SEARCH · Engineering Papers
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.
There is a lack of state-of-the-art quantum computing simulation software that scales on heterogeneous systems like Titan. Tensor Network Quantum Virtual Machine (TNQVM) provides a quantum simulator that leverages a distributed network of GPUs to simulate quantum circuits in a manner that leverages recent results from tensor network theory.
MARBLES (Multi-scale Adaptively Refined Boltzmann LatticE Solver) is an open-source computational fluid dynamics package powered by the lattice Boltzmann equations and built on AMReX. In the lattice Boltzmann method, local collisions between meso-scale fictitious particles drive the governing equations which enables MARBLES to easily simulate flow around complex and/or moving geometry without the generation of a body-conforming mesh. Using AMReX data structures and operations ensures a high level of computational performance and parallel scaling on heterogenous architectures while also naturally supporting locally enhanced grid resolution and fidelity through automatic mesh refinement. New domains and problem definitions are easily specified through an input file with examples and guidance on all options and variables provided in the MARBLES documentation.
There is a lack of state-of-the-art quantum computing simulation software that scales on heterogeneous systems like Titan. Tensor Network Quantum Virtual Machine (TNQVM) provides a quantum simulator that leverages a distributed network of GPUs to simulate quantum circuits in a manner that leverages recent results from tensor network theory.
Abstract A computational approach by an implementation of the principle component analysis (PCA) with K -means and Gaussian mixture (GM) clustering methods from machine learning algorithms to identify structural and dynamical heterogeneities of supercooled liquids is developed. In this method, a collection of the average weighted coordination numbers ( W C N s ‾ ) of particles calculated from particles’ positions are used as an order parameter to build a low-dimensional representation of feature (structural) space for K -means clustering to sort the particles in the system into few meso-states using PCA. Nano-domains or aggregated clusters are also formed in configurational (real) space from a direct mapping using associated meso-states’ particle identities with some misclassified interfacial particles. These classification uncertainties can be improved by a co-learning strategy which utilizes the probabilistic GM clustering and the information transfer between the structural space and configurational space iteratively until convergence. A final classification of meso-states in structural space and domains in configurational space are stable over long times and measured to have dynamical heterogeneities. Armed with such a classification protocol, various studies over the thermodynamic and dynamical properties of these domains indicate that the observed heterogeneity is the result of liquid–liquid phase separation after quenching to a supercooled state.
The performance of lithium ion electrodes is hindered by unfavorable chemical heterogeneities that pre-exist or develop during operation. Time-resolved spatial descriptions are needed to understand the link between such heterogeneities and a cell's performance. Here, operando high-resolution X-ray diffraction-computed tomography is used to spatially and temporally quantify crystallographic heterogeneities within and between particles throughout both fresh and degraded Li x Mn 2 O 4 electrodes. This imaging technique facilitates identification of stoichiometric differences between particles and stoichiometric gradients and phase heterogeneities within particles. Through radial quantification of phase fractions, the response of distinct particles to lithiation is found to vary; most particles contain localized regions that transition to rock salt LiMnO 2 within the first cycle. Other particles contain monoclinic Li 2 MnO 3 near the surface and almost pure spinel LixMn2O4 near the core. Following 150 cycles, concentrations of LiMnO 2 and Li 2 MnO 3 significantly increase and widely vary between particles.
Purpose: M5L, a fully automated computer-aided detection (CAD) system for the detection and segmentation of lung nodules in thoracic computed tomography (CT), is presented and validated on several image datasets. Methods: M5L is the combination of two independent subsystems, based on the Channeler Ant Model as a segmentation tool [lung channeler ant model (lungCAM)] and on the voxel-based neural approach. The lungCAM was upgraded with a scan equalization module and a new procedure to recover the nodules connected to other lung structures; its classification module, which makes use of a feed-forward neural network, is based of a small number of features (13), so as to minimize the risk of lacking generalization, which could be possible given the large difference between the size of the training and testing datasets, which contain 94 and 1019 CTs, respectively. The lungCAM (standalone) and M5L (combined) performance was extensively tested on 1043 CT scans from three independent datasets, including a detailed analysis of the full Lung Image Database Consortium/Image Database Resource Initiative database, which is not yet found in literature. Results: The lungCAM and M5L performance is consistent across the databases, with a sensitivity of about 70% and 80%, respectively, at eight false positive findings per scan, despite the variable annotation criteria and acquisition and reconstruction conditions. A reduced sensitivity is found for subtle nodules and ground glass opacities (GGO) structures. A comparison with other CAD systems is also presented. Conclusions: The M5L performance on a large and heterogeneous dataset is stable and satisfactory, although the development of a dedicated module for GGOs detection could further improve it, as well as an iterative optimization of the training procedure. The main aim of the present study was accomplished: M5L results do not deteriorate when increasing the dataset size, making it a candidate for supporting radiologists on large scale screenings and clinical programs.
Optimum solutions for geologic modeling and reservoir simulation in industries such as oil and gas recovery and carbon capture and storage require accurate characterization of reservoir properties, which are often heterogeneous. In this study, high-quality micro-computed tomography (CT) images (1.475-μm/pixel resolution) of a sandstone core acquired from the Bell Creek oil field, USA, were used to provide nondestructive analysis of pore- and core-scale heterogeneity across measurement scales of 94–566 μm. In addition to characterizing the as-received sample, the core sample was flooded with brine to evaluate the capacity of the core sample to receive injected fluids. The micro-CT images were systematically segmented into pore spaces and grains via machine learning (ML) steps including image preprocessing, label creation using a traditional ML method based on limited manual image annotation, and finally U-Net segmentation. The segmented image stacks were reconstructed into digital cubes of various scales of voxel lengths. The 3D porosity values were calculated for all the digital cubes, and the fractal dimensions of the cubes were estimated using a box-counting method. The results showed that smaller cubes had greater heterogeneity and that the porosity values could be accurately estimated by fractal dimension and voxel lengths using ML models. For the core sample with brine flooding, the ratio of pores filled by brine to the total pore space was related to the porosity and could also be accurately estimated by porosity, fractal dimension, and voxel lengths using ML models. In conclusion, the results of this study demonstrate that the concept of fractal dimension can be a useful vector to perform upscaling analysis of sandstone rock heterogeneity from the pore to core scale and that fractal dimensions can be used to estimate porosity values and pore space-filling capacity across those scales.
Computational chemistry, especially density functional theory, has experienced a remarkable growth in terms of application over the last few decades. This is attributed to the improvements in theory and computing infrastructure that enable the analysis of systems of unprecedented size and detail at an affordable computational expense. In this perspective, we discuss recent progress and current challenges facing electronic structure theory in the context of heterogeneous catalysis. We specifically focus on the impact of computational chemistry in elucidating and designing catalytic systems in three topics of interest to Haldor Topsøe – ammonia, synthesis, hydrotreating, and NO x reduction. Furthermore, we then discuss the common tools and concepts in computational catalysis that underline these topics and provide a perspective on the challenges and future directions of research in this area of catalysis research.
We present a rigorous homogenization approach for efficient computation of a class of physical problems in a one-dimensional periodic heterogeneous material. This material is represented by a spatially periodic array of unit cells with a length of e. More specifically, the method is applied to the diffusion, heat conduction, and wave propagation problems. Heterogeneous materials can have arbitrary position-dependent continuous or discontinuous materials properties (for example heat conductivity) within the unit cell. The final effective model includes both effective properties at the leading order and high-order contributions due to the microscopic heterogeneity. A dimensionless heterogeneity parameter ß is defined to represent high-order contributions, shown to be in the range of [-1/12, 0], and has a universal expression for all three problems. Both effective properties and heterogeneity parameter ß are independent of e, the microscopic scale of heterogeneity. The homogenized solution describing macroscopic variations can be obtained from the effective model. Solution with sub-unit-cell accuracy can be constructed based on the homogenized solution and its spatial derivatives. The paper represents a general approach to obtain the effective model for arbitrary periodic heterogeneous materials with position-dependent properties.
As large language models scale to trillions of parameters, their computational and memory requirements present critical challenges for efficient training and deployment. While Mixture of Experts (MoE) architectures enable efficient scaling through sparse parameter activation, and state-space models like Mamba offer linear-time complexity, principled methods for combining these paradigms remain undeveloped. We introduce HARMONY (Hybrid Architecture Research for Mamba, Optimized with Neural efficiencY), a multi-objective evolutionary neural architecture search framework for discovering efficient hybrid language models that integrate Transformer attention mechanisms, Mixture-of-Experts routing, and Mamba state-space components. Through large-scale distributed search using 16,384 MI250X GPUs on the Frontier supercomputer, HARMONY explores a comprehensive design space encompassing six attention variants (MHA, MQA, GQA, MLA, SWA, and Mamba-2), variable MoE configurations with both routed and shared experts, and extensive Mamba hyperparameters. Our framework discovers heterogeneous architectures that balance training performance with computational efficiency through multi-objective optimization incorporating latency penalties and fitness-based selection. Analysis of discovered architectures reveals that optimal hybrid designs favor heterogeneous component mixing rather than homogeneous patterns, with Mamba-2 and Multi-Head Latent Attention (MLA) emerging as preferred mechanisms. Discovered architectures demonstrate superior training efficiency: our best configuration achieves a final perplexity of 1.0874 with 2.38B parameters while processing 4,320 tokens/second, outperforming significantly larger manually designed models. Full-scale evaluation shows HARMONY's top architectures achieve better loss trajectories than equivalently-sized models using state-of-the-art configurations including Mixtral, Jamba, and Samba. Additionally, we demonstrate 91% weak scaling efficiency when training discovered 36B-parameter models across 1,024 GPUs. HARMONY is released as an open framework with comprehensive tools for building and training hybrid models using expert-data-pipeline parallelism, democratizing access to automated architecture design for next-generation language models.
Numerical building models are typically forced with weather data from a limited number of “representative cities” or weather stations representing different climate regions. The use of representative weather stations reduces computational costs, but often fails to capture spatial heterogeneity in weather that may be important for simulations aimed at understanding how building stocks respond to a changing climate. We quantify the potential reduction in bias from using an increasing number of weather stations over the western U.S. The approach is based on deriving temperature and load time series using incrementally more weather stations, ranging from 8 to roughly 150, to capture weather across different seasons. Using 8 stations, one from each climate zone, across the western U.S. results in an average absolute summertime temperature bias of 7.2°F with respect to a spatially-resolved gridded dataset. The mean absolute bias drops to 2.8°F using all available weather stations. Temperature biases of this magnitude could translate to absolute summertime mean simulated load biases as high as 13.8%, a significant error for capacity expansion planners who may use these types of simulations. Increasing the size of the domain over which biases are calculated reduces their magnitude as positive and negative biases may cancel out. Using 8 representative weather stations can lead to a 20-40% overestimation of peak building loads during both summer and winter. Using weather stations close to population centers reduces both mean and peak load biases. This approach could be used by others designing aggregate building simulations to understand the sensitivity to their choice of weather stations used to drive the models.
A new method for computing homogenized assembly neutron transport cross sections and dif- fusion coefficients that is both rigorous and computationally efficient is proposed in this paper. In the limit of a homogeneous hydrogen slab, the new method is equivalent to the long-used, and only-recently-published CASMO transport method. The rigorous method is used to demonstrate the sources of inaccuracy in the commonly applied “out-scatter” transport correction. It is also demonstrated that the newly developed method is directly applicable to lattice calculations per- formed by Monte Carlo and is capable of computing rigorous homogenized transport cross sections for arbitrarily heterogeneous lattices. Comparisons of several common transport cross section ap- proximations are presented for a simple problem of infinite medium hydrogen. The new method has also been applied in computing 2-group diffusion data for an actual PWR lattice from BEAVRS benchmark.
Heterogeneity continues to increase at all levels of computing, with the rise of accelerators such as GPUs, FPGAs, and other co-processors into everything from desktops to supercomputers. As a consequence, efficiently managing such disparate resources has become increasingly complex. CoreTSAR seeks to reduce this complexity by adaptively worksharing parallel-loop regions across compute resources without requiring any transformation of the code within the loop. Lastly, our results show performance improvements of up to three-fold over a current state-of-the-art heterogeneous task scheduler as well as linear performance scaling from a single GPU to four GPUs for many codes. In addition, CoreTSAR demonstrates a robust ability to adapt to both a variety of workloads and underlying system configurations.
Most applications of machine learning in heterogeneous catalysis thus far have used black-box models to predict computable physical properties (descriptors), such as adsorption or formation energies, that can be related to catalytic performance (that is, activity or stability). Here, extracting meaningful physical insights from these black-box models has proved challenging, as the internal logic of these black-box models is not readily interpretable due to their high degree of complexity. Interpretable machine learning methods that merge the predictive capacity of black-box models with the physical interpretability of physics-based models offer an alternative to black-box models. In this Perspective, we discuss the various interpretable machine learning methods available to catalysis researchers, highlight the potential of interpretable machine learning to accelerate hypothesis formation and knowledge generation, and outline critical challenges and opportunities for interpretable machine learning in heterogeneous catalysis.
The article mainly focuses on the enrichment of the output power obtained from Spin torque nano-oscillator by introducing the heterogeneous structure in multilayer nanopillar device. Here we devised two homogeneous and two heterogeneous devices having NiFe and Co materials. The dynamics of the devices are governed by a famous Landu- Lifshitz -Gilbert-Solencskwei (LLGS) equation which can be solved numerically using embedded RK-4 procedure. The current density and the external magnetic field for four devices are taken as 5×l0{sup 11}A/m{sup 2} and 6×l0{sup −3} A/m respectively. The applied dc current is converted into spin polarized dc current while it passes through pinned layer. The generated spin polarized dc currents produces spin transfer torque with the free layer magnetization via spacer. Thus the magnetization of the free layer gets a sustained oscillation. The results obtained from the heterogeneous STNOs are really fascinating. The frequency of the NiFe/ Cu/NiFe and Co/Cu/NiFe devices have the same frequency but there is a tremendous change in the output power which is exactly twice that the NiFe/Cu/NiFe device. The similar behaviour is also obtained from Co/Cu/Co and NiFe/Cu/Co devices. The line width and the Q-factor of the output microwave signal are also computed. Among the four devices, the NiFe/Cu/Co heterogeneous device has low linewidth (408 MHz) and high Q-factor (4.77).
The pre-exascale systems are expected to have a significant amount of hierarchical and heterogeneous on-node memory, and this trend of system architecture in extreme-scale systems is expected to continue into the exascale era. Along with hierarchical-heterogeneous memory, the system typically has a high-performing network and a compute accelerator. This system architecture is not only effective for running traditional High Performance Computing (HPC) applications (Big-Compute), but also running data-intensive HPC applications and Big-Data applications. As a consequence, there is a growing desire to have a single system serve the needs of both Big-Compute and Big-Data applications. Though the system architecture supports the convergence of the Big-Compute and Big-Data, the programming models have yet to evolve to support either hierarchical-heterogeneous memory systems or the convergence. In this work, we propose and develop the programming abstraction called SHARed data-structure centric Programming abstraction (SharP) to address both of these goals, i.e., provide (1) a simple, usable, and portable abstraction for hierarchical-heterogeneous memory and (2) a unified programming abstraction for Big-Compute and Big-Data applications. To evaluate SharP, we implement a Stencil benchmark using SharP, port QMCPack, a petascale-capable application, and adapt Memcached ecosystem, a popular Big-Data framework, to use SharP, and quantify the performance and productivity advantages. Additionally, we demonstrate the simplicity of using SharP on different memories including DRAM, High-bandwidth Memory (HBM), and non-volatile random access memory (NVRAM).
Cell membranes contain a large variety of lipid types and are crowded with proteins, endowing them with the plasticity needed to fulfill their key roles in cell functioning. The compositional complexity of cellular membranes gives rise to a heterogeneous lateral organization, which is still poorly understood. Computational models, in particular molecular dynamics simulations and related techniques, have provided important insight into the organizational principles of cell membranes over the past decades. Now, we are witnessing a transition from simulations of simpler membrane models to multicomponent systems, culminating in realistic models of an increasing variety of cell types and organelles. Here, we review the state of the art in the field of realistic membrane simulations and discuss the current limitations and challenges ahead.
A collection of functions for statistical computing and data manipulation. These include routines for rapidly aggregating heterogeneous matrices, manipulating file names, loading R objects, sourcing multiple R files, formatting datetimes, multi-core parallel computing, stream editing, specialized plotting, etc. Smisc-package A collection of miscellaneous functions allMissing Identifies missing rows or columns in a data frame or matrix as.numericSilent Silent wrapper for coercing a vector to numeric comboList Produces all possible combinations of a set of linear model predictors cumMax Computes the maximum of the vector up to the current index cumsumNA Computes the cummulative sum of a vector without propogating NAs d2binom Probability functions for the sum of two independent binomials dataIn A flexible way to import data into R. dbb The Beta-Binomial Distribution df2list Row-wise conversion of a data frame to a list dfplapply Parallelized single row processing of a data frame dframeEquiv Examines the equivalence of two dataframes or matrices dkbinom Probability functions for the sum of k independent binomials factor2character Converts all factor variables in a dataframe to character variables findDepMat Identify linearly dependent rows or columns in a matrix formatDT Converts date or datetime strings into alternate formats getExtension Filename manipulations: remove the extension or path, extract the extension or path getPath Filename manipulations: remove the extension or path, extract the extension or path grabLast Filename manipulations: remove the extension or path, extract the extension or path ifelse1 Non-vectorized version of ifelse integ Simple numerical integration routine interactionPlot Two-way Interaction Plot with Error Bar linearMap Linear mapping of a numerical vector or scalar list2df Convert a list to a data frame loadObject Loads and returns the object(s) in an ".Rdata" file more Display the contents of a file to the R terminal movAvg2 Calculate the moving average using a 2-sided window openDevice Opens a graphics device based on the filename extension p2binom Probability functions for the sum of two independent binomials padZero Pad a vector of numbers with zeros parseJob Parses a collection of elements into (almost) equal sized groups pbb The Beta-Binomial Distribution pcbinom A continuous version of the binomial cdf pkbinom Probability functions for the sum of k independent binomials plapply Simple parallelization of lapply plotFun Plot one or more functions on a single plot PowerData An example of power data pvar Prints the name and value of one or more objects qbb The Beta-Binomial Distribution rbb And numerous others (space limits reporting).