Search

In addition to providing general constraints on probability distributions, fluctuation theorems allow to infer essential information on the role played by temperature in heat exchange phenomena. In this numerical study, we measure the temperature of an out of equilibrium active bath using a fluctuation theorem that relates the fluctuations of the heat exchanged between two baths to their temperatures. Our setup consists of a single particle moving between two wells of a quartic potential accommodating two different baths. The heat exchanged between the two baths is monitored according to two definitions: as the kinetic energy carried by the particle whenever it jumps from one well to the other and as the work performed by the particle on one of the two baths when immersed in it. First, we consider two equilibrium baths at two different temperatures and verify that a fluctuation theorem featuring the baths temperatures holds for both heat definitions. Then, we introduce an additional Gaussian coloured noise in one of the baths, so as to make it effectively an active (out-of-equilibrium) bath. We find that a fluctuation theorem is still satisfied with both heat definitions. Interestingly, in this case the temperature obtained through the fluctuation theorem for the active bath corresponds to the kinetic temperature when considering the first heat definition, while it is larger with the second one. We interpret these results by looking at the particle jump phenomenology., Comment: 29 pages, 7 figures

Due to its inherent intertwinement with irreversibility, entropy production is a prime observable to monitor in systems of active particles. In this numerical study, entropy production in the liquid, hexatic and solid phases of a two-dimensional system of active Brownian particles is examined at both average and fluctuation level. The trends of averages as functions of density show no singularity and marked changes in their derivatives at the hexatic-solid transition. Distributions show instead peculiar tail structures interpreted by looking at microscopic configurations. Particles in regions of low local order generate tail values according to different dynamical mechanisms: they move towards empty regions or bounce back and forth into close neighbours. The tail structures are reproduced by a simple single-particle model including an intermittent harmonic potential.

Multi-modal Large Language Models (MLLMs) are gaining significant attention for their ability to process multi-modal data, providing enhanced contextual understanding of complex problems. MLLMs have demonstrated exceptional capabilities in tasks such as Visual Question Answering (VQA); however, they often struggle with fundamental engineering problems, and there is a scarcity of specialized datasets for training on topics like digital electronics. To address this gap, we propose a benchmark dataset called ElectroVizQA specifically designed to evaluate MLLMs' performance on digital electronic circuit problems commonly found in undergraduate curricula. This dataset, the first of its kind tailored for the VQA task in digital electronics, comprises approximately 626 visual questions, offering a comprehensive overview of digital electronics topics. This paper rigorously assesses the extent to which MLLMs can understand and solve digital electronic circuit questions, providing insights into their capabilities and limitations within this specialized domain. By introducing this benchmark dataset, we aim to motivate further research and development in the application of MLLMs to engineering education, ultimately bridging the performance gap and enhancing the efficacy of these models in technical fields.

Novel devices and novel computing paradigms are key for energy efficient, performant future computing systems. However, designing devices for new applications is often time consuming and tedious. Here, we investigate the design and optimization of spin orbit torque and spin transfer torque magnetic tunnel junction models as the probabilistic devices for true random number generation. We leverage reinforcement learning and evolutionary optimization to vary key device and material properties of the various device models for stochastic operation. Our AI guided codesign methods generated different candidate devices capable of generating stochastic samples for a desired probability distribution, while also minimizing energy usage for the devices.

In question-answering scenarios, humans can assess whether the available information is sufficient and seek additional information if necessary, rather than providing a forced answer. In contrast, Vision Language Models (VLMs) typically generate direct, one-shot responses without evaluating the sufficiency of the information. To investigate this gap, we identify a critical and challenging task in the Visual Question Answering (VQA) scenario: can VLMs indicate how to adjust an image when the visual information is insufficient to answer a question? This capability is especially valuable for assisting visually impaired individuals who often need guidance to capture images correctly. To evaluate this capability of current VLMs, we introduce a human-labeled dataset as a benchmark for this task. Additionally, we present an automated framework that generates synthetic training data by simulating ``where to know'' scenarios. Our empirical results show significant performance improvements in mainstream VLMs when fine-tuned with this synthetic data. This study demonstrates the potential to narrow the gap between information assessment and acquisition in VLMs, bringing their performance closer to humans., Comment: NeurIPS 2024

This paper presents and analyzes two robust, efficient, and optimally accurate fully discrete finite element algorithms for computing the parameterized Navier-Stokes Equations (NSEs) flow ensemble. The timestepping algorithms are linearized, use the backward-Euler method for approximating the temporal derivative, and Ensemble Eddy Viscosity (EEV) regularized. The first algorithm is a coupled ensemble scheme, and the second algorithm is decoupled using projection splitting with grad-div stabilization. We proved the stability and convergence theorems for both algorithms. We have shown that for sufficiently large grad-div stabilization parameters, the outcomes of the projection scheme converge to the outcomes of the coupled scheme. We then combine the Stochastic Collocation Methods (SCMs) with the proposed two Uncertainty Quantification (UQ) algorithms. A series of numerical experiments are given to verify the predicted convergence rates and performance of the schemes on benchmark problems, which shows the superiority of the splitting algorithm., Comment: 27 pages, 4 figures. arXiv admin note: text overlap with arXiv:2310.17779

Magnetic Resonance Imaging (MRI) is the gold standard for measuring fat and iron content non-invasively in the body via measures known as Proton Density Fat Fraction (PDFF) and $R_2^*$, respectively. However, conventional PDFF and $R_2^*$ quantification methods operate on MR images voxel-wise and require at least three measurements to estimate three quantities: water, fat, and $R_2^*$. Alternatively, the two-point Dixon MRI protocol is widely used and fast because it acquires only two measurements; however, these cannot be used to estimate three quantities voxel-wise. Leveraging the fact that neighboring voxels have similar values, we propose using a generative machine learning approach to learn PDFF and $R_2^*$ from Dixon MRI. We use paired Dixon-IDEAL data from UK Biobank in the liver and a Pix2Pix conditional GAN to demonstrate the first large-scale $R_2^*$ imputation from two-point Dixon MRIs. Using our proposed approach, we synthesize PDFF and $R_2^*$ maps that show significantly greater correlation with ground-truth than conventional voxel-wise baselines.

Developing questions that are pedagogically sound, relevant, and promote learning is a challenging and time-consuming task for educators. Modern-day large language models (LLMs) generate high-quality content across multiple domains, potentially helping educators to develop high-quality questions. Automated educational question generation (AEQG) is important in scaling online education catering to a diverse student population. Past attempts at AEQG have shown limited abilities to generate questions at higher cognitive levels. In this study, we examine the ability of five state-of-the-art LLMs of different sizes to generate diverse and high-quality questions of different cognitive levels, as defined by Bloom's taxonomy. We use advanced prompting techniques with varying complexity for AEQG. We conducted expert and LLM-based evaluations to assess the linguistic and pedagogical relevance and quality of the questions. Our findings suggest that LLms can generate relevant and high-quality educational questions of different cognitive levels when prompted with adequate information, although there is a significant variance in the performance of the five LLms considered. We also show that automated evaluation is not on par with human evaluation.

This work investigates the problem of instance-level image retrieval re-ranking with the constraint of memory efficiency, ultimately aiming to limit memory usage to 1KB per image. Departing from the prevalent focus on performance enhancements, this work prioritizes the crucial trade-off between performance and memory requirements. The proposed model uses a transformer-based architecture designed to estimate image-to-image similarity by capturing interactions within and across images based on their local descriptors. A distinctive property of the model is the capability for asymmetric similarity estimation. Database images are represented with a smaller number of descriptors compared to query images, enabling performance improvements without increasing memory consumption. To ensure adaptability across different applications, a universal model is introduced that adjusts to a varying number of local descriptors during the testing phase. Results on standard benchmarks demonstrate the superiority of our approach over both hand-crafted and learned models. In particular, compared with current state-of-the-art methods that overlook their memory footprint, our approach not only attains superior performance but does so with a significantly reduced memory footprint. The code and pretrained models are publicly available at: https://github.com/pavelsuma/ames, Comment: ECCV 2024

We present foundation language models developed to power Apple Intelligence features, including a ~3 billion parameter model designed to run efficiently on devices and a large server-based language model designed for Private Cloud Compute. These models are designed to perform a wide range of tasks efficiently, accurately, and responsibly. This report describes the model architecture, the data used to train the model, the training process, how the models are optimized for inference, and the evaluation results. We highlight our focus on Responsible AI and how the principles are applied throughout the model development.

The transport of DNA polymers through nanoscale pores is central to many biological processes, from bacterial gene exchange to viral infection. In single-molecule nanopore sensing, the detection of nucleic acid and protein analytes relies on the passage of a long biopolymer through a nanoscale aperture. Understanding the dynamics of polymer translocation through nanopores, especially the relation between ionic current signal and polymer conformations is thus essential for the successful identification of targets. Here, by analyzing ionic current traces of dsDNA translocation, we reveal that features up to now uniquely associated with knots are instead different structural motifs: plectonemes. By combining experiments and simulations, we demonstrate that such plectonemes form because of the solvent flow that induces rotation of the helical DNA fragment in the nanopore, causing torsion propagation outwards from the pore. Molecular dynamic simulations reveal that plectoneme initialization is dominated by the applied torque while the translocation time and size of the plectonemes depend on the coupling of torque and pulling force, a mechanism that might also be relevant for in vivo DNA organization. Experiments with nicked DNA constructs show that the number of plectonemes depends on the rotational constraints of the translocating molecules. Thus, our work introduces plectonemes as essential structural features that must be considered for accurate analysis of polymer transport in the nanopore., Comment: 20 pages, 4 figures

This paper introduces Werewolf Arena, a novel framework for evaluating large language models (LLMs) through the lens of the classic social deduction game, Werewolf. In Werewolf Arena, LLMs compete against each other, navigating the game's complex dynamics of deception, deduction, and persuasion. The framework introduces a dynamic turn-taking system based on bidding, mirroring real-world discussions where individuals strategically choose when to speak. We demonstrate the framework's utility through an arena-style tournament featuring Gemini and GPT models. Our results reveal distinct strengths and weaknesses in the models' strategic reasoning and communication. These findings highlight Werewolf Arena's potential as a challenging and scalable LLM benchmark., Comment: 13 pages, 10 figures

DNA unzipping by nanopore translocation has implications in diverse contexts, from polymer physics to single-molecule manipulation to DNA-enzyme interactions in biological systems. Here we use molecular dynamics simulations and a coarse-grained model of DNA to address the nanopore unzipping of DNA filaments that are knotted. This previously unaddressed problem is motivated by the fact that DNA knots inevitably occur in isolated equilibrated filaments and in vivo. We study how different types of tight knots in the DNA segment just outside the pore impact unzipping at different driving forces. We establish three main results. First, knots do not significantly affect the unzipping process at low forces. However, knotted DNAs unzip more slowly and heterogeneously than unknotted ones at high forces. Finally, we observe that the microscopic origin of the hindrance typically involves two concurrent causes: the topological friction of the DNA chain sliding along its knotted contour and the additional friction originating from the entanglement with the newly unzipped DNA. The results reveal a previously unsuspected complexity of the interplay of DNA topology and unzipping, which should be relevant for interpreting nanopore-based single-molecule unzipping experiments and improving the modeling of DNA transactions in vivo., Comment: 11 pages, 6 figures

We present deep kpc- and pc-scale neutral atomic hydrogen (HI) absorption observations of a very young radio source (< 5000 yrs), 4C 31.04, using the WSRT and the Global VLBI array. Using $z=0.0598$, we detect a broad absorption feature centred at the systemic velocity, and narrow absorption redshifted by 220 km/s both previously observed. Additionally, we detect a new blueshifted, broad, shallow absorption wing. At pc scales, the broad absorption at the systemic velocity is detected across the entire radio source while the shallow wing is only seen against part of the eastern lobe. The velocity dispersion of the gas is overall high ($\geq$40 km/s), and is highest (>60 km/s) in the region including the outflow and the radio hot spot. While we detect a velocity gradient along the western lobe and parts of the eastern lobe, most of the gas along the rest of the eastern lobe exhibits no signs of rotation. We therefore conclude that the radio lobes of 4C 31.04 are expanding into a circumnuclear disc, partially disrupting it and making the gas highly turbulent. The distribution of gas is predominantly smooth at the spatial resolution of ~4 pc studied here. However, clumps of gas are also present, particularly along the eastern lobe. This lobe is strongly interacting with the clouds and driving an outflow ~35 pc from the radio core, with a mass-outflow rate of $0.3 \leq \dot{M} \leq 1.4$ M$_\odot$/yr. We compare our observations with a model on the survival of atomic gas clouds in radio-jet-driven outflows and find that the existence of a sub-kpc outflow implies high gas density and inefficient mixing of the cold gas with the hot medium, leading to shorter cooling times. Overall, this provides further evidence of the strong impact of young radio jets on cold ISM and supports the predictions of simulations regarding jet$-$ISM interactions and the nature of the gas into which the jets expand., Comment: 9 pages, 4 figures. Accepted for publication in Astronomy & Astrophysics

In recent years, advancements in deep learning techniques have considerably enhanced the efficiency and accuracy of medical diagnostics. In this work, a novel approach using multi-task learning (MTL) for the simultaneous classification of lung sounds and lung diseases is proposed. Our proposed model leverages MTL with four different deep learning models such as 2D CNN, ResNet50, MobileNet and Densenet to extract relevant features from the lung sound recordings. The ICBHI 2017 Respiratory Sound Database was employed in the current study. The MTL for MobileNet model performed better than the other models considered, with an accuracy of74\% for lung sound analysis and 91\% for lung diseases classification. Results of the experimentation demonstrate the efficacy of our approach in classifying both lung sounds and lung diseases concurrently. In this study,using the demographic data of the patients from the database, risk level computation for Chronic Obstructive Pulmonary Disease is also carried out. For this computation, three machine learning algorithms namely Logistic Regression, SVM and Random Forest classifierswere employed. Among these ML algorithms, the Random Forest classifier had the highest accuracy of 92\%.This work helps in considerably reducing the physician's burden of not just diagnosing the pathology but also effectively communicating to the patient about the possible causes or outcomes.

Close relationships are irreplaceable social resources, yet prone to high-risk conflict. Building on findings from the fields of HCI, virtual reality, and behavioral therapy, we evaluate the unexplored potential of retrospective VR-embodied perspective-taking to fundamentally influence conflict resolution in close others. We develop a biographically-accurate Retrospective Embodied Perspective-Taking system (REPT) and conduct a mixed-methods evaluation of its influence on close others' reflection and communication, compared to video-based reflection methods currently used in therapy (treatment as usual, or TAU). Our key findings provide evidence that REPT was able to significantly improve communication skills and positive sentiment of both partners during conflict, over TAU. The qualitative data also indicated that REPT surpassed basic perspective-taking by exclusively stimulating users to embody and reflect on both their own and their partner's experiences at the same level. In light of these findings, we provide implications and an agenda for social embodiment in HCI design: conceptualizing the use of `embodied social cognition,' and envisioning socially-embodied experiences as an interactive context., Comment: 18 pages, 5 figures, Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems

Yttrium orthosilicate (Y2SiO5 - YSO) is one of the most promising crystals to host rare-earth (RE) ions for quantum technologies applications. In this matrix, they indeed exhibit narrow optical and spin linewidths that can be exploited to develop quantum memories or quantum information processing capabilities. In this paper, we propose a new method to grow RE doped silicate thin films on silicon wafers based on direct liquid injection chemical vapour deposition (DLI-CVD). We optimize the deposition and annealing conditions to achieve formation of the high temperature X2-YSO phase. The phase purity and crystalline quality of the films are assessed by evaluating the optical properties of Eu3+ ions embedded in this oxide matrix. In view of the results, we discuss the possible phase formation mechanisms, and the potential of this new wafer-compatible form of YSO for quantum technologies applications., Comment: 14 pages, 7 figures

Molecular dynamics simulations were employed to investigate the phase separation process of a two-dimensional active Brownian dumbbell model. We evaluated the time dependence of the typical size of the dense component using the scaling properties of the structure factor, along with the averaged number of clusters and their radii of gyration. The growth observed is faster than in active particle (disk) models, and this effect is further enhanced under stronger activity. Next, we focused on studying the hexatic order of the clusters. The length associated to the orientational order grows algebraically and faster than for active Brownian particles. Under weak active forces, most clusters exhibit a uniform internal orientational order. However, under strong forcing, large clusters consist of domains with different orientational orders. We demonstrated that the latter configurations are not stable, and given sufficient time to evolve, they eventually reach homogeneous configurations as well. No gas bubbles are formed within the clusters, even when there are patches of different hexatic order. Finally, attention was directed towards the geometry and motion of the clusters themselves. By employing a tracking algorithm, we showed that clusters smaller than the typical size at the observation time exhibit regular shapes, while larger ones display fractal characteristics. In between collisions or break-ups, the clusters behave as solid bodies. Their centers of mass undergo circular motion, with radii increasing with the cluster size. The center of mass angular velocity equals that of the constituents with respect to their center of mass. These observations were rationalised with a simple mechanical model., Comment: 18 pages, 18 figures

Much progress has been made in elucidating the inner workings of voltage-gated ion channels, but less understood is the influence of lipid rafts on gating kinetics. Here we propose that state-dependent channel affinity for different lipid species provides a unified explanation for the experimentally observed behaviors of clustering, cooperativity, and hysteresis. We develop models of diffusing lipids and channels engaged in Ising-like interactions to investigate the collective behaviors driven by raft formation in critical membranes close to the demixing transition. The model channels demonstrate lipid-mediated long-range interactions, activation curve steepening, and long-term memory in ionic currents. These behaviors likely play a role in channel-mediated cellular signaling and suggest a universal mechanism for self-organization of biomolecular assemblies., Comment: 14 pages, 6 figures