KITopen
Not a member yet
63711 research outputs found
Sort by
Exploring room-temperature microplasticity in a B2-strengthened refractory compositionally complex alloy
Get PDFMechanoelectric sensitivity reveals destructive quantum interference in single-molecule junctions
Get PDFQuantum interference plays an important role in charge transport through single-molecule junctions, even at room temperature. Of special interest is the measurement of the destructive quantum interference dip itself. Such measurements are especially demanding when performed in a continuous mode of operation. Here, we use mechanical modulation experiments at ambient conditions to reconstruct the destructive quantum interference dip of conductance versus displacement. Simultaneous measurements of the Seebeck coefficient show a sinusoidal response across the dip without sign change. Calculations that include electrode distance and energy alignment variations explain both observations quantitatively, emphasizing the crucial role of thermal fluctuations for measurements under ambient conditions. Our results open the way for establishing a closer link between break-junction experiments and theory in explaining single-molecule transport phenomena, especially when describing sharp features in the transmission
Phase-field based shape optimization of uni- and multiaxially loaded nature-inspired porous structures while maintaining characteristic properties
Get PDFTriply periodic minimal surfaces (TPMS) are highly versatile porous formations that can be defined by formulas. Computationally based, load-specific shape optimization enables tailoring these structures for their respective application areas and thereby enhance their potential. In this investigation, individual sheet-based gyroid structures with varying porosities are specifically optimized with respect to their stiffness. A modified phase-field method is employed to establish a simulation framework for the shape optimization process. Despite constant volume and the preservation of the periodicity of the unit cells, volume redistribution occurs through displacement of the interfaces. The phase-field-based optimization process is detailed using unidirectional loading on three gyroidal unit cells with porosities of 75 %, 80 %, and 85 %. Subsequently, the gyroidal unit cell with a porosity of 85 % is shape-optimized under multidirectional loading. A subsequent experimental validation of the unidirectionally loaded cells confirms that the shape-optimized structures exhibit, on average, higher stiffness than the non-optimized structures. The highest increase of 40 % in effective modulus is achieved with the gyroid structure having a porosity of 75 %, while maintaining minimal alteration to the surface-to-volume ratio and preserving periodicity. Additionally, the experimental data show that the optimization process resulted in a shift in the linear elasticity and plasticity range. In summary, the phase-field method proves to be a valid optimization technique for complex porous structures, allowing the preservation of characteristic properties
Calibrating Bayesian generative machine learning for Bayesiamplification
Get PDFRecently, combinations of generative and Bayesian deep learning have been introduced in particle physics for both fast detector simulation and inference tasks. These neural networks aim to quantify the uncertainty on the generated distribution originating from limited training statistics. The interpretation of a distribution-wide uncertainty however remains ill-defined. We show a clear scheme for quantifying the calibration of Bayesian generative machine learning models. For a Continuous Normalizing Flow applied to a low-dimensional toy example, we evaluate the calibration of Bayesian uncertainties from either a mean-field Gaussian weight posterior, or Monte Carlo sampling network weights, to gauge their behaviour on unsteady distribution edges. Well calibrated uncertainties can then be used to roughly estimate the number of uncorrelated truth samples that are equivalent to the generated sample and clearly indicate data amplification for smooth features of the distribution
PAL - Parallel active learning for machine-learned potentials
Get PDFConstructing datasets representative of the target domain is essential for training effective machine learning models. Active learning (AL) is a promising method that iteratively extends training data to enhance model performance while minimizing data acquisition costs. However, current AL workflows often require human intervention and lack parallelism, leading to inefficiencies and underutilization of modern computational resources. In this work, we introduce PAL, an automated, modular, and parallel active learning library that integrates AL tasks and manages their execution and communication on shared- and distributed-memory systems using the Message Passing Interface (MPI). PAL provides users with the flexibility to design and customize all components of their active learning scenarios, including machine learning models with uncertainty estimation, oracles for ground truth labeling, and strategies for exploring the target space. We demonstrate that PAL significantly reduces computational overhead and improves scalability, achieving substantial speed-ups through asynchronous parallelization on CPU and GPU hardware. Applications of PAL to several real-world scenarios - including ground-state reactions in biomolecular systems, excited-state dynamics of molecules, simulations of inorganic clusters, and thermo-fluid dynamics - illustrate its effectiveness in accelerating the development of machine learning models. Our results show that PAL enables efficient utilization of high-performance computing resources in active learning workflows, fostering advancements in scientific research and engineering applications
EDITORIAL: Chemical Compound Space Exploration by Multiscale High-Throughput Screening and Machine Learning
Get PDF
INSIDE seismic monitoring approaches: Capabilities versus costs
Get PDFThis report evaluates the capabilities of different seismic monitoring approaches applied during the INSIDE project. These approaches cover traditional surface and borehole seismometers, a mini-array of geophones, and a Distributed Acoustic Sensing (DAS) system. This analysis uses as a prerequisite the catalogue of seismic events described in a separate report.
The monitoring concept developed in the frame of the project is briefly summarized in Section 2. Then, we present the monitoring results obtained from “unconventional” monitoring stations, i.e. a DAS station and a mini-array. We focus on comparing the ability of different technologies to detect seismic events (Section 3), characterize their source properties (Section 4). In addition, we evaluate the use of ambient noise for site characterization (Section 5). Finally, a cost comparison is provided, considering both installation and maintenance expenses for various monitoring solutions.
In the Section 7 of the report, the results are summarized, and recommendations are provided based on experience gained in INSIDE
The Large-Scale Helmholtz Research Infrastructure GeoLaB
Get PDFAgainst the background of climate change and the geopolitical situation, the worldwide pressure mounts to reduce the dependence on fossil fuels and to accelerate the energy transition as quickly as possible. Geothermal technologies have a key role to play in supplying and storing heat. The greatest, yet untapped geothermal potential lies in the crystalline basement with important hotspots in tectonically stressed areas. New targeted, science-based strategies are the key to harness this energy under safe, sustainable, predictable, and efficient conditions. The planned GeoLaB (Geothermal Laboratory in the Crystalline Basement) will address the fundamental challenges of reservoir technology and wellbore safety for deep geothermal projects. Experiments will contribute significantly to understanding the coupled, nonlinear processes associated with high flow rates in crystalline reservoir rocks. The application and development of cutting- edge monitoring, online analysis and visualization tools will provide fundamental knowledge essential for the safe and environmentally sound operation of geothermal energy. As an interdisciplinary and international research platform, GeoLaB will collaborate with universities, industry partners, and professional associations to foster synergies. The transparent nature of the envisaged data acquisition will allow the geoscience community to participate in GeoLaB