Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “dataset use case study”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 55 records · Page 3
All sourcesNASA NTRSDOE OSTI
NASA NTRS2012

Drought Monitoring for 3 North American Case Studies Based on the North American Land Data Assimilation System (NLDAS)

Both NLDAS Phase 1 (1996-2007) and Phase 2 (1979-present) datasets have been evaluated against in situ observational datasets, and NLDAS forcings and outputs are used by a wide variety of users. Drought indices and drought monitoring from NLDAS were recently examined by Mo et al. (2010) and Sheffield et al. (2010). In this poster, we will present results analyzing NLDAS Phase 2 forcings and outputs for 3 North American Case studies being analyzed as part of the NOAA MAPP Drought Task Force: (1) Western US drought (1998- 2004); (2) plains/southeast US drought (2006-2007); and (3) Current Texas-Mexico drought (2011-). We will examine percentiles of soil moisture consistent with the NLDAS drought monitor.

Peters-Lidard, Christa D.↗
DOE OSTI2023

Reinforcement learning for real-time process control in high-temperature superconductor manufacturing

With high efficiency and low energy loss, high-temperature superconductors (HTS) have demonstrated their profound applications in various fields, such as medical imaging, transportation, accelerators, microwave devices, and power systems. The high-field applications of HTS tapes have raised the demand for producing cost-effective tapes with long lengths in superconductor manufacturing. However, achieving the uniform and enhanced performance of a long HTS tape is challenging due to the unstable growth conditions in the manufacturing process. Although it is confirmed that the process parameters during the advanced metal organic chemical vapor deposition (A-MOCVD) process influence the uniformity of the produced HTS tapes, the high-dimensional process parameter signals and their complicated interactions make it difficult to develop an effective control policy. In this paper, we propose a local measure for the uniformity of HTS tapes to provide instant feedback for our control policy. Then, we model the manufacturing of HTS tapes as a Markov decision process (MDP) with continuous state and action spaces to assess the instant reward in real time in our feedback control model. As our MDP involves continuous and high-dimensional state and action spaces, a neural fitted Q-iteration (NFQ) algorithm is adopted to solve the MDP with artificial neural network (ANN) function approximation. The collinearity of process parameters can restrict our capability of adjusting the process parameters, which is addressed by the principal component analysis (PCA) in our method. The control policy adjusts the PCA of process parameters using the NFQ algorithm. In conclusion, based on our case studies on real A-MOCVD dataset, the obtained control policy increases the average uniformity of tapes by 5.6% and performs especially well on sample HTS tapes with a low uniformity.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2025

A Neural Optimizer With Decision-Focused Learning for Optimal Energy Storage Operation

Here, this article introduces a neural optimizer-based framework for optimizing battery energy storage system (BESS) control for grid services, including demand charge and energy cost reduction. By leveraging decision-focused learning (DFL), the proposed framework ensures seamless integration and adaptation, significantly enhancing control performance. A patch time-series transformer is employed for peak load forecasting, incorporating aleatoric uncertainty quantification to account for forecasting uncertainties within the decision-making process. The framework utilizes a solver-in-the-loop approach to generate optimal BESS actions, which are then used to train the neural optimizer-based agent. By co-optimizing both BESS operational modes and output power within the NN, the system achieves improved performance and robustness. After initial training, the forecasting and control models are jointly fine-tuned to account for forecasting errors, further improving decision precision and efficiency through DFL. Case studies are performed to validate the performance of the framework using multiple real-world datasets, demonstrating superior performance in monthly peak load forecasting compared to state-of-the-art models. In addition, the results are compared against existing decision-making approaches. The results demonstrate a reduction in monthly peak forecasting error by approximately 15% across various performance measures and achieve an optimization gap for BESS operation that is about three times smaller compared to existing methods.

Kim, Hyeonjin [Pacific Northwest National Laborato↗
NASA NTRS2017

An Integrated Examination of AMSR2 Products over Ocean

Integrated processing and analysis are presented for oceanic retrievals of precipitation, sea ice concentration, columnar water vapor and cloud water, sea surface temperature, and near surface wind from the advanced microwave scanning radiometer-2 (AMSR2) sensor. By developing a common algorithmic framework and permitting iterative interaction between historically separate science algorithms, ambiguous and contradictory retrieval results are minimized. The integration also serves to improve each algorithm individually, decreasing reliance on ancillary datasets. Case studies are presented that exemplify both ongoing challenges for these retrievals and potential research uses of such an integrated satellite product. Two cases are presented each for precipitation at high latitudes and marginal sea ice detection, with analysis supplemented by satellite data from CloudSat profiles and Himawari-8 imagery. Detection and retrieval of snowfall remains a challenge, while light rainfall detection can be aided by a variational algorithm. Sea ice concentrations below 20% cause disagreements, but the algorithms otherwise agree well on the sea ice edge. Potential mitigation strategies for ambiguous areas of light rainfall and marginal sea ice are discussed. The analysis demonstrates potential avenues for future algorithm development, but also some physical limitations of remote sensing with the AMSR2 frequencies.

variational methods↗
NASA NTRS2011

Assessment of NASA's Physiographic and Meteorological Datasets as Input to HSPF and SWAT Hydrological Models

This paper documents the use of simulated Moderate Resolution Imaging Spectroradiometer land use/land cover (MODIS-LULC), NASA-LIS generated precipitation and evapo-transpiration (ET), and Shuttle Radar Topography Mission (SRTM) datasets (in conjunction with standard land use, topographical and meteorological datasets) as input to hydrological models routinely used by the watershed hydrology modeling community. The study is focused in coastal watersheds in the Mississippi Gulf Coast although one of the test cases focuses in an inland watershed located in northeastern State of Mississippi, USA. The decision support tools (DSTs) into which the NASA datasets were assimilated were the Soil Water & Assessment Tool (SWAT) and the Hydrological Simulation Program FORTRAN (HSPF). These DSTs are endorsed by several US government agencies (EPA, FEMA, USGS) for water resources management strategies. These models use physiographic and meteorological data extensively. Precipitation gages and USGS gage stations in the region were used to calibrate several HSPF and SWAT model applications. Land use and topographical datasets were swapped to assess model output sensitivities. NASA-LIS meteorological data were introduced in the calibrated model applications for simulation of watershed hydrology for a time period in which no weather data were available (1997-2006). The performance of the NASA datasets in the context of hydrological modeling was assessed through comparison of measured and model-simulated hydrographs. Overall, NASA datasets were as useful as standard land use, topographical , and meteorological datasets. Moreover, NASA datasets were used for performing analyses that the standard datasets could not made possible, e.g., introduction of land use dynamics into hydrological simulations

Alacron, Vladimir J.↗
NASA NTRS2023

A Blind Convolutional Deep Autoencoder for Spectral Unmixing of Hyperspectral Images Over Waterbodies

Harmful algal blooms have dangerous repercussions for biodiversity, the ecosystem, and public health. Automatic identification based on remote sensing hyperspectral image analysis provides a valuable mechanism for extracting the spectral signatures of harmful algal blooms and their respective percentage in a region of interest. This paper proposes a new model called a non-symmetrical autoencoder for spectral unmixing to perform endmember extraction and fractional abundance estimation. The model is assessed in benchmark datasets, such as Jasper Ridge and Samson. Additionally, a case study of the HSI2 image acquired by NASA over Lake Erie in 2017 is conducted for extracting optical water types. The results using the proposed model for the benchmark datasets improve unmixing performance, as indicated by the spectral angle distance compared to five baseline algorithms. Improved results were obtained for various metrics. In the Samson dataset, the proposed model outperformed other methods for water (0.060) and soil (0.025) endmember extraction. Moreover, the proposed method exhibited superior performance in terms of mean spectral angle distance compared to the other five baseline algorithms. The non-symmetrical autoencoder for the spectral unmixing approach achieved better results for abundance map estimation, with a root mean square error of 0.091 for water and 0.187 for soil, compared to the ground truth. For the Jasper Ridge dataset, the non-symmetrical autoencoder for the spectral unmixing model excelled in the tree (0.039) and road (0.068) endmember extraction and also demonstrated improved results for water abundance maps (0.1121). The proposed model can identify the presence of chlorophyll-a in waterbodies. Chlorophyll-a is an essential indicator of the presence of the different concentrations of macrophytes and cyanobacteria. The non-symmetrical autoencoder for spectral unmixing achieves a value of 0.307 for the spectral angle distance metric compared to a reference ground truth spectral signature of chlorophyll-a. The source code for the proposed model, as implemented in this manuscript, can be found at https://github.com/EstefaniaAlfaro/autoencoder_owt_spectral.git.

hyperspectral imaging↗
DOE OSTI2025

Developing machine learning for heterogeneous catalysis with experimental and computational data

Machine learning techniques have emerged as a useful tool for identifying complex patterns and correlations in large datasets, such as associating catalyst performance to its physicochemical properties. In the heterogeneous catalysis communities, machine learning models have mostly been developed using high-throughput quantum chemistry calculations, with only a few case studies resulting in experimentally validated catalyst improvements. This limited success may be due to the use of simplified catalyst structures in computational studies and the lack of comprehensive experimental datasets. In this Review, we bring together studies integrating high-throughput approaches and machine learning for the advancement of solid heterogeneous catalysis, leveraging both experimental and computational data. We systematically analyze trends in the field, based on the descriptors used as model input and output; the materials, devices, or reactions investigated; the dataset size; and the overall achievements. Furthermore, for models reporting unitless R 2 values, we compare the performances based on these mentioned trends.

Computational chemistry↗
NASA NTRS2014

An Analysis of Earth Science Data Analytics Use Cases

The increase in the number and volume, and sources, of globally available Earth science data measurements and datasets have afforded Earth scientists and applications researchers unprecedented opportunities to study our Earth in ever more sophisticated ways. In fact, the NASA Earth Observing System Data Information System (EOSDIS) archives have doubled from 2007 to 2014, to 9.1 PB (Ramapriyan, 2009; and https:earthdata.nasa.govaboutsystem-- performance). In addition, other US agency, international programs, field experiments, ground stations, and citizen scientists provide a plethora of additional sources for studying Earth. Co--analyzing huge amounts of heterogeneous data to glean out unobvious information is a daunting task. Earth science data analytics (ESDA) is the process of examining large amounts of data of a variety of types to uncover hidden patterns, unknown correlations and other useful information. It can include Data Preparation, Data Reduction, and Data Analysis. Through work associated with the Earth Science Information Partners (ESIP) Federation, a collection of Earth science data analytics use cases have been collected and analyzed for the purpose of extracting the types of Earth science data analytics employed, and requirements for data analytics tools and techniques yet to be implemented, based on use case needs. ESIP generated use case template, ESDA use cases, use case types, and preliminary use case analysis (this is a work in progress) will be presented.

data analytics↗
DOE OSTI2020

Scalable Data-Intensive Geocomputation: A Design for Real-Time Continental Flood Inundation Mapping

The convergence of data-intensive and extreme-scale computing enables an integrated software and data ecosystem for scientific discovery. Developments in this realm will fuel transformative research in data-driven interdisciplinary domains. Geocomputation provides computing paradigms in Geographic Information Systems (GIS) for interactive computing of geographic data, processes, models, and maps. Because GIS is data-driven, the computational scalability of a geocomputation workflow is directly related to the scale of the GIS data layers, their resolution and extent, as well as the velocity of the geo-located data streams to be processed. Unique in high user interactivity and low end-to-end latency requirements, geocomputation applications will dramatically benefit from the convergence of high-end data analytics (HDA) and high-performance computing (HPC). The application level challenge, however, is to identify and eliminate computational bottlenecks that arise along a geocomputation workflow. Indeed, poor scalability at any of the workflow components is detrimental to the entire end-to-end pipeline. Here, we study a large geocomputation use case in flood inundation mapping that handles multiple national-scale geospatial datasets and targets low end-to-end latency. We discuss benefits and challenges for harnessing both HDA and HPC for data-intensive geospatial data processing and intensive numerical modeling of geographic processes. We propose an HDA+HPC geocomputation architecture design that couples HDA (e.g., Spark)-based spatial data handling and HPC-based parallel data modeling. Key techniques for coupling HDA and HPC to bridge the two different software stacks are reviewed and discussed.

Liu, Yan↗
DOE OSTI2024

Gray-Box Modeling for Distribution Systems With Inverter-Based Resources: Integrating Physics-Based and Data-Driven Approaches

Here, in this paper, we develop a novel gray-box modeling approach for distribution systems with inverter-based resources (IBRs). The proposed gray-box modeling method aims to improve estimation accuracy by taking advantages of both physics-based (white-box) and data-driven (black-box) modeling approaches. To this end, we utilize partial physical knowledge of the system, including the inverters’ structures and control diagrams, as well as the equivalent network model simplified through Kron reduction. The white-box model containing unknown parameters is then constructed with mathematical equations and an optimization-based method is subsequently employed to identify these unknown parameters within the white-box model. Next, the graybox modeling framework is then constructed by embedding the output variables of the white-box model into the input vector of a black-box model (represented using a neural network). Finally, the black-box section is trained using the collected input-output datasets and the gray-box model is then obtained. Furthermore, case studies demonstrate that our gray-box modeling approach effectively improves estimation accuracy compared to purely physics-based or data-driven methods.

42 ENGINEERING↗
DOE OSTI2021

Integrating Cambium Marginal Costs into Electric Sector Decisions: Opportunities to Integrate Cambium Marginal Cost Data into Berkeley Lab Analysis and Technical Assistance

NREL’s Cambium tool generates forward-looking simulations of marginal wholesale electricity costs associated with NREL’s Standard Scenarios. The scenarios include growing shares of variable renewable energy (VRE, i.e. wind and solar), among other power sector assumptions, between 2018 and 2050. The tool’s primary output—hourly costs at more than 130 balancing areas—could serve as public and transparent data source that supports electric-sector decision-making processes across the U.S. Berkeley Lab conducts a large range of analyses that use historical and forward-looking wholesale electricity prices to inform electric-sector decisions. In this report, Berkeley Lab uses its expertise to evaluate the Cambium cost data. We compare Cambium data with historical wholesale prices for the year 2018 and other modeled prices for the year 2030. We then present eight case studies in which Berkeley Lab researchers use Cambium data to replicate previous analyses based on other price datasets. We describe where primary findings and underlying key price dynamics align or differ, and highlight possible novel insights from the Cambium data. Finally, we qualitatively evaluate the suitability of Cambium costs in ten additional Berkeley Lab studies, though a direct comparison with alternative price data was not feasible at this time. The goal is to inform how electric-sector decision-makers and DOE program offices may be able to use this cohesive dataset, and to highlight what improvements to Cambium may make it even more useful.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
NASA NTRSDate not supplied

Characterizing Wildfires in Western US.: A Cloud-based Case Study for Interdisciplinary Research using NASA Resources

This presentation will demonstrate a case study of interdisciplinary research done in the Amazon Web Services (AWS) cloud platform, in addition to in the local machine. We conduct data analysis next to data by leveraging various cloud-based data in NASA Earthdata Cloud, which are distributed by different missions/NASA Distributed Active Archive Centers (DAACs), and cloud computing resources at NASA. For instance, we directly access multiple datasets stored in the AWS Simple Storage Service (S3) buckets using a Python Jupyter notebook through a JupyterHub interface hosted in AWS (without having to download data), and conduct data analysis next to data in the cloud. We will also show how to share the research results following Open Source policy. This case study characterizes the change in wildfire events in the western United States during the past 20 years. In particular, we focus on the wildfires in California in 2021, one of the most severe wildfire years occurring in the most recent 20 years in California. We will analyze the possible causes of wildfires, such as drought conditions and climate variability, and examine the impacts of wildfires on air quality and atmospheric composition, and on land cover. We will examine the data distributed by the NASA Goddard Earth Sciences Data and Information Services Center (GES DISC), including aerosols and meteorological data from the NASA Modern-Era Retrospective analysis for Research and Applications version 2 (MERRA-2), precipitation from the Global Precipitation Measurement (GPM) and Global Precipitation Climate Project (GPCP), and aerosol index from Ozone Monitoring Instrument (OMI). We also utilize the data distributed by the Physical Oceanography (PO) DAAC, such as Sea Surface Temperature (SST) data from the Group for High Resolution Sea Surface Temperature (GHRSST), and the data distributed by Land Processes (LP) DAAC, such as Normalized Difference Vegetation Index (NDVI).

Xiaohua Pan↗
DOE OSTI2022

Deep Learning Based Frequency Stability Assessment in Power Grid with High Renewables

Frequency stability assessment is one critical aspect of power system security assessment. Traditional N-1 screening method is based on the simulations of a few typical daily and seasonal operation scenarios. However, the increasing integration of inverter-based renewables and the retirement of conventional synchronous generators result in decreasing system inertia and growing complexity of system operating conditions. Selecting a few typical operation scenarios cannot cover all operating conditions, and the time-domain simulation of all operation conditions requires tremendous time. This paper proposes a more efficient frequency stability assessment method based on deep learning. The affinity propagation clustering algorithm is used to divide the dataset into different clusters, so the selected dataset for training can cover the diversified operating conditions as much as possible. Also, feature normalization is applied to both the training dataset and testing dataset in order to remove any unnecessary bias. Especially, trained model based on full dataset normalization has bounded error in the prediction. The case study on the reduced 240-bus WECC system demonstrates that the proposed method can predict accurate frequency nadir with limited training dataset. The deep learning model using the revised feature normalization can predict more accurate frequency nadir than that using the traditional feature normalization and has very small maximum prediction error.

affinity propagation↗
NASA NTRS2005

Impacts and environmental catastrophes: A study of the effects of impact events on the climate system

The goal of this work is to investigate the perturbation of the climate system due to large impact events. Impacts are among the most important mechanisms for the evolution, distribution, and destruction of life in the universe. However, the possible climatic effects of an impact were not seriously considered until 1980, when Louis and Walter Alvarez suggested that the profound end-Cretaceous extinction might have been caused by the impact of an asteroid or comet about 10 km in diameter. Since then, the climatic change associated with the end-Cretaceous impact has become one of the most interesting and still unresolved questions in linking the well-known Chicxulub impact event and the end- Cretaceous mass extinction. While the end-Cretaceous impact offers the best-documented case of an impact affecting the Earth's climate and biota, even smaller (and more frequent in time) impacts could introduce significant perturbations of the climate comparable, if not larger, to the largest known volcanic perturbations. We propose to study the mechanical and thermal state of the atmosphere following an impact event. This will be done by using both one-dimensional and three-dimensional climate models. When necessary, modifications of the state-of-the-art general circulation models will b e carried out. We want to use the end-Cretaceous impact event as a case study. This allows us to take advantage of the extensive modeling of this impact event that has already been carried out through a previous Exobiology grant. Furthermore, a large experimental dataset, that can be used to constrain and test our models, is associated with the end-Cretaceous mass extinction (one of the largest of the Phanerozoic) and impact event.

Pierazzo, E.↗
NASA NTRS2010

Advances in Landslide Hazard Forecasting: Evaluation of Global and Regional Modeling Approach

A prototype global satellite-based landslide hazard algorithm has been developed to identify areas that exhibit a high potential for landslide activity by combining a calculation of landslide susceptibility with satellite-derived rainfall estimates. A recent evaluation of this algorithm framework found that while this tool represents an important first step in larger-scale landslide forecasting efforts, it requires several modifications before it can be fully realized as an operational tool. The evaluation finds that the landslide forecasting may be more feasible at a regional scale. This study draws upon a prior work's recommendations to develop a new approach for considering landslide susceptibility and forecasting at the regional scale. This case study uses a database of landslides triggered by Hurricane Mitch in 1998 over four countries in Central America: Guatemala, Honduras, EI Salvador and Nicaragua. A regional susceptibility map is calculated from satellite and surface datasets using a statistical methodology. The susceptibility map is tested with a regional rainfall intensity-duration triggering relationship and results are compared to global algorithm framework for the Hurricane Mitch event. The statistical results suggest that this regional investigation provides one plausible way to approach some of the data and resolution issues identified in the global assessment, providing more realistic landslide forecasts for this case study. Evaluation of landslide hazards for this extreme event helps to identify several potential improvements of the algorithm framework, but also highlights several remaining challenges for the algorithm assessment, transferability and performance accuracy. Evaluation challenges include representation errors from comparing susceptibility maps of different spatial resolutions, biases in event-based landslide inventory data, and limited nonlandslide event data for more comprehensive evaluation. Additional factors that may improve algorithm performance accuracy include incorporating additional triggering factors such as tectonic activity, anthropogenic impacts and soil moisture into the algorithm calculation. Despite these limitations, the methodology presented in this regional evaluation is both straightforward to calculate and easy to interpret, making results transferable between regions and allowing findings to be placed within an inter-comparison framework. The regional algorithm scenario represents an important step in advancing regional and global-scale landslide hazard assessment and forecasting.

Kirschbaum, Dalia B.↗
DOE OSTI2021

Mitigating Catastrophic Forgetting in Deep Learning in a Streaming Setting Using Historical Summary

Recent advancements in scientific equipment and the adaptation of electronics and the Internet of Things (IoT) in our everyday lives resulted in large and complex data production at a high rate. Making meaningful and timely knowledge discovery at a modest cost from this big data is difficult for computing power and storage limitations. Training deep learning models incrementally in a streaming setting can help us with overcoming these limitations. However, in a well-known phenomenon named catastrophic forgetting, incrementally trained models increasingly perform poorly on the past data. To mitigate catastrophic forgetting in training in a streaming setting, we propose constructing a historical summary over time and use the summary with newly arrived data during incremental training. We propose various data summarization techniques such as random sampling, micro clustering, coreset computation, and Auto Encoders to counteract catastrophic forgetting. We built a pipeline for incremental training with a historical summary for training deep learning models for streaming data. We demonstrate the effectiveness of historical summary in mitigating catastrophic forgetting using three case studies involving three different deep learning applications: an Artificial Neural Network (ANN) for classification task on MNIST dataset, a language model (RNN-LM) on the WikiText2 dataset, and a Convolutional Neural Network (CNN), ResNet50 to classify the ImageNet dataset. Through the training of the models, we observe that catastrophic forgetting is evident in ANN and CNN but not in an RNN. For the first task, our method recovers up to 47.9% lost accuracy due to catastrophic forgetting. For the third task, the historical summary recovers classification accuracy by up to 25%. For the second task, though there is not proof of catastrophic forgetting, the training performance (PPL) improves by up to 26% with historical summary.

Dash, Sajal↗
DOE OSTI2022

Optimal Energy Storage System and Smart Switch Placement in Dynamic Microgrids With Applications to Marine Energy Integration

Here this paper studies a dynamic microgrid (DMG) planning problem that places energy storage systems (ESSs) and smart switches (SSWs) optimally in the system. We apply the proposed methodology to applications concerning marine renewable energy (MRE). MRE is an emerging clean energy resource with enormous capacity but volatile and intermittent energy output profiles. Innovative grid-integration technologies designed to enhance the reliability of an MRE-integrated system are needed. However, there are still limited studies in this regard. Existing works have shown the promising prospect of using a dynamic microgrid (DMG) operational concept to accommodate renewable resources in distribution systems, but they usually assume fixed ESS and SSW installations. To further improve the operational flexibility of DMGs, we propose a DMG planning methodology that optimally places ESSs and SSWs so that a DMG with MRE is warranted with proper resource adequacy and topological flexibility in both the contingency and normal operations. We use realistic case studies based on a real-world distribution network and the U.S. Department of Energy's MRE dataset to verify the value and validity of the proposed work.

25 ENERGY STORAGE↗
DOE OSTI2025

Development and transferability of neural-network models for plasma-surface interactions

Plasma-surface interactions are increasingly critical to modern technologies; yet, accurate molecular dynamics simulations remain limited by the capabilities of interatomic potentials. Deep Potentials (DPs) promise to revolutionize the field by providing a systematic method for producing accurate interatomic potentials. The primary challenge of DP development is selecting a dataset, which efficiently spans the set of atomic environments one expects to encounter in the subsequent molecular dynamics simulations. The computational cost of density functional theory calculations, which are the typical basis for DP development, makes it impossible to directly verify the quality of a given DP. To address this challenge, we explore the development of a deep-learned interatomic potential, “DeepREBO,” trained to reproduce the behavior of the REBO2 empirical potential, enabling direct validation of training methodology and transferability. Using an active learning framework, we begin with a minimal dataset and iteratively expand it to train a Deep Potential-Smooth Edition model that faithfully reproduces REBO2 results for 25 eV hydrogen bombardment of diamond (001), a particularly challenging case. We show that small, carefully curated datasets can outperform large, unguided ones, with effective models requiring fewer than 15 000 snapshots. Subsequent transferability tests demonstrate that while DeepREBO generalizes well to diamond (111) surfaces, performance degrades for amorphous carbon or higher-energy impacts, highlighting the need for use-case-specific training data. We also evaluate methods to improve short-range repulsion. This study outlines best practices for training robust deep potentials and underscores the importance of dataset design for predictive plasma simulations.

Ab-initio molecular dynamics↗