Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “ITERATIVE NETWORK”

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 451 records · Page 25

The Astronomy Commons Platform: A Deployable Cloud-based Analysis Platform for Astronomy

Abstract We present a scalable, cloud-based science platform solution designed to enable next-to-the-data analyses of terabyte-scale astronomical tabular data sets. The presented platform is built on Amazon Web Services (over Kubernetes and S3 abstraction layers), utilizes Apache Spark and the Astronomy eXtensions for Spark for parallel data analysis and manipulation, and provides the familiar JupyterHub web-accessible front end for user access. We outline the architecture of the analysis platform, provide implementation details and rationale for (and against) technology choices, verify scalability through strong and weak scaling tests, and demonstrate usability through an example science analysis of data from the Zwicky Transient Facility’s 1Bn+ light-curve catalog. Furthermore, we show how this system enables an end user to iteratively build analyses (in Python) that transparently scale processing with no need for end-user interaction. The system is designed to be deployable by astronomers with moderate cloud engineering knowledge, or (ideally) IT groups. Over the past 3 yr, it has been utilized to build science platforms for the DiRAC Institute, the ZTF partnership, the LSST Solar System Science Collaboration, and the LSST Interdisciplinary Network for Collaboration and Computing, as well as for numerous short-term events (with over 100 simultaneous users). In a live demo instance, the deployment scripts, source code, and cost calculators are accessible. 4 4 http://hub.astronomycommons.org/

79 ASTRONOMY AND ASTROPHYSICS↗

STOMP-ECKEChem: An Engineering Perspective on Reactive Transport in Geologic Media

ECKEChem (Equilibrium, Conservation, Kinetic Equation Chemistry) is a reactive transport module for the STOMP suite of multifluid subsurface flow and transport simulators that was developed from an engineering perspective. STOMP comprises a suite of operational modes that are distinguished by the solved coupled conservation equations with capabilities for a variety of subsurface applications (e.g., environmental remediation and stewardship, geologic sequestration of greenhouse gases, gas hydrate production, and oil shale production). The ECKEChem module was designed to provide integrated reactive transport capabilities across the suite of STOMP simulator operational modes. The initial application for the ECKEChem module was in the simulation of the mineralization reactions that occurred with the injection of supercritical carbon dioxide into deep Columbia River basalt formations, where it was implemented in the STOMP-CO2 simulator. The STOMP-ECKEChem solution approach to modeling reactive transport in multifluid geologic media is founded on an engineering perspective: (1) sequential non-iterative coupling between the flow and reactive transport is sufficient, (2) reactive transport can be modeled by operator splitting with local geochemistry and global transport, (3) geochemistry can be expressed as a system of coupled nonlinear equilibrium, conservation and kinetic equations, (4) a limited number of kinetic equation forms are used in geochemical practice. This chapter describes the conceptual approach to converting a geochemical reaction network into a series of equilibrium, conservation and kinetic equations, the implementation of ECKEChem in STOMP, the numerical solution approach, and a demonstration of the simulator on a complex application involving desorption of uranium from contaminated field-textured sediments.

04 OIL SHALES AND TAR SANDS↗

Computationally efficient Bayesian estimation of graphical networks for omics data

Graphical networks are useful, widely-used modeling approaches to represent complex biological processes with biological measurements generated by platforms such as mass spectrometry. Bayesian analyses of graphical networks for omics data have several advantages over their frequentist counterparts, such as the inclusion of prior knowledge in the estimation of models. However, Bayesian approaches to date have only been feasible for data with a couple hundred biomolecules due to prohibitive computational time, but omics data often contains tens of thousands of biomolecules. Here, we present and illustrate a more computationally efficient approach named BPlane (Bayesian PseudoLikelihood-based Algorithm for Network Estimation) to extend Bayesian modeling capabilities for larger-sized datasets, such as most untargeted proteomics data. Via simulation, we demonstrate that BPlane produces substantial computational savings over a current state-of-the-art Bayesian algorithm while maintaining competitive edge detection accuracy. On a SARS-CoV2 proteomics data with 7000 proteins, the competing algorithm takes three times as long to complete the first iteration as BPlane takes to converge after over 100 iterations.

EM algorithm↗

wa-hls4ml and lui-gnn: A benchmark and GNN-based surrogate model for hls4ml resource and latency estimation

As machine learning (ML) increasingly serves as a tool for addressing real-time challenges in scientific applications, the development of advanced tooling has significantly reduced the time required to iterate on various designs. These advancements have solved major obstacles, but also exposed new challenges. For example, processes that were not previously considered bottlenecks, such as model synthesis, are now becoming limiting factors in the rapid iteration of designs. To reduce these emerging constraints, multiple efforts are being launched toward designing an ML-based surrogate model that estimates resource usage of synthesized accelerator architectures. This model would reduce the design iteration time, especially when designing within a set of given hardware constraints. This approach shows considerable potential, but as it stands, the effort is early and would benefit from coordination and standardization to assist future work as it emerges. We introduce wa-hls4ml, a benchmark for ML accelerator resource and latency estimation, and its corresponding initial dataset of more than 100,000 fully connected neural networks, all synthesized using hls4ml and targeting Xilinx FPGAs. In addition to the resource utilization and latency data provided, the dataset includes generated artifacts and log files for many of the synthesized neural networks, in order to support future research in ML-based code generation. The benchmark evaluates the performance of resource and latency predictors against several common ML model architectures, primarily originating from scientific domains, as exemplar models, as well as the average performance across a subset of the dataset. We measure the performance of a given predictor model through multiple metrics, including $R^2$ score and SMAPE on regression tasks, as well as inference time to further characterize the estimator under test. Additionally, we introduce the latency/utilization inference graph neural network (lui-gnn), a surrogate model that uses a graph neural network to represent input architectures in the form of a directed graph. This graph representation allows for a diverse set of model architectures to all be effectively handled by a surrogate model. We present the architecture and performance of the model, as evaluated by the new proposed benchmark, including SMAPE, $R^2$ score, and inference times, and find that lui-gnn generally predicts latency and utilization for the 75\% quantile within several percent of the synthesized resources on the synthetic test dataset, indicating that this approach of estimating resource and latency via a surrogate models has promise and warrants further research.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Machine learning guided selection of broad-spectrum epitope-specific functional antibodies for "Disease X"

Our project established and demonstrated a transfer learning framework that enables prediction of antibody–antigen interactions across related viruses. The approach focused on three major activities: 1. Conserved region and epitope identification – We compared viral protein structures and sequences to identify shared receptor-binding domains and neutralizing epitope regions across variants and related viruses. These conserved features formed the foundation for discovering broadly functional antibodies. 2. Machine learning model development – We built neural network–based models that integrate epitope features with antibody sequence information. Instead of relying solely on structural or physical properties, the models learned transferable patterns that describe antibody binding potential across different viral families. 3. Transfer learning and validation – Using SARS-CoV-2 and Ebola as source systems, we successfully transferred learned epitope features to predict antibody interactions for SARS CoV-1 and Marburg virus. Iterative cycles of dataset generation, retraining, and evaluation improved generalization and predictive power, ensuring the framework can adapt to new threats.

59 BASIC BIOLOGICAL SCIENCES↗

Scale-invariant machine-learning model accelerates the discovery of quaternary chalcogenides with ultralow lattice thermal conductivity

We design an advanced machine-learning (ML) model based on crystal graph convolutional neural network that is insensitive to volumes (i.e., scale) of the input crystal structures to discover novel quaternary chalcogenides, AMM'Q 3 (A/M/M' = alkali, alkaline earth, post-transition metals, lanthanides, and Q = chalcogens). These compounds are shown to possess ultralow lattice thermal conductivity (κ l ), a desired requirement for thermal-barrier coatings and thermoelectrics. Upon screening the thermodynamic stability of ~1 million compounds using the ML model iteratively and performing density-functional theory (DFT) calculations for a small fraction of compounds, we discover 99 compounds that are validated to be stable in DFT. Taking several DFT-stable compounds, we calculate their κ l using Peierls–Boltzmann transport equation, which reveals ultralow κ l (<2 Wm -1 K -1 at room temperature) due to their soft elasticity and strong phonon anharmonicity. Our work demonstrates the high efficiency of scale-invariant ML model in predicting novel compounds and presents experimental-research opportunities with these new compounds.

36 MATERIALS SCIENCE↗

Reinforcement Learning of Structured Stabilizing Control for Linear Systems With Unknown State Matrix

This paper delves into designing feedback control gains for a continuous-time linear quadratic regulator (LQR) problem that is constrained to certain predefined structure with unknown state matrix. We bring forth the ideas from reinforcement learning (RL) in conjunction with sufficient stability and performance guarantees in order to design these structured gains using the trajectory measurements of states and controls. Here we first formulate a model-based framework using dynamic programming (DP) to embed the structural constraint to the LQR gain computation in the continuous-time setting, and then subsequently, formulate a policy iteration RL algorithm that can alleviate the requirement of known state matrix in conjunction with maintaining the feedback gain structure. The design enables a distributed learning control design which is necessary for many large-scale cyber-physical systems. Theoretical guarantees are provided for stability and convergence of the structured reinforcement learning (SRL) algorithm. We validate our theoretical results with numerical simulations on a multi-agent networked linear time-invariant (LTI) dynamic system.

42 ENGINEERING↗

A search for the Higgs boson in H→WW→lvlv with 802 fb -1

This thesis documents a search for Standard Model (SM) Higgs boson to WW production in the final state of two charged leptons (e, μ) and two neutrino final state resulting from in pp collisions at √ s = 1.96 TeV. The data collected by the CDF II detector at the Tevatron collider at Fermilab, corresponds to an integrated luminosity of 8.2 fb -1 . This iteration of the analysis incorporated several improvements to increase Higgs sensitivity. We reevaluated lepton isolation values to recover signal events in which leptons mutually spoil each other’s isolation requirements. We additionally incorporated new lepton identification selections, namely likelihood based forward electrons and a better recovery of electrons passing through detector cracks. The base analysis searches for opposite-sign dilepton events. We employed three separately trained neural networks to distinguish signal from background processes for events with 0, one, or two jets. To further increase sensitivity for 0-jet events, we constructed a likelihood ratio discriminant based on Matrix Element calculations as an additional input to the neural network. For the base opposite-sign analysis, summing over all jet multiplicities, we observed 3513 events in data compared with a background expectation of 3409 ± 233 and a signal expectation of 53.6 ± 9.4 for a mass of 165 GeV/c 2 . To further increase signal acceptance, we incorporated several separate search regions in addition to the base analysis. We include subchannels searching for Higgs production in events with low dilepton invariant mass and in events having final states with either same-sign dileptons or trileptons With the combination of the separate analyses, we observed a 95% C.L. upper production limit of 0.77 times the SM expectation for a Higgs mass of 165 GeV/c 2 to be compared with the value for the median of the expected limit (0.78). We additionally present results for eighteen Higgs masses ranging from 110 GeV/c 2 to 200 GeV/c 2 . We excluded at the 95% C.L., a SM Higgs boson in the mass range between 156 and 175 GeV/c 2 .

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Hierarchical Distributed Optimal Power Flow of HV and MV Distribution Networks With Continuous and Discrete Devices

With large-scale distributed photovoltaics (PVs) being integrated into distribution networks (DNs), coordinated optimal power flow (OPF) of high voltage (HV) and medium voltage (MV) DNs should be investigated to optimally dispatch the distributed PVs and other network devices. Here, this paper presents a hierarchical distributed OPF method for HV and MV DNs with on-load tap changers, reactive power compensators, feeder switches and distributed PVs. A hierarchical master-slave control architecture is applied to implement coordinated OPF of two-layer DNs. The HV master problem and MV subproblems are transformed into mixed-integer convex problems respectively with second order cone programming and LinDistFlow approximation. Since there is no efficient distributed algorithm to solve such OPF models with integer subproblems, a novel distributed algorithm is proposed in this paper to efficiently solve the hierarchical coordinated OPF model with integer subproblems in a distributed manner. In the proposed algorithm, the coordinated OPF model is solved in a branch-and-bound framework, where in each branch node generalized Benders decomposition (GBD) algorithm is applied to decompose the coordinated OPF model into a master problem and relaxed subproblems and solves them iteratively to get optimal solution. The GBD optimal and feasible cutting planes generated in a branch node are proved to be valid for its descendants. Moreover, three acceleration techniques are introduced into the proposed algorithm to improve computational efficiency. Finally, the effectiveness and accuracy of the proposed method are verified via simulation tests in Jinzhai DNs of China.

42 ENGINEERING↗

Addressing numerical challenges in introducing a reactive transport code into a land surface model: a biogeochemical modeling proof-of-concept with CLM–PFLOTRAN 1.0

Abstract. We explore coupling to a configurable subsurface reactive transport code as a flexible and extensible approach to biogeochemistry in land surface models. A reaction network with the Community Land Model carbon–nitrogen (CLM-CN) decomposition, nitrification, denitrification, and plant uptake is used as an example. We implement the reactions in the open-source PFLOTRAN (massively parallel subsurface flow and reactive transport) code and couple it with the CLM. To make the rate formulae designed for use in explicit time stepping in CLMs compatible with the implicit time stepping used in PFLOTRAN, the Monod substrate rate-limiting function with a residual concentration is used to represent the limitation of nitrogen availability on plant uptake and immobilization. We demonstrate that CLM–PFLOTRAN predictions (without invoking PFLOTRAN transport) are consistent with CLM4.5 for Arctic, temperate, and tropical sites.Switching from explicit to implicit method increases rigor but introduces numerical challenges. Care needs to be taken to use scaling, clipping, or log transformation to avoid negative concentrations during the Newton iterations. With a tight relative update tolerance (STOL) to avoid false convergence, an accurate solution can be achieved with about 50 % more computing time than CLM in point mode site simulations using either the scaling or clipping methods. The log transformation method takes 60–100 % more computing time than CLM. The computing time increases slightly for clipping and scaling; it increases substantially for log transformation for half saturation decrease from 10-3 to 10-9 mol m-3, which normally results in decreasing nitrogen concentrations. The frequent occurrence of very low concentrations (e.g. below nanomolar) can increase the computing time for clipping or scaling by about 20 %, double for log transformation. Overall, the log transformation method is accurate and robust, and the clipping and scaling methods are efficient. When the reaction network is highly nonlinear or the half saturation or residual concentration is very low, the allowable time-step cuts may need to be increased for robustness for the log transformation method, or STOL may need to be tightened for the clipping and scaling methods to avoid false convergence.As some biogeochemical processes (e.g., methane and nitrous oxide reactions) involve very low half saturation and thresholds, this work provides insights for addressing nonphysical negativity issues and facilitates the representation of a mechanistic biogeochemical description in Earth system models to reduce climate prediction uncertainty.

58 GEOSCIENCES↗

Nonlinear extensions of a fractal-multifractal approach for environmental modeling

We present the extension of a deterministic fractal geometric procedure aimed at representing the complexity of the spatio-temporal patterns encountered in environmental applications. The original procedure, which is based on transformations of multifractal distributions via fractal functions, is extended through the introduction of nonlinear perturbations to the underlying iterated linear maps. We demonstrate how the nonlinear perturbations generate yet a richer collection of patterns by means of various simulations that include evolutions of patterns based on changes in their parameters and in their statistical and multifractal properties. It is shown that the nonlinear extensions yield structures that closely resemble complex hydrologic temporal data sets, such as rainfall and runoff time series, and width-functions of river networks as a function of distance from the basin outlet. The implications of this nonlinear approach for environmental modeling and prediction are discussed.

54 ENVIRONMENTAL SCIENCES↗

Grappolo

Community detection has become a fundamental operation in numerous graph-theoretic applications. It is used to reveal natural divisions that exist within real world networks without imposing prior size or cardinality constraints on the set of communities. Despite its potential for application, there is only limited support for community detection on large-scale parallel computers, largely owing to the irregular and inherently sequential nature of the underlying heuristics. In this paper, we present parallelization heuristics for fast community detection using the Louvain method as the serial template. The Louvain method is an iterative heuristic for modularity optimization. Originally developed by Blondel et al. in 2008, the method has become increasingly popular owing to its ability to detect high modularity community partitions in a fast and memory-efficient manner. However, the method is also inherently sequential, thereby limiting its scalability. Here, we observe certain key properties of this method that present challenges for its parallelization, and consequently propose heuristics that are designed to break the sequential barrier. For evaluation purposes, we implemented our heuristics using OpenMP multithreading, and tested them over real world graphs derived from multiple application domains (e.g., internet, citation, biological). Compared to the serial Louvain implementation, our parallel implementation is able to produce community outputs with a higher modularity for most of the inputs tested, in comparable number of iterations, while providing real speedups of up to 8x using 32 threads. In addition, our parallel implementation was able to exhibit weak scaling properties on up to 32 threads.

Halappanavar, Mahantesh↗

Parallel Heuristics for Scalable Community Detection

Community detection has become a fundamental operation in numerous graph-theoretic applications. It is used to reveal natural divisions that exist within real world networks without imposing prior size or cardinality constraints on the set of communities. Despite its potential for application, there is only limited support for community detection on large-scale parallel computers, largely owing to the irregular and inherently sequential nature of the underlying heuristics. In this paper, we present parallelization heuristics for fast community detection using the Louvain method as the serial template. The Louvain method is an iterative heuristic for modularity optimization. Originally developed by Blondel et al. in 2008, the method has become increasingly popular owing to its ability to detect high modularity community partitions in a fast and memory-efficient manner. However, the method is also inherently sequential, thereby limiting its scalability to problems that can be solved on desktops. Here, we observe certain key properties of this method that present challenges for its parallelization, and consequently propose multiple heuristics that are designed to break the sequential barrier. Our heuristics are agnostic to the underlying parallel architecture. For evaluation purposes, we implemented our heuristics on shared memory (OpenMP) and distributed memory (MapReduce-MPI) machines, and tested them over real world graphs derived from multiple application domains (internet, biological, natural language processing). Experimental results demonstrate the ability of our heuristics to converge to high modularity solutions comparable to those output by the serial algorithm in nearly the same number of iterations, while also drastically reducing time to solution.

clustering, graph algorithms↗

Improving photovoltaic hosting capacity of distribution networks with coordinated inverter control: A case study of the EPRI J1 feeder

Abstract Adding photovoltaic (PV) systems in distribution networks, while desirable for reducing the carbon footprint, can lead to voltage violations under high solar‐low load conditions. The inability of traditional volt‐VAr control in eliminating all the violations is also well‐known. This article presents a novel coordinated inverter control methodology that leverages system‐wide situational awareness to significantly improve hosting capacity (HC). The methodology employs a real‐time voltage‐reactive power (VQ) sensitivity matrix in an iterative linear optimizer to calculate the minimum reactive power intervention from PV inverters needed for mitigating over‐voltage without resorting to active power curtailing or requiring step voltage regulator setting changes. The algorithm is validated using the EPRI J1 feeder under an extensive set of realistic use cases and is shown to provide 3x improvement in HC under all scenarios.

Dalal, Dhaval [School of Electrical, Computer, and↗

MultiLoad-GAN: A GAN-Based Synthetic Load Group Generation Method Considering Spatial-Temporal Correlations

This paper presents a deep-learning framework, Multi-load Generative Adversarial Network (MultiLoad-GAN), for generating a group of synthetic load profiles (SLPs) simultaneously. The main contribution of MultiLoad-GAN is the capture of spatial-temporal correlations among a group of loads that are served by the same distribution transformer. This enables the generation of a large amount of correlated SLPs required for microgrid and distribution system studies. Here, the novelty and uniqueness of the MultiLoad-GAN framework are three-fold. First, to the best of our knowledge, this is the first method for generating a group of load profiles bearing realistic spatial- temporal correlations simultaneously. Second, two complementary realisticness metrics for evaluating generated load profiles are developed: computing statistics based on domain knowledge and comparing high-level features via a deep-learning classifier. Third, to tackle data scarcity, a novel iterative data augmentation mechanism is developed to generate training samples for enhancing the training of both the classifier and the MultiLoad-GAN model. Simulation results show that MultiLoad- GAN can generate more realistic load profiles than existing approaches, especially in group level characteristics. With little finetuning, MultiLoad-GAN can be readily extended to generate a group of load or PV profiles for a feeder or a service area.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Physics-informed neural network with transfer learning (TL-PINN) based on domain similarity measure for prediction of nuclear reactor transients

Nuclear reactor safety and efficiency can be enhanced through the development of accurate and fast methods for prediction of reactor transient (RT) states. Physics informed neural networks (PINNs) leverage deep learning methods to provide an alternative approach to RT modeling. Applications of PINNs in monitoring of RTs for operator support requires near real-time model performance. However, as with all machine learning models, development of a PINN involves time-consuming model training. Here, we show that a transfer learning (TL-PINN) approach achieves significant performance gain, as measured by reduction of the number of iterations for model training. Using point kinetic equations (PKEs) model with six neutron precursor groups, constructed with experimental parameters of the Purdue University Reactor One (PUR-1) research reactor, we generated different RTs with experimentally relevant range of variables. The RTs were characterized using Hausdorff and Fréchet distance. We have demonstrated that pre-training TL-PINN on one RT results in up to two orders of magnitude acceleration in prediction of a different RT. The mean error for conventional PINN and TL-PINN models prediction of neutron densities is smaller than 1%. We have developed a correlation between TL-PINN performance acceleration and similarity measure of RTs, which can be used as a guide for application of TL-PINNs.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Efficient Mixed-Precision Matrix Factorization of the Inverse Overlap Matrix in Electronic Structure Calculations with AI-Hardware and GPUs

In recent years, a new kind of accelerated hardware has gained popularity in the artificial intelligence (AI) community which enables extremely high-performance tensor contractions in reduced precision for deep neural network calculations. In this article, we exploit Nvidia Tensor cores, a prototypical example of such AI-hardware, to develop a mixed precision approach for computing a dense matrix factorization of the inverse overlap matrix in electronic structure theory, S –1 . This factorization of S –1 , written as ZZT = S –1 , is used to transform the general matrix eigenvalue problem into a standard matrix eigenvalue problem. Here we present a mixed precision iterative refinement algorithm where Z is given recursively using matrix–matrix multiplications and can be computed with high performance on Tensor cores. To understand the performance and accuracy of Tensor cores, comparisons are made to GPU-only implementations in single and double precision. Additionally, we propose a nonparametric stopping criteria which is robust in the face of lower precision floating point operations. The algorithm is particularly useful when we have a good initial guess to Z, for example, from previous time steps in quantum-mechanical molecular dynamics simulations or from a previous iteration in a geometry optimization.

36 MATERIALS SCIENCE↗

AWB-GCN: A Graph Convolutional Network Accelerator with Runtime Workload Rebalancing

The recent development of deep learning has been mostly focusing on Euclidean data, such as images, videos, audios, etc. However, most real-world information and relation are often expressed as graphs. To efficiently learn from graph data, graph convolutional networks (GCNs) emerge as a promising approach, showing advantages in several practical applications such as social network analysis, knowledge discovery, 3D modeling, motion capturing, etc. Real-world graphs are usually extremely large and imbalanced, posting significant performance demand and design challenges on the hardware dedicated for GCN inference. In this paper, we propose an architecture design called UW-GCN to accelerate graph convolutional network inference. To tackle the major performance bottleneck from workload imbalance, we propose dynamic neighborhood stealing and remote chunk shuffling techniques, relying on hardware flexibility to achieve hardware auto-tuning under negligible area or delay overhead. Specifically, UW-GCN is able to smartly profile the sparse graph pattern while continuously adjusting the workload distribution via routing reconfiguration among parallel processing elements (PEs). The ideal configuration is then reused in the remaining iterations. To the best of our knowledge, this is the first accelerator design particularly for GCN and the first work relying on hardware auto-tuning, which is normally based on software, to achieve near-optimal workload balance in processing sparse structures.

Geng, Tong↗