Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “random graphs”

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 91 records · Page 5
All sourcesNASA NTRSDOE OSTI
DOE OSTI2020

Polaritons and excitons: Hamiltonian design for enhanced coherence

The primary questions motivating this report are: Are there ways to increase coherence and delocalization of excitation among many molecules at moderate electronic coupling strength? Coherent delocalization of excitation in disordered molecular systems is studied using numerical calculations. The results are relevant to molecular excitons, polaritons, and make connections to classical phase oscillator synchronization. In particular, it is hypothesized that it is not only the magnitude of electronic coupling relative to the standard deviation of energetic disorder that decides the limits of coherence, but that the structure of the Hamiltonian—connections between sites (or molecules) made by electronic coupling—is a significant design parameter. Inspired by synchronization phenomena in analogous systems of phase oscillators, some properties of graphs that define the structure of different Hamiltonian matrices are explored. The report focuses on eigenvalues and ensemble density matrices of various structured, random matrices. Some reasons for the special delocalization properties and robustness of polaritons in the single-excitation subspace (the star graph) are discussed. The key result of this report is that, for some classes of Hamiltonian matrix structure, coherent delocalization is not easily defeated by energy disorder, even when the electronic coupling is small compared to disorder.

Science & Technology - Other Topics↗
DOE OSTI2023

Machine Learning Prediction of the Experimental Transition Temperature of Fe(II) Spin-Crossover Complexes

Spin-crossover (SCO) complexes are materials that exhibit changes in the spin state in response to external stimuli, with potential applications in molecular electronics. It is challenging to know a priori how to design ligands to achieve the delicate balance of entropic and enthalpic contributions needed to tailor a transition temperature close to room temperature. Here, we leverage the SCO complexes from the previously curated SCO-95 data set [Vennelakanti et al. J. Chem. Phys. 159, 024120 (2023)] to train three machine learning (ML) models for transition temperature (T 1/2 ) prediction using graph-based revised autocorrelations as features. We perform feature selection using random forest-ranked recursive feature addition (RF-RFA) to identify the features essential to model transferability. Of the ML models considered, the full feature set RF and recursive feature addition RF models perform best, achieving moderate correlation to experimental T 1/2 values. We then compare ML T 1/2 predictions to those from three previously identified best-performing density functional approximations (DFAs) which accurately predict SCO behavior across SCO-95, finding that the ML models predict T 1/2 more accurately than the best-performing DFAs. In addition, we study ML model predictions for a set of 18 SCO complexes for which only estimated T 1/2 values are available. Upon excluding outliers from this set, the RF-RFA RF model shows a strong correlation to estimated T 1/2 values with a Pearson’s r of 0.82. In contrast, DFA-predicted T 1/2 values have large errors and show no correlation to estimated T 1/2 values over the same set of complexes. Overall, our study demonstrates slightly superior performance of ML models in comparison with some of the best-performing DFAs, and we expect ML models to improve further as larger data sets of SCO complexes are curated and become available for model training.

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

Graph Sparsification by Approximate matrix Multiplication

Graphs arising in statistical problems, signal processing, large networks, combinatorial optimization, and data analysis are often dense, which causes both computational and storage bottlenecks. One way of sparsifying a weighted graph, while sharing the same vertices as the original graph but reducing the number of edges, is through spectral sparsification. We study this problem through the perspective of RandNLA. Specifically, we utilize randomized matrix multiplication to give a clean and simple analysis of how sampling according to edge weights gives a spectral approximation to graph Laplacians, without requiring spectral information. Through the CR–MM algorithm, we attain a simple and computationally efficient sparsifier whose resulting Laplacian estimate is unbiased and of minimum variance. Here, we define a new notion of additive spectral sparsifiers, which has not been considered in the literature.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Contributions of vegetation heterogeneity within tower footprint to CO 2 flux estimations through graph neural network modeling

Net ecosystem exchange of CO 2 (Fc) measured directly by eddy covariance towers is based on various assumptions, including large, flat and homogenous land cover type. In reality, often a tower site is not large enough for flux measurements, and landscapes consist of patches of different land cover types within the flux footprint. In addition, some portions of fluxes are contributed by different cover types when a footprint exceeds the size of the target ecosystem. The contributions of non-dominant patches to Fc are often ignored. Here, in this study, we propose a novel integrated modeling framework that combines random forest (RF) and XGBoost with a residual correction module based on a deep graph convolutional network (DeeperGCN) to simulate Fc for seven flux measurement sites in southwest Michigan. High-resolution remote sensing vegetation indices, soil properties, meteorological variables, and footprint-weighted spatial features were used as model inputs at three spatial resolutions (10 m, 20 m, 30 m), and their importance in predicting Fc with DeeperGCN was assessed. We found that residual correction using DeeperGCN significantly improved prediction accuracy, with the R 2 increasing from 0.9098 to 0.9479 for RF and from 0.9235 to 0.9433 for XGBoost. At site level, the maximum improvement in R 2 reached 0.1617. Paired t-tests confirmed that these improvements were statistically significant (p < 0.05). Among all predictors, leaf area index and incoming shortwave radiation emerged as the dominant drivers of spatial residual variation, followed by precipitation, relative humidity, and selected vegetation indices. The 20 m resolution yielded the best balance between model performance and computational efficiency. In conclusion, our modeling framework effectively captures both spatial heterogeneity and nonlinear interactions, offering a robust solution for spatially explicit flux modeling in structurally diverse ecosystems beyond the study sites.

footprint model↗
DOE OSTI2021

Efficient Sampling of Complex Interdependent and Multiplex Networks

Efficient sampling of interdependent and multiplex infrastructure networks is critical for effectively applying failure and recovery algorithms in real-world settings, as well as to generate property-preserving reduced-order graph-based ensembles that address topological uncertainties. In this paper, we first explore the performance, i.e. the success in preserving graph properties, of graph sampling algorithms for interdependent and multiplex networks with synthetic and real-world graphs. We simulate sampling algorithms under different parameter settings. These settings include probabilistic graph generators, coupling patterns, and various performance metrics. Our results show that while Random Node and Random Walk sampling algorithms perform best for interdependent networks, Random Edge and Forest Fire sampling algorithms perform best for multiplex networks. Second, we propose and implement a novel similarity-based sampling algorithm for multiplex networks that samples only log(N) number of layers of an N-layer multiplex network while yielding computational savings with performance guarantees. Experimental results show that similarity sampling outperforms complete sampling of all layers while decreasing performance costs from a linear scale to a logarithmic one. Our results also indicate that similarity-based sampling outperforms complete sampling and random selection in nearly all scenarios when tested with real-world data.

Subasi, Omer↗
NASA NTRSDate not supplied

Graph Representation Learning for Dengue Forecasting

In 2017, the largest recorded dengue outbreak in Sri Lanka’s history occurred. Since then, dengue has continued to threaten national health across Sri Lanka. The development of an effective Early Warning System (EWS) for dengue outbreaks is essential for Sri Lanka’s Ministry of Health to take preventative measures. We propose the use of Graph Neural Networks as EWS. Using earth observational data from NASAs global satellites and dengue incidence data from Sri Lanka s Ministry of Health, we developed a series of traditional and graph representation EWS to forecast Dengue cases across Sri Lanka’s 25 districts between 2013 and 2022. We demonstrate empirically that Graph Neural Networks which incorporate spatiotemporal relations significantly outperform traditional EWS such as Autoregressive Integrated Moving Average (ARIMA), Random Forest, and Long Short-Term Memory (LSTM). Our source code is available on GitHub and will be provided in the final submission.

Graph Neural Networks↗
DOE OSTI2022

WPGNN and PLayGen (Wind Plant Graph Neural Network and Plant Layout Generator) [SWR-21-90]

WPGNN is the graph neural network machine learning based surrogate model and software that provides a streamlined approximation of wind plant wake models. It can rapidly estimate the energy production of the plant and turbines for any arbitrarily sized wind plant and layout under any inflow condition. Associated tools include graphing and visualization capabilities as well as a wind plant generator capable of creating randomized realizations of canonical wind plant layouts. The WPGNN architecture and application is extendable to multiple wake models, turbine technologies and features, and supports downstream optimizations of wind plant layouts and control strategies. In addition to the WPGNN, we include the code for the plant layout generator (PLayGen) playgen.py. This generator can produce random realizations of realistic wind plant layouts from one of the four canonical styles: cluster, single string, multiple string, or parallel string. The PLayGen_demo.ipynb notebook provides a demonstration of how to use the generator tool.

Harrison-Atlas, Dylan↗
NASA NTRS1966

Flowgraph techniques for closed systems.

Flowgraph techniques for closed systems, discussing properties, approximation method, topology equation, frequency response, constraints, oscillatory and stochastic processes, etc

HARMONIC OSCILLATOR↗
NASA NTRS1974

A computer-controlled, on-board data acquisition system for wind-tunnel testing

A computer-controlled data acquisition system has been developed for the 40x80-foot wind tunnel at Ames Research Center. The system, consisting of several small onboard units installed in the model and a data-managing, data-displaying ground station, is capable of sampling up to 256 channels of raw data at a total sample rate of 128,000 samples/sec. Complete signal conditioning is contained within the on-board units. The sampling sequence and channel gain selection is completely random and under total control of the ground station. Outputs include a bar-graph display, digital-to-analog converters, and digital interface to the tunnel's central computer, an SEL 840MP. The system can be run stand-alone or under the control of the SEL 840MP.

Finger, H. J.↗
DOE OSTI2024

Graph Analytics on Jellyfish topology

Because large unstructured datasets is important for many science domains, distributed graph analytics is critical to many scientists. Unfortunately, obtaining scaling and performance for irregular communication is challenging because contemporary network interconnects are primarily designed to maximize bandwidths of fixed-neighborhoods large-message exchanges (e.g., stencils). Although there is no consensus on the “best” network topologies for irregular communication, unstructured graph-based interconnects can be more suitable. We analyze three popular graph workloads – clustering, pattern enumeration, and traversal — on comparable networks (in terms of resources and costs) constructed from Jellyfish Random Regular, Dragonfly and Fat tree topologies, varying the routing algorithms. Using packet-level simulations, we demonstrate up to 60% improvement in communication time with Jellyfish due to diversity of the short paths between arbitrary endpoints, which can reduce overall network stalls and congestion.

Graph Analytics, network topology, interconnect, H↗
DOE OSTI2023

Short-Depth QAOA circuits and Quantum Annealing on Higher-Order Ising Models (Rev.2)

The Quantum Alternating Operator Ansatz (QAOA) and Quantum Annealing (QA) are quantum algorithms that are both based on the adiabatic theorem and both have the goal of sampling the optimal solution(s) of combinatorial optimization problems. Quantum annealing has been physically instantiated on D-Wave devices using superconducting flux qubits, and QAOA can be programmed on digital gate-model quantum computers such as the programmable superconducting transmon qubits devices of the IBMQ series, for instance ibm washington. QAOA and QA address the same types of problems, but it is unclear how they will scale to large problem sizes and to larger and higher-fidelity quantum computers. In this article, we present a direct comparison between QAOA, one and two rounds, run on all 127 qubits of ibm washington and QA run on D-Wave Advantage system4.1 and Advantage system6.1. The problems which allow for this comparison are random Ising model problems whose connectivity matches the heavy hexagonal lattice topology of ibm washington and the Pegasus graph connectivity of the two D-Wave devices. We create two classes of problem instances for this comparison: one with higher order terms (ZZZ variable interactions), linear terms, and quadratic terms, and a separate problem type with only linear and quadratic terms. Our QAOA circuits are novel and extremely short depth, with a CNOT depth of 6 per round, which allows whole chip usage of ibm washington’s heavy hexagonal lattice and can be applied to future heavy-hex chips. We also test the effectiveness of the error suppression technique digital dynamical decoupling on the QAOA circuits. The QAOA circuits compiled to ibm washington are composed of several thousand circuit instructions, approximately 3, 000 depending on the details of the circuit, making these some the largest quantum circuits ever executed on a digital quantum processor. QAOA and QA are compared against the classical heuristic algorithm of simulated annealing and all problem instances are exactly solved using CPLEX in order to evaluate which samplers, if any, correctly found the ground state solution(s) of the problem instances. We find that (i) QA outperforms QAOA on all problem instances, (ii) QAOA samples the problems better than random sampling, and (iii) QAOA angle computation exhibits clear parameter concentration across the ensemble of Ising models.

127 Qubits↗
DOE OSTI2022

RanCompute: Computational Security in Embedded Devices via Random Input and Output Encodings

An embedded device in an insecure environment is subject to additional security risk through capture and reverse-engineering by a capable adversary. If this device contains a microchip performing sensitive computations, capture of the chip may leak functionality to an adversary. In this paper we propose a novel method in which we randomly encode the input operands and the outputs of a computation, thus not revealing the arithmetic operations being performed. The operations are sequenced in a graph representing the overall application. Once the initialization values are overwritten and lost, the results of these computations are indecipherable by the device performing the calculations as well as by any adversary. The result is transmitted back to a secure server which has stored the initialization values and so can decode the results which appear random to the adversary.

Embedded computing↗
DOE OSTI2024

MDLoader: A Hybrid Model-Driven Data Loader for Distributed Graph Neural Network Training

Scalable data management is essential for processing large scientific dataset on HPC platforms for distributed deep learning. In-memory distributed storage is preferred for its speed, enabling rapid, random, and frequent data access required by stochastic optimizers. Processes use one-sided or collective communication to fetch remote data, with optimal performance depending on (i) dataset characteristics, (ii) training scale, and (iii) interconnection network. Empirical analysis shows collective communication excels with larger mini-batch sizes and/or fewer processes, whereas one-sided communication outperforms at larger scales. We propose MDLoader, a hybrid in-memory data loader for distributed graph neural network training. MDLoader features a model-driven performance estimator that dynamically selects between one-sided and collective communication at the beginning of training using Tree of Parzen Estimators (TPE). Evaluations on NERSC Perlmutter and OLCF Summit show MDLoader outperforms single-backend loaders by up to 2.83 × and predicts the suitable communication method with 96.3% (Perlmutter) and 94.3% (Summit) success rate.

Bae, Jonghyun↗
DOE OSTI2024

4-Clique network minor embedding for quantum annealers

Quantum annealing is a quantum algorithm for computing solutions to combinatorial optimization problems. This study proposes a method for minor embedding optimization problems onto sparse quantum annealing hardware graphs called 4-clique network minor embedding. This method is in contrast to the standard minor embedding technique of using a path of linearly connected qubits in order to represent a logical variable state. The 4-clique minor embedding is possible on Pegasus graph connectivity, which is the native hardware graph for some of the current D-Wave quantum annealers. The Pegasus hardware graph contains many cliques of size 4, making it possible to form a graph composed entirely of paths of connected 4-cliques on which a problem can be minor-embedded. The 4-clique chains come at the cost of additional qubit usage on the hardware graph, but they allow for stronger coupling within each chain, thereby increasing chain integrity, reducing chain breaks, and allow for greater usage of the available energy scale for programming logical problem coefficients on current quantum annealers. The 4-clique minor embedding technique is compared with the standard linear path minor embedding with experiments on two D-Wave quantum annealing processors with Pegasus hardware graphs. We show proof-of-concept experiments where the 4-clique minor embeddings can use weak chain strengths while successfully carrying out the computation of minimizing random all-to-all spin glass problem instances. Published by the American Physical Society 2024

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2023

DDStore: Distributed Data Store for Scalable Training of Graph Neural Networks on Large Atomistic Modeling Datasets

Graph neural networks (GNNs) are a class of Deep Learning models used in designing atomistic materials for effective screening of large chemical spaces. To ensure robust prediction, GNN models must be trained on large volumes of atomistic data on leadership class supercomputers. Even with the advent of modern architectures that consist of multiple storage layers that include node-local NVMe devices in addition to device memory for caching large datasets, extreme-scale model training faces I/O challenges at scale.We present DDStore, an in-memory distributed data store designed for GNN training on large-scale graph data. DDStore provides a hierarchical, distributed, data caching technique that combines data chunking, replication, low-latency random access, and high throughput communication. DDStore achieves near-linear scaling for training a GNN model using up to 1000 GPUs on the Summit and Perlmutter supercomputers, and reaches up to a 6.15x reduction in GNN training time compared to state-of-the-art methodologies.

Choi, Jong Youl↗
NASA NTRS2021

Identifying Molecules as Biosignatures with Assembly Theory and Mass Spectrometry

The search for evidence of life elsewhere in the universe is hard because it is not obvious what signatures are unique to life. Here we postulate that complex molecules found in high abundance are universal biosignatures as they cannot form by chance. To explore this, we developed the first intrinsic measure of molecular complexity that can be experimentally determined, and this is based upon a new approach called assembly theory which gives the molecular assembly number (MA) of a given molecule. MA allows us to compare the intrinsic complexity of molecules using the minimum number of steps required to construct the molecular graph starting from basic objects, and a probabilistic model shows how the probability of any given molecule forming randomly drops dramatically as its MA increases. To map chemical space, we calculated the MA of ca. 2.5 million compounds, and collected data which showed the complexity of a molecule can be experimentally determined by using three independent techniques including infra-red spectroscopy, nuclear magnetic resonance, and by fragmentation in a mass spectrometer, and this data has an excellent corelation with the values predicted from our assembly theory. We then set out to see if this approach could allow us to identify molecular biosignatures with a set of diverse samples from around the world, outer space, and the laboratory including prebiotic soups. The results show that there is a non-living to living threshold in MA complexity and the higher the MA for a given molecule, the more likely that it had to be produced by a biological process. This work demonstrates it is possible to use this approach to build a life detection instrument that could be deployed on missions to extra-terrestrial locations to detect biosignatures, map the extent of life on Earth, and be used as a molecular complexity scale to quantify the constraints needed to direct prebiotically plausible processes in the laboratory. Such an approach is vital if we are going to find new life elsewhere in the universe or create de-novo life in the lab.

Stuart M Marshall↗
DOE OSTI2022

Scalable algorithms for physics-informed neural and graph networks

Physics-informed machine learning (PIML) has emerged as a promising new approach for simulating complex physical and biological systems that are governed by complex multiscale processes for which some data are also available. In some instances, the objective is to discover part of the hidden physics from the available data, and PIML has been shown to be particularly effective for such problems for which conventional methods may fail. Unlike commercial machine learning where training of deep neural networks requires big data, in PIML big data are not available. Instead, we can train such networks from additional information obtained by employing the physical laws and evaluating them at random points in the space–time domain. Such PIML integrates multimodality and multifidelity data with mathematical models, and implements them using neural networks or graph networks. Here, we review some of the prevailing trends in embedding physics into machine learning, using physics-informed neural networks (PINNs) based primarily on feed-forward neural networks and automatic differentiation. For more complex systems or systems of systems and unstructured data, graph neural networks (GNNs) present some distinct advantages, and here we review how physics-informed learning can be accomplished with GNNs based on graph exterior calculus to construct differential operators; we refer to these architectures as physics-informed graph networks (PIGNs). We present representative examples for both forward and inverse problems and discuss what advances are needed to scale up PINNs, PIGNs and more broadly GNNs for large-scale engineering problems.

42 ENGINEERING↗