pages (6 Figures) in Appendix. Comments are very welcome, enjoy reading! The catalogues will be published upon acceptance

We present a novel method to constrain the formation time of the Milky Way disc using the chrono-kinematic signatures of the inner Galaxy. We construct an O-rich Mira variable sample from the Gaia Long-period Variable catalogue to study the kinematic behaviour of stars with different ages in the inner Galaxy. From the Auriga suite of cosmological zoom-in simulations, we find that the age of the oldest stellar population with imprints of the bar in density and kinematics matches the disc spin-up epoch. This is because stars born before the spin-up show insufficient rotation and are not kinematically cold enough to be efficiently trapped by the bar. We find that the bar kinematic signature disappears for Mira variables with a period shorter than 190 days. Using the period-age relation of Mira variables, we constrain the spin-up epoch of the Milky Way to be younger than $\sim11-12$~Gyr (redshift $\sim3$). We also discuss and compare our method and result to other evidence of the Milky Way spin-up epoch under the context of a realistic age uncertainty. Age uncertainty leads to an overestimation of the disc formation time when performing backward modelling. Our constrain of the spin-up epoch is independent from previous studies because it relies on the kinematics of the inner Galaxy instead of the solar vicinity., Comment: 16 pages, 10 figures. Accepted by MNRAS. DOI: 10.1093/mnras/stae2546

Deep model training on extensive datasets is increasingly becoming cost-prohibitive, prompting the widespread adoption of deep model fusion techniques to leverage knowledge from pre-existing models. From simple weight averaging to more sophisticated methods like AdaMerging, model fusion effectively improves model performance and accelerates the development of new models. However, potential interference between parameters of individual models and the lack of interpretability in the fusion progress remain significant challenges. Existing methods often try to resolve the parameter interference issue by evaluating attributes of parameters, such as their magnitude or sign, or by parameter pruning. In this study, we begin by examining the fine-tuning of linear layers through the lens of subspace analysis and explicitly define parameter interference as an optimization problem to shed light on this subject. Subsequently, we introduce an innovative approach to model fusion called zero-shot Sparse MIxture of Low-rank Experts (SMILE) construction, which allows for the upscaling of source models into an MoE model without extra data or further training. Our approach relies on the observation that fine-tuning mostly keeps the important parts from the pre-training, but it uses less significant or unused areas to adapt to new tasks. Also, the issue of parameter interference, which is intrinsically intractable in the original parameter space, can be managed by expanding the dimensions. We conduct extensive experiments across diverse scenarios, such as image classification and text generation tasks, using full fine-tuning and LoRA fine-tuning, and we apply our method to large language models (CLIP models, Flan-T5 models, and Mistral-7B models), highlighting the adaptability and scalability of SMILE. Code is available at https://github.com/tanganke/fusion_bench, Comment: Code is available at https://github.com/tanganke/fusion_bench