Search

We introduce a generalization of the Kelvin-Voigt model in order to include and characterize heterogeneities in viscoelastic semisolid materials. By considering a microrheological approach, we present analytical expressions for the mean square displacement and for the time-dependent diffusion coefficient of probe particles immersed in a viscoelastic material described by this model. Besides validating our theoretical approach through Brownian dynamics simulations, we show how the model can be used to describe experimental data obtained for polyacrylamide and laponite gels., Comment: 10 pages, 6 figures, 44 references. This manuscript corresponds to the submitted version of the article published as part of the themed collection: Soft Matter Emerging Investigators Series in Soft Matter Journal

We introduce Docling, an easy-to-use, self-contained, MIT-licensed, open-source toolkit for document conversion, that can parse several types of popular document formats into a unified, richly structured representation. It is powered by state-of-the-art specialized AI models for layout analysis (DocLayNet) and table structure recognition (TableFormer), and runs efficiently on commodity hardware in a small resource budget. Docling is released as a Python package and can be used as a Python API or as a CLI tool. Docling's modular architecture and efficient document representation make it easy to implement extensions, new features, models, and customizations. Docling has been already integrated in other popular open-source frameworks (e.g., LangChain, LlamaIndex, spaCy), making it a natural fit for the processing of documents and the development of high-end applications. The open-source community has fully engaged in using, promoting, and developing for Docling, which gathered 10k stars on GitHub in less than a month and was reported as the No. 1 trending repository in GitHub worldwide in November 2024., Comment: Accepted to AAAI 25: Workshop on Open-Source AI for Mainstream Use

This study comprehensively analyzes Atlas's current orbit, focusing on the secular and resonant perturbations caused by Prometheus, Pandora, and Saturn's oblateness. We performed numerical integration of the exact equations of motion for a dense ensemble of Atlas clone satellites. Through spectral analysis and interpretation of these orbits on dynamical maps, we identified the domain of the 54:53 Prometheus-Atlas and 70:67 Pandora-Atlas mean-motion resonances, showing that Atlas lies on the boundary of the separatrices of each of these resonances. We also identified the domains for the multiplets $\Psi_{1}$, $\Psi_{2}$, $\Psi_{3}$ and $\Psi_{4}$ associated with 70:67 resonance. Additionally, we explored the variation in Prometheus's eccentricity, demonstrating that as eccentricity increases (or decreases) in the 54:53 resonance domain correspondingly decreases (or increases). This combined analysis, between the above mappings, revealed qualitatively the overlap between the 54:53 and 70:67 resonances, which are responsible for the chaotic behavior of Atlas's orbit. We quantified chaotic motion in frequency space and found that the vicinity of Atlas is characterized by weak to moderate chaos, rather than strong chaos. Finally, we investigated Atlas's recent past, considering Prometheus's migration under the influence of Saturn's tidal forces. This analysis reveals several resonances crossed in the past, particularly focusing on the Atlas-Prometheus pair, which exhibited a co-orbital configuration., Comment: 29 pages, 16 figures. Published in PSS

Blockchain-based systems are frequently governed through tokens that grant their holders voting rights over core protocol functions and funds. The centralisation occurring in Decentralised Finance (DeFi) protocols' token-based voting systems is typically analysed by examining token holdings' distribution across addresses. In this paper, we expand this perspective by exploring shared token holdings of addresses across multiple DeFi protocols. We construct a Statistically Validated Network (SVN) based on shared governance token holdings among addresses. Using the links within the SVN, we identify influential addresses that shape these connections and we conduct a post-hoc analysis to examine their characteristics and behaviour. Our findings reveal persistent influential links over time, predominantly involving addresses associated with institutional investors who maintain significant token supplies across the sampled protocols. Finally, we observe that token holding patterns and concentrations tend to shift in response to speculative market cycles.

Accurate power forecasting from renewable energy sources (RES) is crucial for integrating additional RES capacity into the power system and realizing sustainability goals. This work emphasizes the importance of integrating decentralized spatio-temporal data into forecasting models. However, decentralized data ownership presents a critical obstacle to the success of such spatio-temporal models, and incentive mechanisms to foster data-sharing need to be considered. The main contributions are a) a comparative analysis of the forecasting models, advocating for efficient and interpretable spline LASSO regression models, and b) a bidding mechanism within the data/analytics market to ensure fair compensation for data providers and enable both buyers and sellers to express their data price requirements. Furthermore, an incentive mechanism for time series forecasting is proposed, effectively incorporating price constraints and preventing redundant feature allocation. Results show significant accuracy improvements and potential monetary gains for data sellers. For wind power data, an average root mean squared error improvement of over 10% was achieved by comparing forecasts generated by the proposal with locally generated ones.

The main challenges limiting the adoption of deep learning-based solutions in medical workflows are the availability of annotated data and the lack of interpretability of such systems. Concept Bottleneck Models (CBMs) tackle the latter by constraining the final disease prediction on a set of predefined and human-interpretable concepts. However, the increased interpretability achieved through these concept-based explanations implies a higher annotation burden. Moreover, if a new concept needs to be added, the whole system needs to be retrained. Inspired by the remarkable performance shown by Large Vision-Language Models (LVLMs) in few-shot settings, we propose a simple, yet effective, methodology, CBVLM, which tackles both of the aforementioned challenges. First, for each concept, we prompt the LVLM to answer if the concept is present in the input image. Then, we ask the LVLM to classify the image based on the previous concept predictions. Moreover, in both stages, we incorporate a retrieval module responsible for selecting the best examples for in-context learning. By grounding the final diagnosis on the predicted concepts, we ensure explainability, and by leveraging the few-shot capabilities of LVLMs, we drastically lower the annotation cost. We validate our approach with extensive experiments across four medical datasets and twelve LVLMs (both generic and medical) and show that CBVLM consistently outperforms CBMs and task-specific supervised methods without requiring any training and using just a few annotated examples. More information on our project page: https://cristianopatricio.github.io/CBVLM/., Comment: This work has been submitted to the IEEE for possible publication

Exoplanets are detected around stars of different ages and birthplaces within the Galaxy. The aim of this work is to infer the Galactic birth radii ($r_\text{birth}$) of stars and, consequently, their planets, with the ultimate goal of studying the Galactic aspects of exoplanet formation. We used photometric, spectroscopic, and astrometric data to estimate the stellar ages of two samples of stars hosting planets and, for comparison, a sample of stars without detected planets. The $r_\text{birth}$ of exoplanets were inferred by projecting stars back to their birth positions based on their estimated age and metallicity [Fe/H]. We find that stars hosting planets have higher [Fe/H], are younger, and have smaller $r_\text{birth}$ compared to stars without detected planets. In particular, stars hosting high-mass planets show higher [Fe/H], are younger, and have smaller $r_\text{birth}$ than stars hosting low-mass planets. We show that the formation efficiency of planets, calculated as the relative frequency of planetary systems, decreases with the galactocentric distance, which relationship is stronger for high-mass planets than for low-mass planets. Additionally, we find that (i) the formation efficiency of high-mass planets increases with time and encompasses a larger galactocentric distance over time; (ii) the formation efficiency of low-mass planets shows a slight increase between the ages of 4 and 8 Gyr and also encompasses a larger galactocentric distance over time; and (iii) stars without detected planets appear to form at larger galactocentric distances over time. We conclude that the formation of exoplanets throughout the Galaxy follows the Galactic chemical evolution, for which our results are in agreement with the observed negative interstellar medium (ISM) metallicity gradient and its enrichment and flattening with time at any radius., Comment: 10 pages, 9 figures, and 1 table

Let $\mathbb{F}_q$ be the finite field with $q=p^s$ elements, where $p$ is an odd prime and $s$ a positive integer. In this paper, we define the function $f(X)=(cX^q+aX)(X^{q}-X)^{n-1}$, for $a,c\in\mathbb{F}_q$ and $n\geq 1$. We study the dynamics of the function $f(X)$ over the finite field $\mathbb{F}_{q^2}$, determining cycle lengths and number of cycles. We also show that all trees attached to cyclic elements are isomorphic, with the exception of the tree hanging from zero. We also present the general shape of such hanging trees, which concludes the complete description of the functional graph of $f(X)$.

This work presents FreeSVC, a promising multilingual singing voice conversion approach that leverages an enhanced VITS model with Speaker-invariant Clustering (SPIN) for better content representation and the State-of-the-Art (SOTA) speaker encoder ECAPA2. FreeSVC incorporates trainable language embeddings to handle multiple languages and employs an advanced speaker encoder to disentangle speaker characteristics from linguistic content. Designed for zero-shot learning, FreeSVC enables cross-lingual singing voice conversion without extensive language-specific training. We demonstrate that a multilingual content extractor is crucial for optimal cross-language conversion. Our source code and models are publicly available.

Using an interface inserted in a background mesh is an alternative way of constructing a complex geometrical shape with a relative low meshing efforts. However, this process may require special treatment of elements cut by the interface. Our study focuses on comparing the integration of cut elements defined by implicit and parametric curves. We investigate the efficiency and robustness of open-source tools such as Algoim [5](a library for quadrature on implicitly defined geometries) and Ginkgo [2](a library for isogeometric analysis on Boolean operations with a parametric description) with numerical examples computing the area defined by the interface and benchmarks for 2D elasticity problem using the open-source code GeoPDEs [7]. It is concluded that none of the two interface descriptions is preferable with respect to the quality of the integration. Thus, the choice of the interface type depends only on the studied problem and the available curve description, but not on the numerical aspects of the integration., Comment: ECCOMAS 2024

We present the experimental production and characterization of a dense cold atomic cloud of \(^{88}\text{Sr}\) atoms, optimized for the future studies of light transport in highly dense regimes. Using narrow-line molasses on the 689 nm transition, combined with a far off-resonant optical dipole trap, we achieve spatial densities as high as \(7.9 \times 10^{13} \, \text{atoms/cm}^3\) and optical depths up to 64. This approach stands out from previous methods by integrating narrow-line molasses with an optical dipole trap, enabling high-density samples without relying on evaporative cooling. Unlike traditional absorption imaging, which becomes inaccurate in such dense regimes, we demonstrate that phase-contrast imaging (PCI) can reliably reconstruct the in-situ density profile even for highly spatially and optically dense samples. The use of a spatial light modulator instead of a fixed phase plate in the PCI setup provides enhanced flexibility and control of imaging parameters, making this imaging technique robust against imaging artifacts and adaptable to varying experimental conditions. Moreover, we derive theoretical conditions for reliable PCI operation in dense regimes and validate these experimentally, showing excellent agreement with time-of-flight measurements even at the highest densities. Our results establish a robust method for producing and characterizing dense atomic clouds., Comment: 11 pages, 7 figures

This work describes our group's submission to the PROCESS Challenge 2024, with the goal of assessing cognitive decline through spontaneous speech, using three guided clinical tasks. This joint effort followed a holistic approach, encompassing both knowledge-based acoustic and text-based feature sets, as well as LLM-based macrolinguistic descriptors, pause-based acoustic biomarkers, and multiple neural representations (e.g., LongFormer, ECAPA-TDNN, and Trillson embeddings). Combining these feature sets with different classifiers resulted in a large pool of models, from which we selected those that provided the best balance between train, development, and individual class performance. Our results show that our best performing systems correspond to combinations of models that are complementary to each other, relying on acoustic and textual information from all three clinical tasks.

We investigate a cosmological model inspired by hybrid inflation, where two scalar fields representing dark energy (DE) and dark matter (DM) interact through a coupling that is proportional to the DE scalar field $1/\phi$. The strength of the coupling is governed solely by the initial condition of the scalar field, $\phi_i$, which parametrises deviations from the standard $\Lambda$CDM model. In this model, the scalar field tracks the behaviour of DM during matter-domination until it transitions to DE while the DM component decays quicker than standard CDM during matter-domination, and is therefore different from some interacting DM-DE models which behaves like phantom dark energy. Using \textit{Planck} 2018 CMB data, DESI BAO measurements and Pantheon+ supernova observations, we find that the model allows for an increase in $H_0$ that can help reduce the Hubble tension. In addition, we find that higher values of the coupling parameter are correlated with lower values of $\omega_m$, and a mild decrease of the weak-lensing parameter $S_8$, potentially relevant to address the $S_8$ tension. Bayesian model comparison, however, reveals inconclusive results for most datasets, unless S$H_0$ES data are included, in which case a moderate evidence in favour of the hybrid model is found., Comment: V1: 17 pages, 9 figures, 6 tables

The Inverse Seesaw mechanism remains one of the most attractive explanations for the lightness of neutrino masses, allowing for natural low-scale realisations. We consider the prospects of a simple extension via 3 generations of sterile fermions - the so called ISS(3,3) - in what concerns numerous lepton flavour observables. In order to facilitate a connection between the Lagrangian parameters and low-energy data, we systematically develop new parametrisations of the Yukawa couplings. Relying on these new parametrisations to explore the parameter space, we discuss the complementary role of charged lepton flavour violation searches in dedicated facilities, as well as in lepton colliders (FCC-ee and $\mu$TRISTAN). Our results reveal the strong synergy of the different indirect searches in probing the distinct flavour sectors of the model. In particular, we show that in the absence of radiative decays $\ell_\alpha\to\ell_\beta\gamma$, sizeable rates for $Z$-penguin dominated observables could hint at a non-trivially mixed and non-degenerate heavy spectrum., Comment: 17 pages, 5 figures

With the growing interest in underwater exploration and monitoring, Autonomous Underwater Vehicles (AUVs) have become essential. The recent interest in onboard Deep Learning (DL) has advanced real-time environmental interaction capabilities relying on efficient and accurate vision-based DL models. However, the predominant use of sonar in underwater environments, characterized by limited training data and inherent noise, poses challenges to model robustness. This autonomy improvement raises safety concerns for deploying such models during underwater operations, potentially leading to hazardous situations. This paper aims to provide the first comprehensive overview of sonar-based DL under the scope of robustness. It studies sonar-based DL perception task models, such as classification, object detection, segmentation, and SLAM. Furthermore, the paper systematizes sonar-based state-of-the-art datasets, simulators, and robustness methods such as neural network verification, out-of-distribution, and adversarial attacks. This paper highlights the lack of robustness in sonar-based DL research and suggests future research pathways, notably establishing a baseline sonar-based dataset and bridging the simulation-to-reality gap.

The use of exogenous pulmonary surfactant (EPS) to deliver other relevant drugs to the lung is a promising strategy for combined therapy. We evaluated the interaction of polymyxin B (PxB) with clinically used EPS, the poractant alfa Curosurf (PSUR). The effect of PxB on the protein-free model system (MS) composed of four phospholipids (diC16:0PC/16:0-18:1PC/16:0-18:2PC/16:0-18:1PG) was examined in parallel to distinguish the specificity of the composition of PSUR. We used several experimental techniques (differential scanning calorimetry, small-and wide-angle X-ray scattering, small angle neutron scattering, fluorescence spectroscopy, and electrophoretic light scattering) to characterize the binding of PxB to both EPS. Electrostatic interactions PxB -EPS are dominant. The results obtained support the concept of cationic PxB molecules lying on the surface of the PSUR bilayer, strengthening the multilamellar structure of the PSUR as derived from SAXS and SANS. A protein-free MS mimics natural EPS well but was found to be less resistant to penetration of PxB into the lipid bilayer. PxB does not affect the gel-to-fluid phase transition temperature Tm of PSUR, while Tm increased by ~ +2 $^\circ$C in MS. The decrease of the thickness of the lipid bilayer (dL) of PSUR upon PxB binding is negligible. The hydrophobic tail of the PxB molecule does not penetrate the bilayer as derived from SANS data analysis and changes in lateral pressure monitored by excimer fluorescence at two depths of the hydrophobic region of the bilayer. Changes in dL of protein-free MS show a biphasic dependence on the adsorbed amount of PxB with a minimum close to the point of electroneutrality of the mixture. Our results do not discourage the concept of a combined treatment with PxBenriched Curosurf. However, the amount of PxB must be carefully assessed (less than 5 wt% relative to the mass of the surfactant) to avoid inversion of the surface charge of the membrane.

Within the context of heavy neutral lepton (HNL) extensions of the Standard Model, we compute the cross-sections for $\mu^+ e^-\to \ell_\alpha^+\ell_\beta^-$ scattering, as well as several angular observables. In particular, we investigate the future sensitivity of a $\mu$TRISTAN collider in discovering such charged lepton flavour violating processes and the potential constraining power of these searches on the parameter space of HNL models. Our results show that while low-energy probes of $\mu-e$ flavour violation do offer the most promising potential, the prospects for $e\tau$ and $\mu\tau$ flavour violation searches at $\mu$TRISTAN can exceed those of related low-energy probes (as well as flavour violating $Z$-pole processes at FCC-ee) by several orders of magnitude., Comment: 31 pages, 13 figures

Safety filters ensure that control actions that are executed are always safe, no matter the controller in question. Previous work has proposed a simple and stealthy false-data injection attack for deactivating such safety filters. This attack injects false sensor measurements to bias state estimates toward the interior of a safety region, making the safety filter accept unsafe control actions. The attack does, however, require the adversary to know the dynamics of the system, the safety region used in the safety filter, and the observer gain. In this work we relax these requirements and show how a similar data-injection attack can be performed when the adversary only observes the input and output of the observer that is used by the safety filter, without any a priori knowledge about the system dynamics, safety region, or observer gain. In particular, the adversary uses the observed data to identify a state-space model that describes the observer dynamics, and then approximates a safety region in the identified embedding. We exemplify the data-driven attack on an inverted pendulum, where we show how the attack can make the system leave a safe set, even when a safety filter is supposed to stop this from happening.

The thermodynamic behavior of mixed systems containing the anionic surfactant sodium dodecyl sulfate (SDS) and the nonionic surfactant polyethylene glycol dodecyl ether (Brij L4) in aqueous solutions was investigated. Electrical conductivity and interfacial tension measurements were employed to investigate the concentration-dependent properties of these surfactant mixtures. Two main experiments were conducted: i) constant ratio experiment: the overall surfactant concentration was varied while maintaining a fixed ratio between SDS and Brij L4. It was shown that the critical micelle concentration (CMC) determined from electrical conductivity measurements does not indicate the formation of the first mixed micelles as observed using tensiometry, but the point from which the adsorption of counterions by the existing micelles became important. By using the Regular Solution Theory (RST), it was found that the interaction between SDS and Brij L4 is synergistic, driven by dipole-ion interactions of the hydrophilic regions of the two surfactants. ii) Fixed SDS concentration while increase the Brij L4 concentration: the resulting electrical conductivity exhibited non-trivial behavior which complexity arose from simultaneous phenomena of incorporation of free SDS molecules into the micelles as Brij L4 concentration increased, leading to a decrease in electrical conductivity; liberation of some adsorbed counterions from the mixed micelles to the solution, which increases the conductivity and increase of viscosity due to Brij L4 further addition that leads to a sequential decrease in electrical conductivity., Comment: 15 pages, 8 figures. Supplementary Information included

This paper addresses the state estimation problem in continuous LTI systems under attacks with non-periodic and asynchronous sampled measurements. The non-periodic and asynchronous sampling requires sensors to transmit not only the measurement values but also the sampling time-stamps to the fusion center via unprotected communication channels. This communication scheme leaves the system vulnerable to a variety of malicious activities such as (i) manipulating measurement values, (ii) manipulating time-stamps, (iii) hybrid manipulations such as generating fake measurements or eliminating the measurement. To deal with such more powerful attacks, we propose a decentralized local estimation algorithm where each sensor maintains its local state estimate based on its measurements in an asynchronous fashion. The local states are synchronized by time-prediction and fused in an event-triggered manner. In the absence of attacks, local estimates are proved to recover the optimal Kalman estimation by our carefully designed weighted least square problem, given that the sample time is non-pathological. In the presence of attacks, an $\ell_1$ regularized least square problem is proposed to generate secure estimates with uniformly bounded error as long as the observability redundancy is satisfied. The effectiveness of the proposed algorithm is demonstrated through a benchmark example of the IEEE 14-bus system., Comment: 9 pages and 6 figures. arXiv admin note: text overlap with arXiv:2303.17514

Retrieval Augmented Generation (RAG) systems are a widespread application of Large Language Models (LLMs) in the industry. While many tools exist empowering developers to build their own systems, measuring their performance locally, with datasets reflective of the system's use cases, is a technological challenge. Solutions to this problem range from non-specific and cheap (most public datasets) to specific and costly (generating data from local documents). In this paper, we show that using public question and answer (Q&A) datasets to assess retrieval performance can lead to non-optimal systems design, and that common tools for RAG dataset generation can lead to unbalanced data. We propose solutions to these issues based on the characterization of RAG datasets through labels and through label-targeted data generation. Finally, we show that fine-tuned small LLMs can efficiently generate Q&A datasets. We believe that these observations are invaluable to the know-your-data step of RAG systems development., Comment: to be published in the 31st International Conference on Computational Linguistics (COLING 2025)

Matrices can be decomposed via rank-one approximations: the best rank-one approximation is a singular vector pair, and the singular value decomposition writes a matrix as a sum of singular vector pairs. The singular vector tuples of a tensor are the critical points of its best rank-one approximation problem. In this paper, we study tensors that can be decomposed via successive rank-one approximations: compute a singular vector tuple, subtract it off, compute a singular vector tuple of the new deflated tensor, and repeat. The number of terms in such a decomposition may exceed the tensor rank. Moreover, the decomposition may depend on the order in which terms are subtracted. We show that the decomposition is valid independent of order if and only if all singular vectors in the process are orthogonal in at least two factors. We study the variety of such tensors. We lower bound its dimension, showing that it is significantly larger than the variety of odeco tensors., Comment: 26 pages, 1 figure

This paper addresses the security allocation problem in a networked control system under stealthy injection attacks. The networked system is comprised of interconnected subsystems which are represented by nodes in a digraph. An adversary compromises the system by injecting false data into several nodes with the aim of maximally disrupting the performance of the network while remaining stealthy to a defender. To minimize the impact of such stealthy attacks, the defender, with limited knowledge about attack policies and attack resources, allocates several sensors on nodes to impose the stealthiness constraint governing the attack policy. We provide an optimal security allocation algorithm to minimize the expected attack impact on the entire network. Furthermore, under a suitable local control design, the proposed security allocation algorithm can be executed in a scalable way. Finally, the obtained results are validated through several numerical examples., Comment: 8 pages, 5 figures, journal submission

CHANCES, the CHileAN Cluster galaxy Evolution Survey, will study the evolution of galaxies in and around ${\sim}$150 massive galaxy clusters, from the local universe out to z=0.45. CHANCES will use the new 4MOST Spectroscopic Survey Facility on the VISTA 4m telescope to obtain spectra for ${\sim}$500,000 galaxies with magnitudes $r_\mathrm{AB} < 20.5$, providing comprehensive spectroscopic coverage of each cluster out to $5r_{200}$. Its wide and deep scope will trace massive and dwarf galaxies from the surrounding filaments and groups to the cores of galaxy clusters, enabling the study of galaxy pre-processing and the role of the evolving environment on galaxy evolution. In this paper we present and characterize the sample of clusters and superclusters to be targeted by CHANCES. We used literature catalogues based on X-ray emission and Sunyaev-Zel'dovich effect to define the cluster sample in a homogeneous way, with attention to cluster mass and redshift, as well as the availability of ancillary data. We calibrated literature mass estimates from various surveys against each other and provide an initial mass estimate for each cluster, which we used to define the radial extent of the 4MOST coverage. We also present an initial assessment of the structure surrounding these clusters based on the redMaPPer red-sequence algorithm as a preview of some of the science CHANCES will enable., Comment: 11 pages, 9 figures, plus references and appendix containing catalog tables, submitted to A&A

Effective structural assessment of urban infrastructure is essential for sustainable land use and resilience to climate change and natural hazards. Seismic wave methods are widely applied in these areas for subsurface characterization and monitoring, yet they often rely on time-consuming inversion techniques that fall short in delivering comprehensive geological, hydrogeological, and geomechanical descriptions. Here, we explore the effectiveness of a passive seismic approach coupled with artificial intelligence (AI) for monitoring geological structures and hydrogeological conditions in the context of sinkhole hazard assessment. We introduce a deterministic petrophysical inversion technique based on a language model that decodes seismic wave velocity measurements to infer soil petrophysical and mechanical parameters as textual descriptions. Results successfully delineate 3D subsurface structures with their respective soil nature and mechanical characteristics, while accurately predicting daily water table levels. Validation demonstrates high accuracy, with a normalized root mean square error of 8%, closely rivaling with conventional stochastic seismic inversion methods, while delivering broader insights into subsurface conditions 2,000 times faster. These findings underscore the potential of advanced AI techniques to significantly enhance subsurface characterization across diverse scales, supporting decision-making for natural hazard mitigation., Comment: 70 pages, 33 figures

The main challenges hindering the adoption of deep learning-based systems in clinical settings are the scarcity of annotated data and the lack of interpretability and trust in these systems. Concept Bottleneck Models (CBMs) offer inherent interpretability by constraining the final disease prediction on a set of human-understandable concepts. However, this inherent interpretability comes at the cost of greater annotation burden. Additionally, adding new concepts requires retraining the entire system. In this work, we introduce a novel two-step methodology that addresses both of these challenges. By simulating the two stages of a CBM, we utilize a pretrained Vision Language Model (VLM) to automatically predict clinical concepts, and a Large Language Model (LLM) to generate disease diagnoses based on the predicted concepts. We validate our approach on three skin lesion datasets, demonstrating that it outperforms traditional CBMs and state-of-the-art explainable methods, all without requiring any training and utilizing only a few annotated examples. The code is available at https://github.com/CristianoPatricio/2-step-concept-based-skin-diagnosis., Comment: Preprint submitted for review

Recent advances in large language models (LLMs) have led to significant progress in robotics, enabling embodied agents to better understand and execute open-ended tasks. However, existing approaches using LLMs face limitations in grounding their outputs within the physical environment and aligning with the capabilities of the robot. This challenge becomes even more pronounced with smaller language models, which are more computationally efficient but less robust in task planning and execution. In this paper, we present a novel modular architecture designed to enhance the robustness of LLM-driven robotics by addressing these grounding and alignment issues. We formalize the task planning problem within a goal-conditioned POMDP framework, identify key failure modes in LLM-driven planning, and propose targeted design principles to mitigate these issues. Our architecture introduces an ``expected outcomes'' module to prevent mischaracterization of subgoals and a feedback mechanism to enable real-time error recovery. Experimental results, both in simulation and on physical robots, demonstrate that our approach significantly improves task success rates for pick-and-place and manipulation tasks compared to both larger LLMs and standard baselines. Through hardware experiments, we also demonstrate how our architecture can be run efficiently and locally. This work highlights the potential of smaller, locally-executable LLMs in robotics and provides a scalable, efficient solution for robust task execution., Comment: Submitted to ICRA 2025

In this paper, we introduce a new fraud-proof algorithm that offers an unprecedented combination of decentralization, security, and liveness. The resources that must be mobilized by an honest participant to defeat an adversary grow only logarithmically with what the adversary ultimately loses. As a consequence, there is no need to introduce high bonds that prevent an adversary from creating too many Sybils. This makes the system very inclusive and frees participants from having to pool resources among themselves to engage the protocol. Finally, the maximum delay to finalization also grows only logarithmically with total adversarial expenditure, with the smallest multiplicative factor to date. In summary: the entire dispute completes in 2--5 challenge periods, the only way to break consensus is to censor the honest party for more than one challenge period, and the costs of engaging in the dispute are minimal.

The Variational Quantum Eigensolver (VQE) is a promising hybrid algorithm, utilizing both quantum and classical computers to obtain the ground state energy of molecules. In this context, this study applies VQE to investigate the ground state of protocatechuic acid, analyzing its performance with various Ans\"atze and active spaces. Subsequently, all VQE results were compared to those obtained with the Hartree-Fock (HF) method. The results demonstrate that Ans\"atze, like unitary coupled-cluster singles and doubles (UCCSD) and variations of Hardware-Efficient, generally achieve accuracy close to that of HF. Furthermore, the increase in active space has led to the models becoming more difficult to converge to values with a greater distance from the correct energy. In summary, the findings of this study reinforce the use of VQE as a powerful tool for analyzing molecular ground state energies. Finally, the results underscore the critical importance of Ansatz selection and active space size in VQE performance, providing valuable insights into its potential and limitations., Comment: 10 pages, 6 figures, 1 table

Noble gas radiation detectors with optical readout are gaining popularity in fields like astrophysics and particle physics due to their ability to produce both ionization and scintillation signals in response to ionizing radiation interaction. In addition, the amplification of primary ionization signals can be achieved by promoting secondary scintillation in the gas. Noble gas purity, especially concerning impurities like H$_{2}$O, N$_{2}$, O$_{2}$, CO$_{2}$, and hydrocarbons, greatly influences its performance. These impurities can cause the loss of primary electrons and quench the scintillation signal. A very high purity level of the gas is required. In the early 90's, a simple method was developed for noble gas purification in sealed, small volume (up to few litters) gas radiation detectors. Gas purification is achieved promoting gas circulation through Zr-based hot getters, simply maintaining the gas circulation by convection. The effectiveness of this method has been only confirmed by the energy resolutions achieved in those detectors, which were similar to that achieved in other high-performance noble gas detectors. In this work, we used waveform analysis of the primary and secondary scintillation signals and we were able to evaluate the impact of the attachment and quenching caused by impurities in one of our detectors filled with pure Xe, and estimate upper values for the impurity content in the gas. The maximum overall impurity concentration was estimated to be below 4 ppm, considering nearly all the impurities, and below 82 ppm if N$_{2}$ is considered. The electron lifetime was measured to be 2.1 $\pm$ 0.1 ms, in line with those achieved in other high-performance optical detectors.

Sensing will be a key technology in 6G networks, enabling a plethora of new sensing-enabled use cases. Some of the use cases relate to deployments over a wide physical area that needs to be sensed by multiple sensing sources at different locations. The efficient management of the sensing resources is pivotal for sustainable sensing-enabled mobile network designs. In this paper, we provide an example of such use case, and show the energy consumption due to sensing has potential to scale to prohibitive levels. We then propose architectural enhancements to solve this problem, and discuss energy saving and energy efficient strategies in sensing, that can only be properly quantified and applied with the proposed architectural enhancements., Comment: 6 pages, 2 figures

Modern scientific discovery increasingly relies on high-performance computing for complex modeling and simulation. A key challenge in improving parallel program performance is efficiently mapping tasks to processors and data to memory, a process dictated by intricate, low-level system code known as mappers. Developing high-performance mappers demands days of manual tuning, posing a significant barrier for domain scientists without systems expertise. We introduce a framework that automates mapper development with generative optimization, leveraging richer feedback beyond scalar performance metrics. Our approach features the Agent-System Interface, which includes a Domain-Specific Language (DSL) to abstract away low-level complexity of system code and define a structured search space, as well as AutoGuide, a mechanism that interprets raw execution output into actionable feedback. Unlike traditional reinforcement learning methods such as OpenTuner, which rely solely on scalar feedback, our method finds superior mappers in far fewer iterations. With just 10 iterations, it outperforms OpenTuner even after 1000 iterations, achieving 3.8X faster performance. Our approach finds mappers that surpass expert-written mappers by up to 1.34X speedup across nine benchmarks while reducing tuning time from days to minutes.

How do networks of relationships evolve over time? We analyse a dataset tracking the social interactions of 900 individuals over four years. Despite continuous shifts in individual relationships, the macroscopic structural properties of the network remain stable, fluctuating within predictable bounds. We connect this stability to the concept of equilibrium in statistical physics. Specifically, we demonstrate that the probabilities governing network dynamics are stationary over time, and key features like degree, edge, and triangle abundances align with theoretical predictions from equilibrium dynamics. Moreover, the dynamics satisfies the detailed balance condition. Remarkably, equilibrium persists despite constant turnover as people join, leave, and change connections. This suggests that equilibrium arises not from specific individuals but from the balancing act of human needs, cognitive limits, and social pressures. Practically, this equilibrium simplifies data collection, supports methods relying on single network snapshots (like Exponential Random Graph Models), and aids in designing interventions for social challenges. Theoretically, it offers new insights into collective human behaviour, revealing how emergent properties of complex social systems can be captured by simple mathematical models., Comment: 14 pages, 3 figures, under peer-review

This paper introduces ROSAR, a novel framework enhancing the robustness of deep learning object detection models tailored for side-scan sonar (SSS) images, generated by autonomous underwater vehicles using sonar sensors. By extending our prior work on knowledge distillation (KD), this framework integrates KD with adversarial retraining to address the dual challenges of model efficiency and robustness against SSS noises. We introduce three novel, publicly available SSS datasets, capturing different sonar setups and noise conditions. We propose and formalize two SSS safety properties and utilize them to generate adversarial datasets for retraining. Through a comparative analysis of projected gradient descent (PGD) and patch-based adversarial attacks, ROSAR demonstrates significant improvements in model robustness and detection accuracy under SSS-specific conditions, enhancing the model's robustness by up to 1.85%. ROSAR is available at https://github.com/remaro-network/ROSAR-framework.

Major Depressive Disorder and anxiety disorders affect millions globally, contributing significantly to the burden of mental health issues. Early screening is crucial for effective intervention, as timely identification of mental health issues can significantly improve treatment outcomes. Artificial intelligence (AI) can be valuable for improving the screening of mental disorders, enabling early intervention and better treatment outcomes. AI-driven screening can leverage the analysis of multiple data sources, including facial features in digital images. However, existing methods often rely on controlled environments or specialized equipment, limiting their broad applicability. This study explores the potential of AI models for ubiquitous depression-anxiety screening given face-centric selfies. The investigation focuses on high-risk pregnant patients, a population that is particularly vulnerable to mental health issues. To cope with limited training data resulting from our clinical setup, pre-trained models were utilized in two different approaches: fine-tuning convolutional neural networks (CNNs) originally designed for facial expression recognition and employing vision-language models (VLMs) for zero-shot analysis of facial expressions. Experimental results indicate that the proposed VLM-based method significantly outperforms CNNs, achieving an accuracy of 77.6%. Although there is significant room for improvement, the results suggest that VLMs can be a promising approach for mental health screening., Comment: This article has been accepted for publication in HEALTHINF25 at the 18th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2025)

Solitons in liquid crystals are spatially localised stable configuration of the liquid crystal orientational order parameter that exhibit emergent particle-like properties such as mutual interaction, translational motion and reconfigurable self-asembling behaviour in oscillating electric fields. Understanding solitons' out-of-equilibrium interactions remains a significant challenge. In this study, we present a minimalist model capable to address this challenge. We derive analytically elastic multipole interaction potentials between two solitons assuming two dimensional LC configurations. We determine the soliton's elastic dipole and quadrupole moments based on the equilibrium director textures obtained as minimizers of the Frank-Oseen free energy functional. Incorporating the time dependence of the soliton's multipole moments in response to electric field oscillations, we carry out particle-based simulations that demonstrate the formation of soliton chains and two dimensional clusters by varying the amplitude and/or the frequency of the electric field, in agreement with experiments. The presented formalism may serve as a benchmark model for future theoretical developments incorporating, e.g. higher order elastic multipoles or/and many body interactions.

We revisit the Bootstrap Percolation model, leveraging recent mathematical advances linking it with its local counterpart. This new perspective resolves, for the first time, historic discrepancies between Monte Carlo simulations and theoretical results: previously, those predictions disagreed even in the first-order asymptotics of the model. In contrast, our framework achieves excellent agreement between numerics and theory, which now match up to the third-order expansion, as the infection probability approaches zero. Our algorithm allows us to generate novel predictions for the model., Comment: 6 pages, 2 figures, 3 page supplemental material

This book provides a self-contained introduction to geometric group theory. The topics range from an introduction of Cayley and Schreier graphs to Gromov's theorem on groups of polynomial growth and amenability. We discuss the ping-pong lemma, quasi-isometries, growth of groups, hyperbolicity, and other related notions. The book is based on graduate courses and can be used for such a course or for independent study., Comment: 277 pages, in Portuguese language

We perform a complete classification of all 56 subgroups of the two-qubit Clifford group containing the two-qubit Pauli group. We provide generators for these groups using gates familiar to the quantum information community and we reference these groups against the group libraries provided in GAP. We also list several families of groups in higher levels of the two-qubit Clifford hierarchy.

We consider random walks in dynamic random environments and propose a criterion which, if satisfied, allows to decompose the random walk trajectory into i.i.d. increments, and ultimately to prove limit theorems. The criterion involves the construction of a random field built from the environment, that has to satisfy a certain random Markov property along with some mixing estimates. We apply this criterion to correlated environments such as Boolean percolation and renewal chains featuring polynomial decay of correlations.

We design flexible fault tolerant gate gadgets that allow the data and the ancilla to be encoded using different codes. By picking a stabilizer code for the ancilla we are able to perform both Clifford and non-Clifford gates fault tolerantly on generic quantum codes, including both stabilizer and non-additive codes. This allows us to demonstrate the first universal fault tolerant gate set for non-additive codes. We consider fault tolerance both with respect to a dephasing channel and a depolarizing channel.

We developed a jitted compiler for training Artificial Neural Networks using C++, LLVM and Cuda. It features object-oriented characteristics, strong typing, parallel workers for data pre-processing, pythonic syntax for expressions, PyTorch like model declaration and Automatic Differentiation. We implement the mechanisms of cache and pooling in order to manage VRAM, cuBLAS for high performance matrix multiplication and cuDNN for convolutional layers. Our experiments with Residual Convolutional Neural Networks on ImageNet, we reach similar speed but degraded performance. Also, the GRU network experiments show similar accuracy, but our compiler have degraded speed in that task. However, our compiler demonstrates promising results at the CIFAR-10 benchmark, in which we reach the same performance and about the same speed as PyTorch. We make the code publicly available at: https://github.com/NoSavedDATA/NoSavedKaleidoscope, Comment: 12 pages, 3 figures and 3 tables

Microgravity mass gauging has gained increasing importance in recent years due to the acceleration in planning for long-term space missions as well as in-space refueling and transfer operations. It is of particular importance with cryogenic propellants where periodic tank venting maneuvers and leak detection place a special emphasis on accurate mass gauging. Several competing technologies have arisen, but capacitance mass gauging has several distinct advantages due to its low mass, non-intrusiveness, and whole volume interrogation technique. Capacitance based measurement has also seen recent success in measuring cryogenic liquid nitrogen and hydrogen volume fraction and flow rate, showing its compatibility with cryogenic propellants. However, the effects of gravity on fluid behavior make the calibration and testing of these sensors difficult on the ground. In this paper a prototype sensor is constructed that can emulate fluid positions in microgravity and earth gravity configurations. Experimental propellant fills and drains are conducted using a simulant fluid with similar electrical properties to cryogenic propellants. This expanded dataset is compared with previous simulation results and used to construct a machine learning model capable of calculating the fluid mass in tanks both with and without propellant management devices., Comment: 6 figures, 8 references

This study presents the first global analysis of on-demand video streaming over Low Earth Orbit (LEO) satellite networks, using data from over one million households across 85 countries. We highlight Starlink's role as a major LEO provider, enhancing connectivity in underserved regions. Our findings reveal that while overall video quality on Starlink matches that of traditional networks, the inherent variability in LEO conditions -- such as throughput fluctuations and packet loss -- leads to an increase in bitrate switches and rebuffers. To further improve the quality of experience for the LEO community, we manipulate existing congestion control and adaptive bitrate streaming algorithms using simulation and real A/B tests deployed on over one million households. Our results underscore the need for video streaming and congestion control algorithms to adapt to rapidly evolving network landscapes, ensuring high-quality service across diverse and dynamic network types.

Strong collective interaction of atoms with an optical cavity causes normal mode splitting of the cavity's resonances, whose width is given by the collective coupling strength. At low optical density of the atomic cloud the intensity distribution of light in the cavity is ruled by the cavity's mode function, which is solely determined by its geometry. In this regime the dynamics of the coupled atom-cavity system is conveniently described by the open Dicke model, which we apply to calculating normal mode splitting generated by periodically ordered clouds in linear and ring cavities. We also show how to use normal mode splitting as witness for Wannier-Bloch oscillations in the tight-binding limit. At high optical density the atomic distribution contributes to shaping the mode function. This regime escapes the open Dicke model, but can be treated by a transfer matrix model provided the saturation parameter is low. Applying this latter model to an atomic cloud periodically ordered into a one-dimensional lattice, we observe the formation of photonic bands gaps competing with the normal mode splitting. We discuss the limitations of both models and point out possible pathways to generalized theories., Comment: 10 pages, 14 figures

Cell cortex contraction is essential for shaping cells, enabling movement, ensuring proper division, maintaining tissue integrity, guiding development, and responding to mechanical signals - all critical for the life and health of multicellular organisms. Differential contractions in cell membranes, particularly when cells of different types interact, play a crucial role in the emergence of segregation. In this study, we introduce a model where rings composed of active particles interact through differential membrane contraction within a specified cutoff distance. We demonstrate that segregation arises solely from differential contraction, with the activity of the rings functioning similarly to an effective temperature. Additionally, we observed that segregation involves cluster fusion-diffusion process. However, the decay exponent of the segregation parameter we found is close to $\lambda \sim -1/3$, which differs from the $\lambda \sim -1/4$ predicted by previous theoretical approaches and simulations.

Photometric wide-field surveys are imaging the sky in unprecedented detail. These surveys face a significant challenge in efficiently estimating galactic photometric redshifts while accurately quantifying associated uncertainties. In this work, we address this challenge by exploring the estimation of Probability Density Functions (PDFs) for the photometric redshifts of galaxies across a vast area of 17,000 square degrees, encompassing objects with a median 5$\sigma$ point-source depth of $g$ = 24.3, $r$ = 23.9, $i$ = 23.5, and $z$ = 22.8 mag. Our approach uses deep learning, specifically integrating a Recurrent Neural Network architecture with a Mixture Density Network, to leverage magnitudes and colors as input features for constructing photometric redshift PDFs across the whole DECam Local Volume Exploration (DELVE) survey sky footprint. Subsequently, we rigorously evaluate the reliability and robustness of our estimation methodology, gauging its performance against other well-established machine learning methods to ensure the quality of our redshift estimations. Our best results constrain photometric redshifts with the bias of $-0.0013$, a scatter of $0.0293$, and an outlier fraction of $5.1\%$. These point estimates are accompanied by well-calibrated PDFs evaluated using diagnostic tools such as Probability Integral Transform and Odds distribution. We also address the problem of the accessibility of PDFs in terms of disk space storage and the time demand required to generate their corresponding parameters. We present a novel Autoencoder model that reduces the size of PDF parameter arrays to one-sixth of their original length, significantly decreasing the time required for PDF generation to one-eighth of the time needed when generating PDFs directly from the magnitudes., Comment: 22 pages, 14 figures