Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “transfer learning.”

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 721 records · Page 40
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

Charge accumulation and solvation in $β$-NiOOH: Surface chemistry of an OER catalyst from ML-aided simulations

Electrochemical water splitting is a key technology for a sustainable energy transition, providing a route to store surplus electricity from renewable sources. A central bottleneck is the sluggish oxygen evolution reaction (OER), which drives the search for catalysts that are active, stable, and inexpensive enough for large-scale deployment. Within this context, pure and doped NiO x H y combine high activity with low cost, making them prime candidates for alkaline OER. Yet, despite extensive study, the atomistic structure of NiOOH under operando conditions and the associated reaction mechanisms remain debated. Here, we investigate the structural complexity of pure β-NiOOH, the scaffold for its doped derivatives. We systematically investigate the oxidation of the surface adsorbates via proton-coupled electron transfer steps across relevant facets and sites, identifying the most probable sequence of deprotonation events. Our results reveal asymmetric charge accumulation on Wulff-relevant surfaces and show how applied potential can promote morphological restructuring. Explicit solvation is included through machine-learning interatomic potential molecular dynamics of the NiOOH/water interface, which allows us to resolve the hydrophobic and hydrophilic character of different surfaces and the associated interfacial water structure. Together, these insights demonstrate how surface chemistry and solvation jointly govern the stability of NiOOH and the accumulation of surface charge, with possible implications for catalytic performance.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2023

Improving Knowledge Base with Network Architecture Diagrams

In order to ensure Cyber Security Team (CST) has access to updated and in-depth information regarding Fermilab systems, capabilities, procedures, tools, and training, the CST created the Knowledge Base. This consists of a wealth of information for current and future CST employees to improve knowledge retention and transfer. The Knowledge Base, however, uses outdated network diagrams that lack many of the previous and upcoming changes to the architecture. For this reason, updated diagrams have been created to reflect the current position of the CST capabilities. By employing the knowledge learned at Fermi National Accelerator Laboratory, three diagrams have been created to highlight the current state of the CST’s operations and capabilities.

Blum, Ethan↗
DOE OSTI2026

Machine learning based prediction of airflow maldistribution in air-to-refrigerant heat exchangers

Flow maldistribution is a common challenge in heat exchanger (HX) design and particularly important for air-to-refrigerant geometries where capacity losses can approach 65%. This has a major impact on central air conditioning systems, as compact duct design motivates the use of A-type HXs which are known to be affected by airflow maldistribution. Because velocity profiles are difficult to predict, components are often oversized leading to increased material cost, system footprint, and refrigerant charge. Several studies detail airflow maldistribution for individual HXs and packages, but findings cannot always be extrapolated to new designs. In this work, a machine learning (ML) based flow profile prediction framework is developed and applied to two common package configurations: (i) A-type and (ii) U-type HXs, across a broad range of HX geometries and flow rates. Porous media CFD simulations are validated against independent data for both package types as well as comprehensive in house measurements for a finless geometry with shape optimized non-round tubes, which validates the framework for new heat transfer surfaces. The ML models are trained on the porous media CFD simulations, predicting volumetric flow rate (VFR) within 1.1% and 1.9% with maximum relative L 2 norm errors of 0.48 and 0.65, respectively, while also delivering 10 5 speed up factor compared to full porous media CFD. HX level simulations show an up to 9% reduction in heat transfer from flow maldistribution, with greater losses occurring at smaller half apex angles. This framework enables rapid and highly accurate prediction of airflow maldistribution induced capacity degradation.

42 ENGINEERING↗
DOE OSTI2022

Review: Induced Seismicity During Geoenergy Development—A Hydromechanical Perspective

The basic triggering mechanism underlying induced seismicity traces back to the mid-1960s that relied on the process of pore-fluid pressure diffusion. The last decade has experienced a renaissance of induced seismicity research and data proliferation. An unprecedent opportunity is presented to us to synthesize the robust growth in knowledge. The objective of this article is to provide a concise review of the triggering mechanisms of induced earthquakes with a focus on hydro-mechanical processes. Four mechanisms are reviewed: pore-fluid pressure diffusion, poroelastic stress, Coulomb static stress transfer, and aseismic slip. For each, an introduction of the concept is presented, followed by case studies. Diving into these mechanisms sheds light on several outstanding questions. For example, why did some earthquakes occur far from fluid injection or after injection stopped? Our review converges on the following conclusions: (a) Pore-fluid pressure diffusion remains a basic mechanism for initiating inducing seismicity in the near-field. (b) Poroelastic stresses and aseismic slip play an important role in inducing seismicity in regions beyond the influence of pore-fluid pressure diffusion. (c) Coulomb static stress transfer from earlier seismicity is shown to be a viable mechanism for increasing stresses on mainshock faults. (d) Multiple mechanisms have operated concurrently or consecutively at most induced seismicity sites. (e) Carbon dioxide injection is succeeding without inducing earthquakes and much can be learned from its success. Future research opportunities exist in deepening the understanding of physical and chemical processes in the nexus of geoenergy development and fluid motion in the Earth’s crust.

58 GEOSCIENCES↗
DOE OSTI2024

Investigating resource-efficient neutron/gamma classification ML models targeting eFPGAs

There has been considerable interest and resulting progress in implementing machine learning (ML) models in hardware over the last several years from the particle and nuclear physics communities. A big driver has been the release of the Python package, hls4ml, which has enabled porting models specified and trained using Python ML libraries to register transfer level (RTL) code. So far, the primary end targets have been commercial field-programmable gate arrays (FPGAs) or synthesized custom blocks on application specific integrated circuits (ASICs). However, recent developments in open-source embedded FPGA (eFPGA) frameworks now provide an alternate, more flexible pathway for implementing ML models in hardware. These customized eFPGA fabrics can be integrated as part of an overall chip design. In general, the decision between a fully custom, eFPGA, or commercial FPGA ML implementation will depend on the details of the end-use application. In this work, we explored the parameter space for eFPGA implementations of fully-connected neural network (fcNN) and boosted decision tree (BDT) models using the task of neutron/gamma classification with a specific focus on resource efficiency. We used data collected using an AmBe sealed source incident on Stilbene, which was optically coupled to an OnSemi J-series silicon photomultiplier (SiPM) to generate training and test data for this study. We investigated relevant input features and the effects of bit-resolution and sampling rate as well as trade-offs in hyperparameters for both ML architectures while tracking total resource usage. The performance metric used to track model performance was the calculated neutron efficiency at a gamma leakage of 10 -3 . The results of the study will be used to aid the specification of an eFPGA fabric, which will be integrated as part of a test chip.

47 OTHER INSTRUMENTATION↗
DOE OSTI2020

On Performance Prediction of Big Data Transfer in High-performance Networks

Big data generated by large-scale scientific and industrial applications need to be transferred between different geographical locations for remote storage, processing, and analysis. High-speed dedicated connections provisioned in High-performance Networks (HPNs) are increasingly utilized to carry out such big data transfer. HPN management highly relies on an important capability of performance (mainly throughput) prediction to reserve sufficient bandwidth and meanwhile avoid over-provisioning that may result in unnecessary resource waste. This capability is critical to improving the resource (mainly bandwidth) utilization of dedicated connections and meeting various user requests for data transfer. Conventional methods conduct performance prediction by fitting prior observed transfer history with predefined loss functions, without considering unobservable latent factors such as competing loads on end hosts. Such latent factors also have a significant impact on the application-level data transfer performance, which may result in an inaccurate prediction model. In this paper, we first investigate the impact of latent factors and propose a clustering-based method to eliminate their negative impact on performance prediction. We then develop a robust machine learning-based performance predictor by: i) incorporating the proposed latent factor elimination method into data preprocessing, and ii) adopting a customized domain guided loss function. Extensive experimental results show that our predictor achieves significantly higher prediction accuracy than several other state-of-the-art methods.

Liu, Wuji↗
DOE OSTI2020

Systematic Analysis of Drivers and Barriers of Cross-Sectorial Learning Between Nuclear and Oil and Gas Decommissioning Projects - 20139

Project and megaproject management in the energy sector have historically focused on the planning and delivery of the construction of infrastructure. Therefore, policies are predominantly oriented to support the construction of infrastructure rather than its decommissioning. However, globally, several facilities have reached (or will soon reach) their end-of-life and need to be decommissioned. These facilities span across the energy sector, including nuclear power plants and oil and gas rigs, whose decommissioning present unprecedented techno-socio-economic challenges. Moreover, globally, their decommissioning cost is estimated to reach hundreds of billions of US dollars and this budget keeps increasing, with limited knowledge-transfer across industries to mitigate the spiraling increase of this figure. Cross-sectorial knowledge-transfer is a way to tackle this matter and improve the planning and delivery of decommissioning and waste management projects. The aim of this research is to build a road-map based on cross-sectorial industrial experience, designed to promote the sharing of good practices between projects both within and across industries dealing with major decommissioning and waste management challenges. To reach this aim, the tacit knowledge of senior practitioners on several large scale nuclear and oil and gas decommissioning projects is made explicit by the means of semi-structured interviews, systematically analyzed through directed content analysis. The results of this analysis include a list of good practices that could and should be transferred to nuclear decommissioning projects, including the need to promote the adoption of incentives explicitly linked to the project deliverables, and to develop long-term plans to support smooth transition from operations to decommissioning. Suggestions about how the knowledge-transfer can be promoted within and across organizations are also discussed. In this way, this research provides practical guidance for project managers and policymakers on how to support the planning and delivery of projects with improved performance. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2023

AI-Accelerated Design of Targeted Covalent Inhibitors for SARS-CoV-2

Direct-acting antivirals for the treatment of the COVID-19 pandemic caused by the SARS-CoV-2 virus are needed to complement vaccination efforts. Given the ongoing emergence of new variants, automated experimentation, and active learning based fast workflows for antiviral lead discovery remain critical to our ability to address the pandemic’s evolution in a timely manner. While several such pipelines have been introduced to discover candidates with noncovalent interactions with the main protease (M pro ), here we developed a closed-loop artificial intelligence pipeline to design electrophilic warhead-based covalent candidates. Here, this work introduces a deep learning-assisted automated computational workflow to introduce linkers and an electrophilic “warhead” to design covalent candidates and incorporates cutting-edge experimental techniques for validation. Using this process, promising candidates in the library were screened, and several potential hits were identified and tested experimentally using native mass spectrometry and fluorescence resonance energy transfer (FRET)-based screening assays. We identified four chloroacetamide-based covalent inhibitors of M pro with micromolar affinities (K I of 5.27 μM) using our pipeline. Experimentally resolved binding modes for each compound were determined using room-temperature X-ray crystallography, which is consistent with the predicted poses. The induced conformational changes based on molecular dynamics simulations further suggest that the dynamics may be an important factor to further improve selectivity, thereby effectively lowering KI and reducing toxicity. These results demonstrate the utility of our modular and data-driven approach for potent and selective covalent inhibitor discovery and provide a platform to apply it to other emerging targets.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2021

Particle track classification using quantum associative memory

Pattern recognition algorithms are commonly employed to simplify the challenging and necessary step of track reconstruction in sub-atomic physics experiments. Aiding in the discrimination of relevant interactions, pattern recognition seeks to accelerate track reconstruction by isolating signals of interest. In high collision rate experiments, such algorithms can be particularly crucial for determining whether to retain or discard information from a given interaction even before the data is transferred to tape. As data rates, detector resolution, noise, and inefficiencies increase, pattern recognition becomes more computationally challenging, motivating the development of higher efficiency algorithms and techniques. Quantum associative memory is an approach that seeks to exploits quantum mechanical phenomena to gain advantage in learning capacity, or the number of patterns that can be stored and accurately recalled. Here, we study quantum associative memory based on quantum annealing and apply it to the particle track classification. We focus on discrimination models based on Ising formulations of quantum associative memory model (QAMM) recall and quantum content-addressable memory (QCAM) recall. We characterize classification performance of these approaches as a function detector resolution, pattern library size, and detector inefficiencies, using the D-Wave 2000Q processor as a testbed. Discrimination criteria is set using both solution-state energy and classification labels embedded in solution states. We find that energy-based QAMM classification performs well in regimes of small pattern density and low detector inefficiency. In contrast, state-based QCAM achieves reasonably high accuracy recall for large pattern density and the greatest recall accuracy robustness to a variety of detector noise sources.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2023

OPERATOR 4.0 FOR HYBRID MANUFACTURING

Hybrid manufacturing, a combination of additive and subtractive manufacturing capabilities in one system, has recently become a more viable production option across several industries. Although current hybrid manufacturing research covers a broad range of topics, there is a lack of focus on how this new technology impacts both the designer and the operator of hybrid systems. This paper identifies areas of literature across design theory and Industry/Operator 4.0 research efforts and presents a path for applying this research to hybrid manufacturing users. The unique relationship between operator and designer is highlighted as they learn new strategies and develop new intuitive judgements over time to become the first experienced/expert users of hybrid manufacturing. The potential impact of excessive cognitive workload due to the novel combination of processes is discussed. This paper begins a critical discussion about proper knowledge transfer to other hybrid designers and operators, as well as towards efforts of monitoring, inspecting, and automating hybrid manufacturing processes.

Fillingim, Blane↗
DOE OSTI2021

Thermodynamic non-ideality and disorder heterogeneity in actinide silicate solid solutions

Non-ideal thermodynamics of solid solutions can greatly impact materials degradation behavior. We have investigated an actinide silicate solid solution system (USiO 4 –ThSiO 4 ), demonstrating that thermodynamic non-ideality follows a distinctive, atomic-scale disordering process, which is usually considered as a random distribution. Neutron total scattering implemented by pair distribution function analysis confirmed a random distribution model for U and Th in first three coordination shells; however, a machine-learning algorithm suggested heterogeneous U and Th clusters at nanoscale (~2 nm). The local disorder and nanosized heterogeneous is an example of the non-ideality of mixing that has an electronic origin. Partial covalency from the U/Th 5f–O 2p hybridization promotes electron transfer during mixing and leads to local polyhedral distortions. The electronic origin accounts for the strong non-ideality in thermodynamic parameters that extends the stability field of the actinide silicates in nature and under typical nuclear waste repository conditions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2026

Enabling accurate chemical modeling of shocked energetic materials using a machine learning interatomic potential

Understanding the complex chemistry of organic materials under dynamic compression is important for many applications, but it is challenging due to the large number of reactions occurring at various time scales. Here, in this study, we develop a machine learning potential based on Chebyshev polynomials to study the insensitive energetic material 1,3,5-triamino-2,4,6-trinitrobenzene (TATB) under detonation. We discuss a strategy for constructing diverse training data needed to capture the complex chemistry of TATB. Our potential demonstrates strong transferability across a wide range of thermodynamic conditions and other explosives, enabling accurate and reliable chemical modeling of organic materials under extreme conditions. The efficiency of our approach allows for simulations over several nanoseconds and for large system sizes, providing detailed insights into the chemistry of shocked TATB. The model accurately reproduces experimental Hugoniot equation of state data, and our simulations reveal the rapid formation of nitrogen-rich carbon clusters following shock. The methods and datasets developed here offer a robust framework for accurate chemical modeling of other shocked organic energetic materials.

Chemistry↗
DOE OSTI2022

Understanding the Impact of Data Staging for Coupled Scientific Workflows

We report the rate of data generated by cutting-edge experimental science facilities and large-scale simulations enabled by current high-performance computing (HPC) systems has continued to grow at a far greater pace than the development of the network and storage capabilities on which these systems rely. To cope with this challenge, scientist are moving toward the creation of autonomous experiments and HPC simulations using machine learning. However, efficiently moving, storing, and processing large amounts of data away from the point of origin presents an incredible challenge. In-memory computing, in situ analysis, data staging, and data streaming are recognized viable alternatives to traditional file-based methods for transferring data between coupled workflows. However, the performance trade-offs and limitations for these methods are not fully understood when used in HPC applications. This article presents a comprehensive performance assessment of the current solutions for data staging when applied to applications that are not necessary I/O intensive which makes them not ideal candidates for these methods. Our study is based on experiments running at scale on Oak Ridge National Laboratory's Summit supercomputer using applications and simulations that cover typical computational motifs and patterns. We investigated the usability and cost/benefit trade-offs of staging algorithms for HPC applications under different scenarios and highlight opportunities for optimizing the dataflow between coupled simulation workflows.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

RuralAI in Tomato Farming: Integrated Sensor System, Distributed Computing, and Hierarchical Federated Learning for Crop Health Monitoring

Precision horticulture is evolving due to scalable sensor deployment and machine learning (ML) integration. These advancements boost the operational efficiency of individual farms, balancing the benefits of analytics with autonomy requirements. However, given concerns that affect wide geographic regions (e.g., climate change), there is a need to apply models that span farms. Federated learning (FL) has emerged as a potential solution. FL enables decentralized ML across different farms without sharing private data. Traditional FL assumes simple two-tier network topologies and, thus, falls short of operating on more complex networks found in real-world agricultural scenarios. Networks vary across crops and farms and encompass various sensor data modes, extending across jurisdictions. New hierarchical FL (HFL) approaches are needed for more efficient and context-sensitive model sharing, accommodating regulations across multiple jurisdictions. Here, we present the RuralAI architecture deployment for tomato crop monitoring, featuring sensor field units for soil, crop, and weather data collection. HFL with personalization is used to offer localized and adaptive insights. Model management, aggregation, and transfers are facilitated via a flexible approach, enabling seamless communication between local devices, edge nodes, and the cloud.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2020

SRS Strategy for Tank 3 Salt Dissolution - Performance and Lessons Learned - 20441

The Savannah River Site (SRS) Tank Farms have 51 underground waste tanks used to store and process liquid nuclear waste materials. There are 4 different tank types, ranging in capacity from 2,840,000 to 4,920,000 L (750,000 to 1,300,000 Gal). Twenty-four of the tanks are older style and do not meet full secondary containment standards. The older style tanks are the initial focus of waste removal efforts for tank closure at SRS. Eight of these twenty-four tanks have completed waste removal and are filled with grout. Prior to salt dissolution, Tank 3 was a dry salt tank that contained 5.06 m (199.3 in) of salt and sludge waste. Additional salt waste was present on cooling coils above the salt layer up to approximately 5.59 m (220 in). Three mixing eductors were installed in Tank 3 to aid in dissolving salt waste in three tank riser access ports. A transfer pump was installed, and the transfer pump suction was located 25.4 cm (10 in) from the tank bottom. Well water was added to the tank through a downcomer until the dry bulk salt was covered with liquid. During the initial fill of Tank 3, approximately 242,000 L (64,000 Gal) of rain water were added from periodic F-Tank Farm (FTF) Catch Tank additions and approximately 17,000 L (4,500 Gal) of well water were added. Following liquid additions to cover the dry bulk salt, well water was added through the three mixing eductors in batches during each stage. Additionally, during the salt dissolution campaigns, the FTF Catch Tank was utilized to add rain water through a downcomer in the center tank riser access port as needed for volume relief in the FTF Catch Tank. Following liquid additions, the water was recirculated (internal to the tank) using the transfer pump, and a sample was pulled to confirm the target specific gravity (SpG) of the dissolved salt solution was achieved. The dissolved salt solution was then transferred to the receipt tank (Tank 7), and the mixing eductors were lowered as close to the new bulk salt layer as possible, to support subsequent dissolution campaigns. While the mixing jets were able to dissolve salt successfully in Tank 3, they did not do so in a completely uniform manner. Throughout dissolution, mounds were discovered under Riser 1, Riser 2, and the Center Riser. The mixing jets dissolved the salt around the edges of the tank well but were not as effective toward the center of the tank. FTF Catch Tank additions through the center tank riser access port downcomer were required to impact the mound under the center tank riser. FTF Catch Tank additions were sporadic as they were dependent on rainfall in the area. Additionally, all mixing jets were rarely able to be lowered to the same elevation, indicating some mounding in the bulk salt layer. Indexing of the mixing jets was also utilized to impact the salt mounds. Overall, Tank 3 salt dissolution was successful as approximately 1,476,000 L (390,000 Gal) of dissolved salt solution was transferred to Tank 7 over six stages. Slightly less than 852,000 L (225,000 Gal) of water were added to Tank 3. The original material balance prediction estimated that the bulk salt level in Tank 3 would be 3.88 m (152.8 in) after six salt dissolution stages. After six stages the actual bulk salt level was 3.38 m (133 in). This paper will discuss the salt removal strategy, each salt dissolution stage, and lessons learned for future salt dissolution. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2023

CoolPINNs: A physics-informed neural network modeling of active cooling in vascular systems

Emerging technologies like hypersonic aircraft, space exploration vehicles, and batteries avail fluid circulation in embedded microvasculatures for efficient thermal regulation. Modeling is vital during the design and operational phases of these engineered systems. However, many challenges exist in developing a modeling framework. What is lacking is an accurate framework that (i) captures sharp jumps in the thermal flux across complex vasculature layouts, (ii) deals with oblique derivatives (involving tangential and normal components), (iii) handles nonlinearity because of radiative heat transfer, (iv) provides a high-speed forecast for real-time monitoring, and (v) facilitates robust inverse modeling. Here, this paper addresses these challenges by availing the power of physics-informed neural networks (PINNs). We develop a fast, reliable, and accurate Scientific Machine Learning (SciML) framework for vascular-based thermal regulation—called CoolPINNs: a PINNs-based modeling framework for active cooling. The proposed mesh-less framework elegantly overcomes all the mentioned challenges. The significance of the reported research is multi-fold. First, the framework is valuable for real-time monitoring of thermal regulatory systems because of rapid forecasting. Second, researchers can address complex thermoregulation designs since the approach is meshless. Finally, the framework facilitates systematic parameter identification and inverse modeling studies, perhaps the most significant utility of the current framework.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Phosphate Limitation Responses in Marine Green Algae Are Linked to Reprogramming of the tRNA Epitranscriptome and Codon Usage Bias

Marine algae are central to global carbon fixation, and their productivity is dictated largely by resource availability. Reduced nutrient availability is predicted for vast oceanic regions as an outcome of climate change; however, there is much to learn regarding response mechanisms of the tiny picoplankton that thrive in these environments, especially eukaryotic phytoplankton. Here, we investigate responses of the picoeukaryote Micromonas commoda, a green alga found throughout subtropical and tropical oceans. Under shifting phosphate availability scenarios, transcriptomic analyses revealed altered expression of transfer RNA modification enzymes and biased codon usage of transcripts more abundant during phosphate-limiting versus phosphate-replete conditions, consistent with the role of transfer RNA modifications in regulating codon recognition. To associate the observed shift in the expression of the transfer RNA modification enzyme complement with the transfer RNAs encoded by M. commoda, we also determined the transfer RNA repertoire of this alga revealing potential targets of the modification enzymes. Codon usage bias was particularly pronounced in transcripts encoding proteins with direct roles in managing phosphate limitation and photosystem-associated proteins that have ill-characterized putative functions in “light stress.” The observed codon usage bias corresponds to a proposed stress response mechanism in which the interplay between stress-induced changes in transfer RNA modifications and skewed codon usage in certain essential response genes drives preferential translation of the encoded proteins. Collectively, we expose a potential underlying mechanism for achieving growth under enhanced nutrient limitation that extends beyond the catalog of up- or downregulated protein-encoding genes to the cell biological controls that underpin acclimation to changing environmental conditions.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2020

Active Learning A Neural Network Model For Gold Clusters & Bulk From Sparse First Principles Training Data

Small metal clusters are of fundamental scientific interest and of tremendous significance in catalysis. These nanoscale clusters display diverse geometries and structural motifs depending on the cluster size; a knowledge of this size-dependent structural motifs and their dynamical evolution has been of longstanding interest. Given the high computational cost of first-principles calculations, molecular modeling and atomistic simulations such as molecular dynamics (MD) has proven to be an important complementary tool to aid this understanding. Classical MD typically employ predefined functional forms which limits their ability to capture such complex size-dependent structural and dynamical transformation. Neural Network (NN) based potentials represent flexible alternatives and in principle, well-trained NN potentials can provide high level of flexibility, transferability and accuracy on-par with the reference model used for training. A major challenge, however, is that NN models are interpolative and requires large quantities (similar to 10 4 or greater) of training data to ensure that the model adequately samples the energy landscape both near and far-from-equilibrium. A highly desirable goal is minimize the number of training data, especially if the underlying reference model is first-principles based and hence expensive. In this work, we introduce an active learning (AL) scheme that trains a NN model on-the-fly with minimal amount of first-principles based training data. Our AL workflow is initiated with a sparse training dataset (similar to 1 to 5 data points) and is updated on-the-fly via a Nested Ensemble Monte Carlo scheme that iteratively queries the energy landscape in regions of failure and updates the training pool to improve the network performance. Using a representative system of gold clusters, we demonstrate that our AL workflow can train a NN with similar to 500 total reference calculations. Using an extensive DFT test set of similar to 1100 configurations, we show that our AL-NN is able to accurately predict both the DFT energies and the forces for clusters of a myriad of different sizes. Our NN predictions are within 30 meV/atom and 40 meV/angstrom of the reference DFT calculations. Moreover, our AL-NN model also adequately captures the various size-dependent structural and dynamical properties of gold clusters in excellent agreement with DFT calculations and available experiments. We finally show that our AL-NN model also captures bulk properties reasonably well, even though they were not included in the training data.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗