We provide a comprehensive overview of meshfree collocation methods for numerically approximating differential operators on continuously labeled unstructured point clouds. Meshfree collocation methods do not require a computational grid or mesh. Instead, they approximate smooth functions and their derivatives at potentially irregularly distributed collocation points, often called particles, to a desired order of consistency. We review several meshfree collocation methods from the literature, trace the historical development of key concepts, and propose a classification of methods according to their principle of derivation. Although some of the methods reviewed are similar or identical, there are subtle yet important differences between many, which we highlight and discuss. We present a unifying formulation of meshfree collocation methods that renders these differences apparent and show how each method can be derived from this formulation. Finally, we propose a generalized derivation for meshfree collocation methods going forward.

We have seen great progress in basic perceptual tasks such as object recognition and detection. However, AI models still fail to match humans in high-level vision tasks due to the lack of capacities for deeper reasoning. Recently the new task of visual question answering (QA) has been proposed to evaluate a model's capacity for deep image understanding. Previous works have established a loose, global association between QA sentences and images. However, many questions and answers, in practice, relate to local regions in the images. We establish a semantic link between textual descriptions and image regions by object-level grounding. It enables a new type of QA with visual answers, in addition to textual answers used in previous work. We study the visual QA tasks in a grounded setting with a large collection of 7W multiple-choice QA pairs. Furthermore, we evaluate human performance and several baseline models on the QA tasks. Finally, we propose a novel LSTM model with spatial attention to tackle the 7W QA tasks.

Hallucinations in large language models (LLMs), plausible but factually inaccurate text, are often viewed as undesirable. However, recent work suggests that such outputs may hold creative potential. In this paper, we investigate whether hallucinations can improve LLMs on molecule property prediction, a key task in early-stage drug discovery. We prompt LLMs to generate natural language descriptions from molecular SMILES strings and incorporate these often hallucinated descriptions into downstream classification tasks. Evaluating seven instruction-tuned LLMs across five datasets, we find that hallucinations significantly improve predictive accuracy for some models. Notably, Falcon3-Mamba-7B outperforms all baselines when hallucinated text is included, while hallucinations generated by GPT-4o consistently yield the greatest gains between models. We further identify and categorize over 18,000 beneficial hallucinations, with structural misdescriptions emerging as the most impactful type, suggesting that hallucinated statements about molecular structure may increase model confidence. Ablation studies show that larger models benefit more from hallucinations, while temperature has a limited effect. Our findings challenge conventional views of hallucination as purely problematic and suggest new directions for leveraging hallucinations as a useful signal in scientific modeling tasks like drug discovery.

Thermal errors in machine tools significantly impact machining precision and productivity. Traditional thermal error correction/compensation methods rely on measured temperature-deformation fields or on transfer functions. Most existing data-driven compensation strategies employ neural networks (NNs) to directly predict thermal errors or specific compensation values. While effective, these approaches are tightly bound to particular error types, spatial locations, or machine configurations, limiting their generality and adaptability. In this work, we introduce a novel paradigm in which NNs are trained to predict high-fidelity temperature and heat flux fields within the machine tool. The proposed framework enables subsequent computation and correction of a wide range of error types using modular, swappable downstream components. The NN is trained using data obtained with the finite element method under varying initial conditions and incorporates a correlation-based selection strategy that identifies the most informative measurement points, minimising hardware requirements during inference. We further benchmark state-of-the-art time-series NN architectures, namely Recurrent NN, Gated Recurrent Unit, Long-Short Term Memory (LSTM), Bidirectional LSTM, Transformer, and Temporal Convolutional Network, by training both specialised models, tailored for specific initial conditions, and general models, capable of extrapolating to unseen scenarios. The results show accurate and low-cost prediction of temperature and heat flux fields, laying the basis for enabling flexible and generalisable thermal error correction in machine tool environments.

We present a topology grounded, multiscale simulation platform for morphogenesis and biological active matter. Morphogenesis and biological active matter represent keystone problems in biology with additional, far-reaching implications across the biomedical sciences. Addressing these problems will require flexible, cross-scale models of tissue shape, development, and dysfunction that can be tuned to understand, model, and predict relevant individual cases. Current approaches to simulating anatomical or cellular subsystems tend to rely on static, assumed shapes. Meanwhile, the potential for topology to provide natural dimensionality reduction and organization of shape and dynamical outcomes is not fully exploited. TopoSPAM combines ease of use with powerful simulation algorithms and methodological advances, including active nematic gels, topological-defect-driven shape dynamics, and an active 3D vertex model of tissues. It is capable of determining emergent flows and shapes across scales.

Driverless train operation for open tracks on urban guided transport and mainline railways requires, among other things automatic detection of actual and potential obstacles, especially humans, in the danger zone of the train's path. Machine learning algorithms have proven to be powerful state-of-the-art tools for this task. However, these algorithms require large amounts of high-quality annotated data containing human beings in railway-specific environments as training data. Unfortunately, the amount of publicly available datasets is not yet sufficient and is significantly inferior to the datasets in the road domain. Therefore, this paper presents RailGoerl24, an on-board visual light Full HD camera dataset of 12205 frames recorded in a railway test center of TÜV SÜD Rail, in Görlitz, Germany. Its main purpose is to support the development of driverless train operation for guided transport. RailGoerl24 also includes a terrestrial LiDAR scan covering parts of the area used to acquire the RGB data. In addition to the raw data, the dataset contains 33556 boxwise annotations in total for the object class 'person'. The faces of recorded actors are not blurred or altered in any other way. RailGoerl24, available at this http URL, can also be used for tasks beyond collision prediction.

Statistical model checking (SMC) randomly samples probabilistic models to approximate quantities of interest with statistical error guarantees. It is traditionally used to estimate probabilities and expected rewards, i.e. means of different random variables on paths. In this paper, we develop methods using the Dvoretzky-Kiefer-Wolfowitz-Massart inequality (DKW) to extend SMC beyond means to compute quantities such as quantiles, conditional value-at-risk, and entropic risk. The DKW provides confidence bounds on the random variable's entire cumulative distribution function, a much more versatile guarantee compared to the statistical methods prevalent in SMC today. We have implemented support for computing new quantities via the DKW in the 'modes' simulator of the Modest Toolset. We highlight the implementation and its versatility on benchmarks from the quantitative verification literature.

We introduce the Push-Forward Signed Distance Morphometric (PF-SDM), a novel method for shape quantification in biomedical imaging that is continuous, interpretable, and invariant to shape-preserving transformations. PF-SDM effectively captures the geometric properties of shapes, including their topological skeletons and radial symmetries. This results in a robust and interpretable shape descriptor that generalizes to capture temporal shape dynamics. Importantly, PF-SDM avoids certain issues of previous geometric morphometrics, like Elliptical Fourier Analysis and Generalized Procrustes Analysis, such as coefficient correlations and landmark choices. We present the PF-SDM theory, provide a practically computable algorithm, and benchmark it on synthetic data.

Social media is nearly ubiquitous in modern life, and concerns have been raised about its putative societal impacts, ranging from undermining mental health and exacerbating polarization to fomenting violence and disrupting democracy. Despite extensive research, consensus on these effects remains elusive, with observational studies often highlighting concerns while randomized controlled trials (RCTs) yield conflicting or null findings. This review examines how the complexity inherent in social systems can account for such discrepancies, emphasizing that emergent societal and long-term outcomes cannot be readily inferred from individual-level effects. In complex systems, such as social networks, feedback loops, hysteresis, multi-scale dynamics, and non-linearity limit the utility of approaches for assessing causality that are otherwise robust in simpler contexts. Revisiting large-scale experiments, we explore how null or conflicting findings may reflect these complexities rather than a true absence of effects. Even in cases where the methods are appropriate, assessing the net impacts of social media provides little actionable insight given that eliminating social media is not a realistic option for whole populations. We argue that progress will require a complexity-minded approach focused on specific design choices of online platforms that triangulates experimental, observational and theoretical methods.

As the demand for efficient, low-power computing in embedded and edge devices grows, traditional computing methods are becoming less effective for handling complex tasks. Stochastic computing (SC) offers a promising alternative by approximating complex arithmetic operations, such as addition and multiplication, using simple bitwise operations, like majority or AND, on random bit-streams. While SC operations are inherently fault-tolerant, their accuracy largely depends on the length and quality of the stochastic bit-streams (SBS). These bit-streams are typically generated by CMOS-based stochastic bit-stream generators that consume over 80% of the SC system's power and area. Current SC solutions focus on optimizing the logic gates but often neglect the high cost of moving the bit-streams between memory and processor. This work leverages the physics of emerging ReRAM devices to implement the entire SC flow in place: (1) generating low-cost true random numbers and SBSs, (2) conducting SC operations, and (3) converting SBSs back to binary. Considering the low reliability of ReRAM cells, we demonstrate how SC's robustness to errors copes with ReRAM's variability. Our evaluation shows significant improvements in throughput (1.39x, 2.16x) and energy consumption (1.15x, 2.8x) over state-of-the-art (CMOS- and ReRAM-based) solutions, respectively, with an average image quality drop of 5% across multiple SBS lengths and image processing tasks.

This contribution describes efforts to model the behavior of individual pedestrians and their interactions in crowds, which generate certain kinds of self-organized patterns of motion. Moreover, this article focusses on the dynamics of crowds in panic or evacuation situations, methods to optimize building designs for egress, and factors potentially causing the breakdown of orderly motion.

Highly spin selective transport of electrons through a helically shaped electrostatic potential is demonstrated in the frame of a minimal model approach. The effect is significant even in the case of weak spin-orbit coupling. Two main factors determine the selectivity, an unconventional Rashba- like spin-orbit interaction, reflecting the helical symmetry of the system, and a weakly dispersive electronic band of the helical system. The weak electronic coupling, associated with the small dispersion, leads to a low mobility of the charges in the system and allows even weak spin-orbit interactions to be effective. The results are expected to be generic for chiral molecular systems displaying low spin-orbit coupling and low conductivity.

The interplay between superconductivity and charge density wave has often been studied from an equilibrium point of view. For example, using static tuning knobs such as doping, magnetic field and pressure, superconductivity can be enhanced or suppressed. The resulting effect on the co-existing charge density wave order, if any, is judged by variations in its ground state properties such as the ordering temperature or the spatial correlation. Such an approach can be understood as coordinated static displacements of two coupled order parameters within a Ginzburg-Landau description, evincing their interplay as either co-operative or competing but does not provide further microscopic information about the interaction. In order to assess such information, we dynamically perturb both orders from equilibrium and observe their coupling directly in the time-domain. We show that high-field multicycle terahertz pulses drive both the Higgs amplitude fluctuations of the superconducting order as well as collective fluctuations of the charge order in an electron-doped cuprate, resulting in characteristic third harmonic generation. A notable time delay is manifested between their respective driven dynamics. We propose that this may signify the important energy scale describing their coupling or imply a terahertz field-depinned charge density wave that destroys macroscopic superconductivity. Our work demonstrates a holistic approach for investigating coupled superconducting and charge density wave orders, which may shed novel light on their intertwined presence and widespread fluctuations in many classes of unconventional superconductors.

Markov decision processes (MDP) are a well-established model for sequential decision-making in the presence of probabilities. In robust MDP (RMDP), every action is associated with an uncertainty set of probability distributions, modelling that transition probabilities are not known precisely. Based on the known theoretical connection to stochastic games, we provide a framework for solving RMDPs that is generic, reliable, and efficient. It is *generic* both with respect to the model, allowing for a wide range of uncertainty sets, including but not limited to intervals, $L^1$- or $L^2$-balls, and polytopes; and with respect to the objective, including long-run average reward, undiscounted total reward, and stochastic shortest path. It is *reliable*, as our approach not only converges in the limit, but provides precision guarantees at any time during the computation. It is *efficient* because -- in contrast to state-of-the-art approaches -- it avoids explicitly constructing the underlying stochastic game. Consequently, our prototype implementation outperforms existing tools by several orders of magnitude and can solve RMDPs with a million states in under a minute.

As robotics advances toward integrating soft structures, anthropomorphic shapes, and complex tasks, soft and highly stretchable mechanotransducers are becoming essential. To reliably measure tactile and proprioceptive data while ensuring shape conformability, stretchability, and adaptability, researchers have explored diverse transduction principles alongside scalable and versatile manufacturing techniques. Nonetheless, many current methods for stretchable sensors are designed to produce a single sensor configuration, thereby limiting design flexibility. Here, we present an accessible, flexible, printing-based fabrication approach for customizable, stretchable sensors. Our method employs a custom-built printhead integrated with a commercial 3D printer to enable direct ink writing (DIW) of conductive ink onto cured silicone substrates. A layer-wise fabrication process, facilitated by stackable trays, allows for the deposition of multiple liquid conductive ink layers within a silicone matrix. To demonstrate the method's capacity for high design flexibility, we fabricate and evaluate both capacitive and resistive strain sensor morphologies. Experimental characterization showed that the capacitive strain sensor possesses high linearity (R^2 = 0.99), high sensitivity near the 1.0 theoretical limit (GF = 0.95), minimal hysteresis (DH = 1.36%), and large stretchability (550%), comparable to state-of-the-art stretchable strain sensors reported in the literature.

The importance of Variational Autoencoders reaches far beyond standalone generative models -- the approach is also used for learning latent representations and can be generalized to semi-supervised learning. This requires a thorough analysis of their commonly known shortcomings: posterior collapse and approximation errors. This paper analyzes VAE approximation errors caused by the combination of the ELBO objective and encoder models from conditional exponential families, including, but not limited to, commonly used conditionally independent discrete and continuous models. We characterize subclasses of generative models consistent with these encoder families. We show that the ELBO optimizer is pulled away from the likelihood optimizer towards the consistent subset and study this effect experimentally. Importantly, this subset can not be enlarged, and the respective error cannot be decreased, by considering deeper encoder/decoder networks.

We propose a new CNN-CRF end-to-end learning framework, which is based on joint stochastic optimization with respect to both Convolutional Neural Network (CNN) and Conditional Random Field (CRF) parameters. While stochastic gradient descent is a standard technique for CNN training, it was not used for joint models so far. We show that our learning method is (i) general, i.e. it applies to arbitrary CNN and CRF architectures and potential functions; (ii) scalable, i.e. it has a low memory footprint and straightforwardly parallelizes on GPUs; (iii) easy in implementation. Additionally, the unified CNN-CRF optimization approach simplifies a potential hardware implementation. We empirically evaluate our method on the task of semantic labeling of body parts in depth images and show that it compares favorably to competing techniques.

The discrete rotational symmetry of crystals leads to the conservation of quantized angular momentum in solids. While the exchange of energy and linear momentum between lattice vibrations (phonons) via anharmonic coupling is a cornerstone of solid-state physics, conservation and transfer of angular momentum within the lattice remains a postulate. Recently, phonon angular momentum, often in the form of chiral phonons, has been linked to giant magnetic fields, thermal Hall conductivity, dynamical multiferroicity, ultrafast demagnetization, or magnetic switching. However, the inherent process of phonon to phonon angular momentum transfer, fundamentally required to reach any magnetization equilibrium and imperative for all spin relaxation phenomena in solids, remains elusive. Here, we demonstrate the coherent transfer of angular momentum from one lattice mode to another by establishing helical nonlinear phononics. We directly observe rotational phonon-phonon Umklapp scattering dictated by pseudo angular momentum conservation and the threefold rotational symmetry of the topological insulator bismuth selenide. We identify nonlinear phonon-phonon coupling as an angular momentum transfer channel, confirmed by our ab-initio calculations. Besides bearing universal implications for angular momentum dissipation abundant in nature, our work actively reverses the natural flow, leading to the nonlinear upconversion of lattice angular momentum. We thus open the field of helical and chiral nonlinear phononics, representing a selective handle for ultrafast control of spins, topology and chiral quasi-particles.

Driver assistance systems support drivers in operating vehicles in a safe, comfortable and efficient way, and thus may induce changes in traffic flow characteristics. This paper puts forward a receding horizon control framework to model driver assistance and cooperative systems. The accelerations of automated vehicles are controlled to optimise a cost function, assuming other vehicles driving at stationary conditions over a prediction horizon. The flexibility of the framework is demonstrated with controller design of Adaptive Cruise Control (ACC) and Cooperative ACC (C-ACC) systems. The proposed ACC and C-ACC model characteristics are investigated analytically, with focus on equilibrium solutions and stability properties. The proposed ACC model produces plausible human car-following behaviour and is unconditionally locally stable. By careful tuning of parameters, the ACC model generates similar stability characteristics as human driver models. The proposed C-ACC model results in convective downstream and absolute string instability, but not convective upstream string instability observed in human-driven traffic and in the ACC model. The control framework and analytical results provide insights into the influences of ACC and C-ACC systems on traffic flow operations.

This position paper addresses the fallacies associated with the improper use of affordances in the opportunistic design of augmented reality (AR) applications. While opportunistic design leverages existing physical affordances for content placement and for creating tangible feedback in AR environments, their misuse can lead to confusion, errors, and poor user experiences. The paper emphasizes the importance of perceptible affordances and properly mapping virtual controls to appropriate physical features in AR applications by critically reflecting on four fallacies of facilitating affordances, namely, the subjectiveness of affordances, affordance imposition and reappropriation, properties and dynamicity of environments, and mimicking the real world. By highlighting these potential pitfalls and proposing a possible path forward, we aim to raise awareness and encourage more deliberate and thoughtful use of affordances in the design of AR applications.