Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Edge computing”

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 343 records · Page 19

An exact and efficient first passage time algorithm for reaction–diffusion processes on a 2D-lattice

We present an exact and efficient algorithm for reaction–diffusion–nucleation processes on a 2D-lattice. The algorithm makes use of first passage time (FPT) to replace the computationally intensive simulation of diffusion hops in KMC by larger jumps when particles are far away from step-edges or other particles. Our approach computes exact probability distributions of jump times and target locations in a closed-form formula, based on the eigenvectors and eigenvalues of the corresponding 1D transition matrix, maintaining atomic-scale resolution of resulting shapes of deposit islands. We have applied our method to three different test cases of electrodeposition: pure diffusional aggregation for large ranges of diffusivity rates and for simulation domain sizes of up to 4096×4096 sites, the effect of diffusivity on island shapes and sizes in combination with a KMC edge diffusion, and the calculation of an exclusion zone in front of a step-edge, confirming statistical equivalence to standard KMC simulations. The algorithm achieves significant speedup compared to standard KMC for cases where particles diffuse over long distances before nucleating with other particles or being captured by larger islands.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

SciDAC Center for Plasma Edge Simulation

This project with a total funding of $592,998 for six years has partially supported four postdoctoral researchers at the University of California, Irvine (UCI). The UCI team has formulated electrostatic and electromagnetic global gyrokinetic particle simulation models with kinetic electrons, implemented these models in the edge code XGC1, performed benchmark between GTC and XGC1, developed computational tools for gyrokinetic particle simulation in tokamak edge geometry, and initiated preparatory study of edge turbulence using GTC code. The research results has been published in 12 papers and presented at many international and national conferences.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Scalable deep text comprehension for Cancer surveillance on high-performance computing

Background: Deep Learning (DL) has advanced the state-of-the-art capabilities in bioinformatics applications which has resulted in trends of increasingly sophisticated and computationally demanding models trained by larger and larger data sets. This vastly increased computational demand challenges the feasibility of conducting cutting-edge research. One solution is to distribute the vast computational workload across multiple computing cluster nodes with data parallelism algorithms. In this study, we used a High-Performance Computing environment and implemented the Downpour Stochastic Gradient Descent algorithm for data parallelism to train a Convolutional Neural Network (CNN) for the natural language processing task of information extraction from a massive dataset of cancer pathology reports. We evaluated the scalability improvements using data parallelism training and the Titan supercomputer at Oak Ridge Leadership Computing Facility. To evaluate scalability, we used different numbers of worker nodes and performed a set of experiments comparing the effects of different training batch sizes and optimizer functions.Results: We found that Adadelta would consistently converge at a lower validation loss, though requiring over twice as many training epochs as the fastest converging optimizer, RMSProp. The Adam optimizer consistently achieved a close 2nd place minimum validation loss significantly faster; using a batch size of 16 and 32 allowed the network to converge in only 4.5 training epochs.Conclusions: We demonstrated that the networked training process is scalable across multiple compute nodes communicating with message passing interface while achieving higher classification accuracy compared to a traditional machine learning algorithm.

Qiu, John X.↗

Combined experimental and theoretical assessments of the lattice dynamics and optoelectronics of TaON and Ta{sub 3}N{sub 5}

Presented herein is a detailed discussion of the properties of the lattice dynamic and optoelectronic properties of tantalum(V) oxynitride (TaON) and tantalum(V) nitride (Ta{sub 3}N{sub 5}), from experimental and theoretical standpoint. The active Raman and infra red (IR) frequencies of TaON and Ta{sub 3}N{sub 5} were measured using confocal Raman and Fourier Transform Infrared spectroscopies (FTIR) and calculated using the linear response method within the density functional perturbation theory (DFPT). The detailed study leads to an exhaustive description of the spectra, including the symmetry of the vibrational modes. Electronic structures of these materials were computed using DFT within the range-separated hybrid HSE06 exchange–correlation formalism. Electronic and ionic contributions to the dielectric constant tensors of these materials were obtained from DFPT within the linear response method using the PBE functional. Furthermore, effective mass of photogenerated holes and electrons at the band edges of these compounds were computed from the electronic band structure obtained at the DFT/HSE06 level of theory. The results suggest that anisotropic nature in TaON and Ta{sub 3}N{sub 5} is present in terms of dielectric constant and effective masses. - Graphical abstract: Detailed investigation has been conducted from combined experimental and theoretical approaches on Raman and IR spectroscopies, electronic structures, dielectric constants and effective masses of TaON and Ta{sub 3}N{sub 5}. - Highlights: • Crystal structures of TaON and Ta{sub 3}N{sub 5} are discussed based on XRD and DFT calculation. • Raman and IR spectra of TaON and Ta{sub 3}N{sub 5} materials are measured and computed by DFPT method. • Optoelectronic properties of TaON and Ta{sub 3}N{sub 5} are discussed. • Dielectric constant and effective masses of TaON and Ta{sub 3}N{sub 5} are calculated.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Bounds on edge shear layer persistence while approaching the density limit

This paper details the theory of edge shear layer collapse as the density approaches the Greenwald density limit. It significantly extends earlier work, which was restricted in applicability. The zonal shear flow screening length is calculated for banana, plateau and Pfirsch–Schluter regimes. Poloidal field scaling persists in the plateau regime. Neoclassical screening and drift wave–zonal flow dynamics are combined in a theory, which is then reduced to a predator–prey model. Zonal noise, due to incoherent mode coupling, is retained. The threshold condition for edge shear layer collapse is computed, and linked to a critical value of the dimensionless parameter ${\rho }_{\mathrm{s}}/\sqrt{{\rho }_{\mathrm{s}\mathrm{c}}{L}_{n}}$. Here ${\rho }_{s}$ is the ion sound radius, ${\rho }_{sc}$ is the zonal flow screening length and ${L}_{n}$ is the equilibrium density scale length. The limiting initial edge density for shear layer collapse is derived and shown to scale favorably with the plasma current. Here, the results are discussed in light of the density limit and Ohmic phenomenology.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Ru L[subscript 2,3] XANES theoretical simulation with DFT: A test of the core-hole treatment

Density functional theory (DFT)-based relativistic calculations were performed to model the Ru L-edge X-ray absorption near edge structure (XANES) spectra of the hexaammineruthenium complex [Ru(NH{sub 3}){sub 6}]{sup 3+} and 'blue dimer' water oxidation catalyst, cis,cis- [(bpy){sub 2}(H{sub 2}O)Ru{sup III}ORu{sup III}(OH{sub 2})(bpy){sub 2}]{sup 4+} (bpy is 2,2-bipyridine). Two computational approaches were compared: simulations without the core-hole and by modeling of the core-hole within the Z+1 approximation. Good agreement between calculated and experimental XANES spectra is achieved without including the core-hole. Simulations with algorithms beyond the Z+1 approximation were only possible in a framework of the scalar relativistic treatment. Time-dependent DFT (TD-DFT) was used to compute the Ru L-edge spectrum for [Ru(NH{sub 3}){sub 6}]{sup 3+} model compound. Three different core-hole treatments were compared in a real-space full multiple scattering XANES modeling within the Green function formalism (implemented in the FEFF9.5 package) for the [Ru(Mebimpy)(bpm)(H{sub 2}O)]{sup 2+} complex. The latter approaches worked well in cases where spin-orbit treatment of relativistic effects is not required.

99 GENERAL AND MISCELLANEOUS↗

Transverse momentum dependent PDFs at N3LO

We compute the quark and gluon transverse momentum dependent parton distribution functions at next-to-next-to-next-to-leading order (N 3 LO) in perturbative QCD. Our calculation is based on an expansion of the differential Drell-Yan and gluon fusion Higgs production cross sections about their collinear limit. This method allows us to employ cutting edge multiloop techniques for the computation of cross sections to extract these universal building blocks of the collinear limit of QCD. The corresponding perturbative matching kernels for all channels are expressed in terms of simple harmonic polylogarithms up to weight five. As a byproduct, we confirm a previous computation of the soft function for transverse momentum factorization at N 3 LO. Our results are the last missing ingredient to extend the q T subtraction methods to N 3 LO and to obtain resummed q T spectra at N 3 LL' accuracy both for gluon as well as for quark initiated processes.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Discontinuity detection in multivariate space for stochastic simulations

Edge detection has traditionally been associated with detecting physical space jump discontinuities in one dimension, e.g. seismic signals, and two dimensions, e.g. digital images. Hence most of the research on edge detection algorithms is restricted to these contexts. High dimension edge detection can be of significant importance, however. For instance, stochastic variants of classical differential equations not only have variables in space/time dimensions, but additional dimensions are often introduced to the problem by the nature of the random inputs. The stochastic solutions to such problems sometimes contain discontinuities in the corresponding random space and a prior knowledge of jump locations can be very helpful in increasing the accuracy of the final solution. Traditional edge detection methods typically require uniform grid point distribution. They also often involve the computation of gradients and/or Laplacians, which can become very complicated to compute as the number of dimensions increases. The polynomial annihilation edge detection method, on the other hand, is more flexible in terms of its geometric specifications and is furthermore relatively easy to apply. This paper discusses the numerical implementation of the polynomial annihilation edge detection method to high dimensional functions that arise when solving stochastic partial differential equations.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Detecting anomalous behavior via user authentication graphs

Significant and aggregate user authentication activity may be analyzed across a population of users and computers in one or more networks to differentiate between authorized users and intruders in a network, and/or to detect inappropriate behavior by otherwise authorized users. Dynamic graphs and graph models over user and computer authentication activity, including time-constrained models, may be used for the purposes of profiling and analyzing user behavior in computer networks. More specifically, an edge-based breadth first search of graphs may be used that enforces time-constraints while maintaining traditional breadth first search computational complexity equivalence.

97 MATHEMATICS AND COMPUTING↗

Detecting anomalous behavior via user authentication graphs

Significant and aggregate user authentication activity may be analyzed across a population of users and computers in one or more networks to differentiate between authorized users and intruders in a network, and/or to detect inappropriate behavior by otherwise authorized users. Dynamic graphs and graph models over user and computer authentication activity, including time-constrained models, may be used for the purposes of profiling and analyzing user behavior in computer networks. More specifically, an edge-based breadth first search of graphs may be used that enforces time-constraints while maintaining traditional breadth first search computational complexity equivalence.

97 MATHEMATICS AND COMPUTING↗

Automated divertor target design by adjoint shape sensitivity analysis and a one-shot method

As magnetic confinement fusion progresses towards the development of first reactor-scale devices, computational tokamak divertor design is a topic of high priority. Presently, edge plasma codes are used in a forward approach, where magnetic field and divertor geometry are manually adjusted to meet design requirements. Due to the complex edge plasma flows and large number of design variables, this method is computationally very demanding. On the other hand, efficient optimization-based design strategies have been developed in computational aerodynamics and fluid mechanics. Such an optimization approach to divertor target shape design is elaborated in the present paper. A general formulation of the design problems is given, and conditions characterizing the optimal designs are formulated. Using a continuous adjoint framework, design sensitivities can be computed at a cost of only two edge plasma simulations, independent of the number of design variables. Furthermore, by using a one-shot method the entire optimization problem can be solved at an equivalent cost of only a few forward simulations. The methodology is applied to target shape design for uniform power load, in simplified edge plasma geometry.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Foundational Tools for Petascale Computing

The Paradyn project has a history of developing algorithms, techniques, and software that push the cutting edge of tool technology for high-end computing systems. Under this funding, we are working on a three-year agenda to make substantial new advances in support of new and emerging Petascale systems. The overall goal for this work is to address the steady increase in complexity of these petascale systems. Our work covers two key areas: (1) The analysis, instrumentation and control of binary programs. Work in this area falls under the general framework of the Dyninst API tool kits. (2) Infrastructure for building tools and applications at extreme scale. Work in this area falls under the general framework of the MRNet scalability framework. Note that work done under this funding is closely related to work done under a contemporaneous grant, “High-Performance Energy Applications and Systems”, SC0004061/FG02-10ER25972, UW PRJ36WV.

97 MATHEMATICS AND COMPUTING↗

SU-E-J-248: Contributions of Tumor and Stroma Phenotyping in Computer-Aided Diagnosis

Purpose: To gain insight into the role of parenchyma stroma in the characterization of breast tumors by incorporating computerized mammographic parenchyma assessment into breast CADx in the task of distinguishing between malignant and benign lesions. Methods: This study was performed on 182 biopsy-proven breast mass lesions, including 76 benign and 106 malignant lesions. For each full-field digital mammogram (FFDM) case, our quantitative imaging analysis was performed on both the tumor and a region-of-interest (ROI) from the normal contralateral breast. The lesion characterization includes automatic lesion segmentation and feature extraction. Radiographic texture analysis (RTA) was applied on the normal ROIs to assess the mammographic parenchymal patterns of these contralateral normal breasts. Classification performance of both individual computer extracted features and the output from a Bayesian artificial neural network (BANN) were evaluated with a leave-one-lesion-out method using receiver operating characteristic (ROC) analysis with area under the curve (AUC) as the figure of merit. Results: Lesion characterization included computer-extracted phenotypes of spiculation, size, shape, and margin. For parenchymal pattern characterization, five texture features were selected, including power law beta, contrast, and edge gradient. Merging of these computer-selected features using BANN classifiers yielded AUC values of 0.79 (SE=0.03) and 0.67 (SE=0.04) in the task of distinguishing between malignant and benign lesions using only tumor phenotypes and texture features from the contralateral breasts, respectively. Incorporation of tumor phenotypes with parenchyma texture features into the BANN yielded improved classification performance with an AUC value of 0.83 (SE=0.03) in the task of differentiating malignant from benign lesions. Conclusion: Combining computerized tumor and parenchyma phenotyping was found to significantly improve breast cancer diagnostic accuracy highlighting the need to consider both tumor and stroma in decision making. Funding: University of Chicago Dean Bridge Fund, NCI U24-CA143848-05, P50-CA58223 Breast SPORE program, and Breast Cancer Research Foundation. COI: MLG is a stockholder in R2 technology/Hologic and receives royalties from Hologic, GE Medical Systems, MEDIAN Technologies, Riverain Medical, Mitsubishi, and Toshiba. MLG is a cofounder and stockholder in Quantitative Insights.

60 APPLIED LIFE SCIENCES↗

Distilling particle knowledge for fast reconstruction at high-energy physics experiments

Knowledge distillation is a form of model compression that allows artificial neural networks of different sizes to learn from one another. Its main application is the compactification of large deep neural networks to free up computational resources, in particular on edge devices. In this article, we consider proton-proton collisions at the High-Luminosity Large Hadron Collider (HL-LHC) and demonstrate a successful knowledge transfer from an event-level graph neural network (GNN) to a particle-level small deep neural network (DNN). Our algorithm, DistillNet, is a DNN that is trained to learn about the provenance of particles, as provided by the soft labels that are the GNN outputs, to predict whether or not a particle originates from the primary interaction vertex. The results indicate that for this problem, which is one of the main challenges at the HL-LHC, there is minimal loss during the transfer of knowledge to the small student network, while improving significantly the computational resource needs compared to the teacher. This is demonstrated for the distilled student network on a CPU, as well as for a quantized and pruned student network deployed on a field programmable gate array. Our study proves that knowledge transfer between networks of different complexity can be used for fast artificial intelligence (AI) in high-energy physics that improves the expressiveness of observables over non-AI-based reconstruction algorithms. Such an approach can become essential at the HL-LHC experiments, e.g. to comply with the resource budget of their trigger stages.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Seismic Waveform Inversion Capability on Resource-Constrained Edge Devices

Seismic full wave inversion (FWI) is a widely used non-linear seismic imaging method used to reconstruct subsurface velocity images, however it is time consuming, has high computational cost and depend heavily on human interaction. Recently, deep learning has accelerated it’s use in several data-driven techniques, however most deep learning techniques suffer from overfitting and stability issues. In this work, we propose an edge computing-based data-driven inversion technique based on supervised deep convolutional neural network to accurately reconstruct the subsurface velocities. Deep learning based data-driven technique depends mostly on bulk data training. In this work, we train our deep convolutional neural network (DCN) (UNet and InversionNet) on the raw seismic data and their corresponding velocity models during the training phase to learn the non-linear mapping between the seismic data and velocity models. The trained network is then used to estimate the velocity models from new input seismic data during the prediction phase. The prediction phase is performed on a resource-constrained edge device such as Raspberry Pi. Raspberry Pi provides real-time and on-device computational power to execute the inference process. In addition, we demonstrate robustness of our models to perform inversion in the presence on noise by performing both noise-aware and no-noise training and feeding the resulting trained models with noise at different signal-to-noise (SNR) ratio values. We make great efforts to achieve very feasible inference times on the Raspberry Pi for both models. Specifically, the inference times per prediction for UNet and InversionNet models on Raspberry Pi were 22 and 4 s respectively whilst inference times for both models on the GPU were 2 and 18 s which are very comparable. Finally, we have designed a user-friendly interactive graphical user interface (GUI) to automate the model execution and inversion process on the Raspberry Pi.

Manu, Daniel (ORCID:0000000154982677)↗

Genomics for Everyone

Genomics — the genetic mapping and DNA sequencing of sets of genes or the complete genomes of organisms, along with related genome analysis and database work — is emerging as one of the transformative sciences of the 21st century. But current bioinformatics tools are not accessible to most biological researchers. Now, a new computational and web-based tool called EDGE Bioinformatics is working to fulfill the promise of democratizing genomics.

59 BASIC BIOLOGICAL SCIENCES↗

FPGA Computing

The articles in this special section focus on cutting-edge research on topics that relate to field programmable gate arrays (FPGA) in computing.

97 MATHEMATICS AND COMPUTING↗