Tree tensor networks (TTNs) are widely used in low-rank approximation and quantum many-body simulation. In this work, we present a formal analysis of the quotient geometry underlying the TTN parameter space. Our framework allows for arbitrary horizontal distributions, and we develop efficient first- and second-order optimization algorithms that exploit this geometry. Additionally, we devise a backpropagation algorithm for training TTNs in a kernel learning setting. We validate our methods through numerical experiments on a representative digit classification task and reveal an important tradeoff between two different horizontal distributions that are available for TTNs: while one offers cleaner geometric statements, the other ultimately leads to more efficient algorithms.
Electrocatalysis is driven by the chemical nature of electrodes and their surfaces. Furthermore, the electrocatalytic activity and selectivity critically depend on the atomic surface arrangement of electrodes, characterized by facets and their crystallographic orientation, surface steps, and defects, i.e., the electrode surface structure. Over the past decades, single-crystal studies have provided fundamental insights into so-called structure–activity relationships, while nanoparticle systems have extended this knowledge toward industrial electrocatalysts. Yet, translating these insights into practical electrodes remains limited by the use of binders, supports, and surfactants, which mask active sites, compromise stability, and hinder reproducibility. This review provides a framework for the electrochemical restructuring of metal electrodes, emphasizing how electrochemical processes can generate nanostructured, binder-free, free-standing electrodes with tailored surface architectures. We critically evaluate restructuring strategies, including electrodeposition, potential cycling, anodic polarization, electrochemical dealloying, and particularly cathodic corrosion, reframing the latter not only as a degradation pathway but also as a versatile tool for fine-tuning metal surfaces and fabricating flexible electrodes. Fundamental insights into cathodic corrosion are discussed in detail, linking atomic-scale restructuring mechanisms with the emergence of features relevant for catalysis and electrocatalysis. We further address stability challenges, parameters influencing restructuring, and applications in electrocatalysis, including HER and OER, CO2 reduction, nitrate reduction, and electro-oxidation of small organic molecules as candidates for fuels in energy storage and conversion. This review bridges electrochemical surface science with engineering of electrode materials to provide a roadmap for advancing surface design in energy technology, electrochemical science, and sustainable catalysis.
Hetero-agglomeration of micron- and submicron-sized particles enables the formation of all-solid-state battery cathode materials with enhanced safety and energy density. This study investigates a turbulent pipe flow process for mixing lithium iron phosphate (LiFePO4, LFP), lithium indium chloride (Li3InCl6, LIC), and carbon black (CB) to form three-component hetero-agglomerates. The analysis aims to identify suitable pipe dimensions and evaluate how different mixing configurations affect agglomerate size, composition, and structure. The investigation employs a dedicated simulation framework utilizing a mesh-free representation of turbulent flow fields derived from direct numerical simulation datasets. Lagrangian particle tracking incorporates an improved discrete random walk model, while stochastic agglomerate structure simulations account for various turbulent collision mechanisms, hydrodynamic interactions, and breakup events. Prior to simulations, virtual CB aggregates were stereologically reconstructed from scanning electron microscopy images, the particle size distributions were determined using laser light diffraction, and the surface energies of all materials were determined from contact angle measurements using the Owens-Wendt-Rabel-Kaelble method. Simulations examined parallel mixing and sequential premixing configurations (LFP-CB, LIC-CB, and LFP-LIC) across various pipe diameters. For parallel mixing under experimental process conditions, an 18 mm pipe diameter proved optimal, achieving maximum particle capture in agglomerates at a downstream distance of 2.73 m. Premixing diameters were selected based on mixing length and deposition. The mixing configuration significantly influences the agglomerate composition, material-specific mixing quality measures, and the contact-pair distribution. The findings provide concrete recommendations for experimental process design regarding pipe dimensions and mixing strategies, while the influence on electrochemical properties requires further investigation.
Artifacts in the electroencephalogram (EEG) degrade signal quality and impact the analysis of brain activity. Current methods for detecting artifacts in sleep EEG range from threshold-based algorithms to machine learning approaches, yet applications remain limited for single-channel mobile EEG. We propose a convolutional neural network (CNN) model incorporating a convolutional block attention module (CNN-CBAM) to detect and localize artifacts in sleep EEG using attention maps. We benchmarked this model against 6 other machine learning and signal processing approaches. We trained/tuned all models on 72 manually annotated EEG recordings obtained during home-based monitoring from 18 healthy participants with a mean (SD) age of 68.05 y (±5.02). We tested them on 26 separate recordings from 6 healthy participants with a mean (SD) age of 68.33 y (±4.08), which contained artifacts in 4% of epochs. CNN-CBAM achieved the highest area under the receiver operating characteristic curve (0.88), sensitivity (0.81), and specificity (0.86) among the tested approaches. Under the ideal choice of an attention threshold of 0.66, the attention maps from CNN-CBAM localized artifacts within detected artifact epochs with a sensitivity of 0.61 and specificity of 0.63. This work demonstrates the feasibility of automating artifact detection and localization in wearable sleep EEG.
A comprehensive understanding of the mechanisms underlying eating behavior is crucial for preventing and treating obesity and eating-related disorders. Gastric interoceptive sensitivity, defined as the perceptual threshold of satiation and fullness, is a key physiological marker of meal termination. Although stress is known to influence eating behavior, its impact on the relationship between gastric interoception and food intake remains insufficiently understood. This study behaviorally examined the association between gastric interoceptive sensitivity and snack consumption under high- and low-stress conditions, while accounting for individual eating traits. In 61 fasted adults, gastric interoceptive sensitivity and self-determined snack intake were assessed following both a stress and a control task across four laboratory sessions. Gastric interoceptive sensitivity was measured using a novel paradigm, the Magic Table (MT). Multilevel analyses showed that both satiation and fullness perception thresholds were significantly associated with snack consumption. Under stress, however, only the fullness threshold remained significantly related to snack intake. Higher emotional and restrained eating scores were linked to a weaker association between gastric interoceptive sensitivity and snack intake, whereas higher uncontrolled eating scores showed the opposite pattern. These findings suggest that satiation and fullness signals guide snack termination under low stress, whereas acute stress and certain eating styles weaken the coupling between gastric signals and behavioral regulation. Dysregulated eating may therefore reflect impaired integration of interoceptive input into decision-making, particularly under stress. Beyond enhancing interoceptive sensitivity, studies on interoceptive-focused interventions may benefit from strengthening the integration of satiation signals into eating decisions, especially in stressful contexts.
