Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “level set methods”

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 1,045 records · Page 58

Removal of Historic Low-Level Radioactive Sediment from the Port Hope Harbour - 13314

At the Port Hope Harbour, located on the north shore of Lake Ontario, the presence of low-level radioactive sediment, resulting from a former radium and uranium refinery that operated alongside the Harbour, currently limits redevelopment and revitalization opportunities. These waste materials contain radium-226, uranium, arsenic and other contaminants. Several other on-land locations within the community of Port Hope are also affected by the low-level radioactive waste management practices of the past. The Port Hope Project is a community initiated undertaking that will result in the consolidation of an estimated 1.2 million cubic metres of the low-level radioactive waste from the various sites in Port Hope into a new engineered above ground long-term waste management facility. The remediation of the estimated 120,000 m{sup 3} of contaminated sediments from the Port Hope Harbour is one of the more challenging components of the Port Hope Project. Following a thorough review of various options, the proposed method of contaminated sediment removal is by dredging. The sediment from the dredge will then be pumped as a sediment-water slurry mixture into geo-synthetic containment tubes for dewatering. Due to the hard substrate below the contaminated sediment, the challenge has been to set performance standards in terms of low residual surface concentrations that are attainable in an operationally efficient manner. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Use of a line-pair resolution phantom for comprehensive quality assurance of electronic portal imaging devices based on fundamental imaging metrics

Image guided radiation therapy solutions based on megavoltage computed tomography (MVCT) involve the extension of electronic portal imaging devices (EPIDs) from their traditional role of weekly localization imaging and planar dose mapping to volumetric imaging for 3D setup and dose verification. To sustain the potential advantages of MVCT, EPIDs are required to provide improved levels of portal image quality. Therefore, it is vital that the performance of EPIDs in clinical use is maintained at an optimal level through regular and rigorous quality assurance (QA). Traditionally, portal imaging QA has been carried out by imaging calibrated line-pair and contrast resolution phantoms and obtaining arbitrarily defined QA indices that are usually dependent on imaging conditions and merely indicate relative trends in imaging performance. They are not adequately sensitive to all aspects of image quality unlike fundamental imaging metrics such as the modulation transfer function (MTF), noise power spectrum (NPS), and detective quantum efficiency (DQE) that are widely used to characterize detector performance in radiographic imaging and would be ideal for QA purposes. However, due to the difficulty of performing conventional MTF measurements, they have not been used for routine clinical QA. The authors present a simple and quick QA methodology based on obtaining the MTF, NPS, and DQE of a megavoltage imager by imaging standard open fields and a bar-pattern QA phantom containing 2 mm thick tungsten line-pair bar resolution targets. Our bar-pattern based MTF measurement features a novel zero-frequency normalization scheme that eliminates normalization errors typically associated with traditional bar-pattern measurements at megavoltage x-ray energies. The bar-pattern QA phantom and open-field images are used in conjunction with an automated image analysis algorithm that quickly computes the MTF, NPS, and DQE of an EPID system. Our approach combines the fundamental advantages of linear systems metrics such as robustness, sensitivity across the full spatial frequency range of interest, and normalization to imaging conditions (magnification, system gain settings, and exposure), with the simplicity, ease, and speed of traditional phantom imaging. The algorithm was analyzed for accuracy and sensitivity by comparing with a commercial portal imaging QA method (PIPSPRO, Standard Imaging, Middleton, WI) on both first-generation lens-coupled and modern a-Si flat-panel based clinical EPID systems. The bar-pattern based QA measurements were found to be far more sensitive to even small levels of degradation in spatial resolution and noise. The bar-pattern based QA methodology offers a comprehensive image quality assessment tool suitable for both commissioning and routine EPID QA.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

A Novel Coarsening Method for Scalable and Efficient Mesh Generation

In this paper, we propose a novel mesh coarsening method called brick coarsening method. The proposed method can be used in conjunction with any graph partitioners and scales to very large meshes. This method reduces problem space by decomposing the original mesh into fixed-size blocks of nodes called bricks, layered in a similar way to conventional brick laying, and then assigning each node of the original mesh to appropriate brick. Our experiments indicate that the proposed method scales to very large meshes while allowing simple RCB partitioner to produce higher-quality partitions with significantly less edge cuts. Our results further indicate that the proposed brick-coarsening method allows more complicated partitioners like PT-Scotch to scale to very large problem size while still maintaining good partitioning performance with relatively good edge-cut metric. Graph partitioning is an important problem that has many scientific and engineering applications in such areas as VLSI design, scientific computing, and resource management. Given a graph G = (V,E), where V is the set of vertices and E is the set of edges, (k-way) graph partitioning problem is to partition the vertices of the graph (V) into k disjoint groups such that each group contains roughly equal number of vertices and the number of edges connecting vertices in different groups is minimized. Graph partitioning plays a key role in large scientific computing, especially in mesh-based computations, as it is used as a tool to minimize the volume of communication and to ensure well-balanced load across computing nodes. The impact of graph partitioning on the reduction of communication can be easily seen, for example, in different iterative methods to solve a sparse system of linear equation. Here, a graph partitioning technique is applied to the matrix, which is basically a graph in which each edge is a non-zero entry in the matrix, to allocate groups of vertices to processors in such a way that many of matrix-vector multiplication can be performed locally on each processor and hence to minimize communication. Furthermore, a good graph partitioning scheme ensures the equal amount of computation performed on each processor. Graph partitioning is a well known NP-complete problem, and thus the most commonly used graph partitioning algorithms employ some forms of heuristics. These algorithms vary in terms of their complexity, partition generation time, and the quality of partitions, and they tend to trade off these factors. A significant challenge we are currently facing at the Lawrence Livermore National Laboratory is how to partition very large meshes on massive-size distributed memory machines like IBM BlueGene/P, where scalability becomes a big issue. For example, we have found that the ParMetis, a very popular graph partitioning tool, can only scale to 16K processors. An ideal graph partitioning method on such an environment should be fast and scale to very large meshes, while producing high quality partitions. This is an extremely challenging task, as to scale to that level, the partitioning algorithm should be simple and be able to produce partitions that minimize inter-processor communications and balance the load imposed on the processors. Our goals in this work are two-fold: (1) To develop a new scalable graph partitioning method with good load balancing and communication reduction capability. (2) To study the performance of the proposed partitioning method on very large parallel machines using actual data sets and compare the performance to that of existing methods. The proposed method achieves the desired scalability by reducing the mesh size. For this, it coarsens an input mesh into a smaller size mesh by coalescing the vertices and edges of the original mesh into a set of mega-vertices and mega-edges. A new coarsening method called brick algorithm is developed in this research. In the brick algorithm, the zones in a given mesh are first grouped into fixed size blocks called bricks. These brick are then laid in a way similar to conventional brick laying technique, which reduces the number of neighboring blocks each block needs to communicate. Contributions of this research are as follows: (1) We have developed a novel method that scales to a really large problem size while producing high quality mesh partitions; (2) We measured the performance and scalability of the proposed method on a machine of massive size using a set of actual large complex data sets, where we have scaled to a mesh with 110 million zones using our method. To the best of our knowledge, this is the largest complex mesh that a partitioning method is successfully applied to; and (3) We have shown that proposed method can reduce the number of edge cuts by as much as 65%.

97 MATHEMATICS AND COMPUTING↗

International Spine Radiosurgery Consortium Consensus Guidelines for Target Volume Definition in Spinal Stereotactic Radiosurgery

Purpose: Spinal stereotactic radiosurgery (SRS) is increasingly used to manage spinal metastases. However, target volume definition varies considerably and no consensus target volume guidelines exist. This study proposes consensus target volume definitions using common scenarios in metastatic spine radiosurgery. Methods and Materials: Seven radiation oncologists and 3 neurological surgeons with spinal radiosurgery expertise independently contoured target and critical normal structures for 10 cases representing common scenarios in metastatic spine radiosurgery. Each set of volumes was imported into the Computational Environment for Radiotherapy Research. Quantitative analysis was performed using an expectation maximization algorithm for Simultaneous Truth and Performance Level Estimation (STAPLE) with kappa statistics calculating agreement between physicians. Optimized confidence level consensus contours were identified using histogram agreement analysis and characterized to create target volume definition guidelines. Results: Mean STAPLE agreement sensitivity and specificity was 0.76 (range, 0.67-0.84) and 0.97 (range, 0.94-0.99), respectively, for gross tumor volume (GTV) and 0.79 (range, 0.66-0.91) and 0.96 (range, 0.92-0.98), respectively, for clinical target volume (CTV). Mean kappa agreement was 0.65 (range, 0.54-0.79) for GTV and 0.64 (range, 0.54-0.82) for CTV (P<.01 for GTV and CTV in all cases). STAPLE histogram agreement analysis identified optimal consensus contours (80% confidence limit). Consensus recommendations include that the CTV should include abnormal marrow signal suspicious for microscopic invasion and an adjacent normal bony expansion to account for subclinical tumor spread in the marrow space. No epidural CTV expansion is recommended without epidural disease, and circumferential CTVs encircling the cord should be used only when the vertebral body, bilateral pedicles/lamina, and spinous process are all involved or there is extensive metastatic disease along the circumference of the epidural space. Conclusions: This report provides consensus guidelines for target volume definition for spinal metastases receiving upfront SRS in common clinical situations.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

High-Grade Glioma Radiation Therapy Target Volumes and Patterns of Failure Obtained From Magnetic Resonance Imaging and {sup 18}F-FDOPA Positron Emission Tomography Delineations From Multiple Observers

Purpose: The objective of this study was to compare recurrent tumor locations after radiation therapy with pretreatment delineations of high-grade gliomas from magnetic resonance imaging (MRI) and 3,4-dihydroxy-6-[{sup 18}F]fluoro-L-phenylalanine ({sup 18}F-FDOPA) positron emission tomography (PET) using contours delineated by multiple observers. Methods and Materials: Nineteen patients with newly diagnosed high-grade gliomas underwent computed tomography (CT), gadolinium contrast-enhanced MRI, and {sup 18}F-FDOPA PET/CT. The image sets (CT, MRI, and PET/CT) were registered, and 5 observers contoured gross tumor volumes (GTVs) using MRI and PET. Consensus contours were obtained by simultaneous truth and performance level estimation (STAPLE). Interobserver variability was quantified by the percentage of volume overlap. Recurrent tumor locations after radiation therapy were contoured by each observer using CT or MRI. Consensus recurrence contours were obtained with STAPLE. Results: The mean interobserver volume overlap for PET GTVs (42% ± 22%) and MRI GTVs (41% ± 22%) was not significantly different (P=.67). The mean consensus volume was significantly larger for PET GTVs (58.6 ± 52.4 cm{sup 3}) than for MRI GTVs (30.8 ± 26.0 cm{sup 3}, P=.003). More than 95% of the consensus recurrence volume was within the 95% isodose surface for 11 of 12 (92%) cases with recurrent tumor imaging. Ten (91%) of these cases extended beyond the PET GTV, and 9 (82%) were contained within a 2-cm margin on the MRI GTV. One recurrence (8%) was located outside the 95% isodose surface. Conclusions: High-grade glioma contours obtained with {sup 18}F-FDOPA PET had similar interobserver agreement to volumes obtained with MRI. Although PET-based consensus target volumes were larger than MRI-based volumes, treatment planning using PET-based volumes may not have yielded better treatment outcomes, given that all but 1 recurrence extended beyond the PET GTV and most were contained by a 2-cm margin on the MRI GTV.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Helical Tomotherapy and Larynx Sparing in Advanced Oropharyngeal Carcinoma: A Dosimetric Study

Intensity-modulated radiation therapy (IMRT) is gaining acceptance as a standard treatment technique for advanced squamous cell carcinoma (SCC) of the oropharynx. Dose to the uninvolved larynx and surrounding structures can pose a problem in patients with significant neck disease, potentially compromising laryngeal function and quality of life. Tomotherapy may allow greater laryngeal sparing. Seven patients with stage IV SCC of the oropharynx were replanned using Tomotherapy version 3.1. All contours/planning target volumes (PTVs) from the original plans were preserved, with the exception of the larynx, which was drawn to include all soft tissue encompassed by the thyroid/cricoid cartilage. A simultaneous integrated boost technique was used with PTV 1, 2, and 3 receiving 69.96, 59.40, and 54.00 Gy, respectively in 33 fractions. Dosimetry was evaluated via the Pinnacle treatment planning system (TPS). Equivalent uniform dose (EUD) was calculated from the dose volume histogram (DVH) using the general method with 'a' = 5.0. Mean larynx dose for all patients was 24.4 Gy. Mean EUD to the larynx was 34.2 Gy. Homogeneity was adequate; average maximum dose was 109.7% of the highest prescription. All other organs at risk (OAR) were adequately spared. Tomotherapy can spare the uninvolved larynx in the setting of advanced SCC of the oropharynx to levels that are similar to or better than those reported with other techniques. Sparing is achieved without compromising target coverage or other OAR sparing. The clinical benefit of this sparing remains to be determined in a prospective study.

61 RADIATION PROTECTION AND DOSIMETRY↗

WE-FG-207B-07: Feasibility of Low Dose Lung Cancer Screening with a Whole-Body Photon Counting CT: First Human Results

Purpose: To evaluate the feasibility of using a whole-body photon counting detector (PCD) CT scanner for low dose lung cancer screening compared to a conventional energy integrating detector (EID) system. Methods: Radiation dose-matched EID and PCD scans of the COPDGene 2 phantom and 2 human volunteers were acquired. Phantom images were acquired at different radiation dose levels (CTDIvol: 3.0, 1.5, and 0.75 mGy) and different tube voltages (120, 100, and 80 kVp), while human images were acquired at vendor recommended low-dose lung cancer screening settings. EID and PCD images were compared for quantitative Hounsfield unit accuracy, noise levels, and contrast-to-noise ratios (CNR) for detection of ground-glass nodules (GGNs) and emphysema. Results: The PCD Hounsfield unit accuracy was better for water at all scan parameters, and for lung, GGN and emphysema equivalent regions of interest (ROIs) at 1.5 and 0.75 mGy. PCD attenuation accuracy was more consistent for all scan parameters (all P<0.01), while Hounsfield units for lung, GGN and emphysema ROIs changed significantly for EID with decreasing dose (all P<0.001). PCD showed lower noise levels at the lowest dose setting at 120, 100 and 80 kVp (15.2±0.3 vs 15.8±0.2, P=0.03; 16.1±0.3 vs 18.0±0.4, P=0.003; and 16.1±0.3 vs 17.9±0.3, P=0.001, respectively), resulting in superior CNR for the detection of GGNs and emphysema at 100 and 80 kVp. Significantly lower PCD noise levels were confirmed in volunteer images. Conclusion: PCD provided better Hounsfield unit accuracy for lung, ground-glass, and emphysema-equivalent foams at 1.5 and 0.75 mGy with less variability than EID. Additionally, PCD showed less noise, and higher CNR at 0.75 mGy for both 100 and 80 kVp. PCD technology may help reduce radiation exposure in lung cancer screening while maintaining diagnostic quality.

60 APPLIED LIFE SCIENCES↗

2010 Neutron Review: ORNL Neutron Sciences Progress Report

During 2010, the Neutron Sciences Directorate focused on producing world-class science, while supporting the needs of the scientific community. As the instrument, sample environment, and data analysis tools at High Flux Isotope Reactor (HFIR ) and Spallation Neutron Source (SNS) have grown over the last year, so has promising neutron scattering research. This was an exciting year in science, technology, and operations. Some topics discussed are: (1) HFIR and SNS Experiments Take Gordon Battelle Awards for Scientific Discovery - Battelle Memorial Institute presented the inaugural Gordon Battelle Prizes for scientific discovery and technology impact in 2010. Battelle awards the prizes to recognize the most significant advancements at national laboratories that it manages or co-manages. (2) Discovery of Element 117 - As part of an international team of scientists from Russia and the United States, HFIR staff played a pivotal role in the discovery by generating the berkelium used to produce the new element. A total of six atoms of ''ununseptium'' were detected in a two-year campaign employing HFIR and the Radiochemical Engineering Development Center at Oak Ridge National Laboratory (ORNL) and the heavy-ion accelerator capabilities at the Joint Institute for Nuclear Research in Dubna, Russia. The discovery of the new element expands the understanding of the properties of nuclei at extreme numbers of protons and neutrons. The production of a new element and observation of 11 new heaviest isotopes demonstrate the increased stability of super-heavy elements with increasing neutron numbers and provide the strongest evidence to date for the existence of an island of enhanced stability for super-heavy elements. (3) Studies of Iron-Based High-Temperature Superconductors - ORNL applied its distinctive capabilities in neutron scattering, chemistry, physics, and computation to detailed studies of the magnetic excitations of iron-based superconductors (iron pnictides and chalcogenides), a class of materials discovered in 2008. This research is yielding new insights into the relationship between magnetism and superconductivity and has established several key features of this family of high-temperature superconducting (HTS ) materials: the maximum magnetic field at which they can function, the nature of the electrons involved in the superconductivity, the dependence of the properties upon chemical substitution, and the character of the magnetic fluctuations in the material. The results suggest that despite important differences between these materials and the HTS copper oxides, a universal mechanism may be responsible for the unconventional superconductivity. (4) Coal Sequestration Research: A New Home for Greenhouse Gases - One possibility for slowing down the increasing levels of carbon dioxide (CO{sub 2}) in the atmosphere is to capture the gas in natural underground features such as coal seams. Critical to the feasibility of this technology is determining how much CO{sub 2} can be stored, no method for which has been found - until now. (5) Accelerator Reliability Passes 92% - In December 2010, SNS set a new record for itself when the accelerator ran at 1 MW with 100% reliability. Target Performance Exceeds All Expectations - The mercury target used at SNS is the first of its kind. During the design and planning for SNS, many people were skeptical that the target would work. In 2010, it was confirmed that the target was working not only well but much better than anyone would have imagined. (6) Changing the World of Data Acquisition - Researchers at SNS are starting to benefit from event-based data analysis. Event data mode captures and stores an individual data set for every single neutron that strikes a detector - precisely when and where the neutron is detected. This technique provides numerous advantages over traditional methods. Event data mode allows researchers to process their data at the highest resolution possible with no loss of data. This method of data collection provides a much more efficient way for users to gather data and get the most from their beam time. (7) New Laboratories for Users - The HFIR and SNS user communities continue to grow dramatically. In 2010, HFIR hosted 862 users and SNS 796, outpacing projections for both facilities. To meet the needs of those users, a new complex of 13 laboratories is now open for users at SNS. (8) Innovative Detectors Provide Relief from Helium-3 Shortage - Helium-3 ({sup 3}He) has been the gas of choice for gaseous detectors since the early days of neutron science. About two years ago, detector scientists worldwide faced the reality that stockpiles of {sup 3}He are dwindling rapidly, while demand for it has risen by a factor of five. The Neutron Sciences Detectors Group has developed two new types of detectors that don't rely on {sup 3}He: the Anger camera and the wavelength shifting fiber neutron detector, both of which use lithium ({sup 6}Li).

High Flux isotope reactor↗

Optimizing a Multiple Right-Hand Side Dslash Kernel for Intel Knights Corner

Lattice Quantum Chromo-dynamics (LQCD) stands unique as the only computationally tractable, non-perturbative, and model-independent quantum field theory of the strong nuclear force. LQCD achieves tractability by discretizing our familiar 4-dimensional space-time onto a hypercubic lattice. The core component of the linear solver used in LQCD is the Wilson-Dirac operator whose finite difference component is called the Dslash operator. Dslash is a nearest neighbor stencil operator summing matrix-vector multiplications over lattice points. The arithmetic intensity of a regular Dslash operator acting on a single vector in single precision is between 0.88-2.29 (depending on the tricks employed) making it memory bandwidth bound on most current architectures, including Xeon Phi. Maximizing the performance of Dslash on Xeon Phi requires deliberate exploitation of its architectural features. One key feature of Xeon Phi is its powerful vector processing unit with lengthy vector registers. There is a class of LQCD computations which aims to solve the Dirac equation for the same lattice gauge fields multiple times. In this case one can gain additional reuse of the gauge field, and potentially simplified vectorization, through applying Dslash to several vectors at the same time; often referred to as Multiple Right Hand Side (MRHS) Dslash. Previous work in this area focused on applying this approach to the Highly Improved Staggered operator, and in cases to the multi-Grid algorithm for Wilson clover fermions. Here, we present two approaches to MRHS Dslash: a vector register blocking approach and one using the software package QPhiX with a custom Code Generator for low-level intrinsics. We explore the Knights Corner instruction set, memory hierarchy, and threading capabilities. For standard lattice sizes used in testing, we observed significant speedups, with sustained performance of over 700 GFLOPS (single precision) in one instance. We achieved up to 27% of theoretical peak performance compared to a maximum of 12% obtained by the previous single RHS method in QPhiX.

Joó, Bálint↗

Network Sensitivity Solutions for Regional Moment Tensor Inversions

Well-resolved moment tensor solutions reveal information about the sources of seismic waves. Here we introduce a new way of assessing confidence in the regional full moment tensor inversion via the introduction of the network sensitivity solution (NSS). The NSS takes into account the unique station distribution, frequency band, and signal-to-noise ratio of a given event scenario. The NSS compares both a hypothetical pure source (for example an explosion or an earthquake) and the actual data with several thousand sets of synthetic data from a uniform distribution of all possible sources. The comparison with a hypothetical pure source provides the theoretically best-constrained source-type region for a given set of stations, and with it one can determine whether further analysis with the data is warranted. The NSS that employs the actual data gives a direct comparison of all other source-types with the best-fit source. In this way, one can choose a threshold level of fit where the solution is comfortably constrained. The method is tested for the well-recorded nuclear test, JUNCTION, at the Nevada Test Site. Sources that fit comparably well to a hypothetical pure explosion recorded with no noise at the JUNCTION data stations have a large volumetric component and are not described well by a double-couple (DC) source. The NSS using the real data from JUNCTION is even more tightly constrained to an explosion since the data contains some energy that precludes fitting with any type of deviatoric source. We also calculate the NSS for the October 2006 North Korea test and a nearby earthquake, where the station coverage is poor and the event magnitude is small. The earthquake solution is very well fit by a DC source, and the best-fit solution to the nuclear test (M{sub W}4.1) is dominantly explosion.

58 GEOSCIENCES↗

Search for V H and Technicolor Producion in the qqbb Final State Using the RunII DØ Detector

A search for dijet resonance production in a four-jet all-hadronic final state from the DØ detector at Fermilab’s Tevatron is presented. The data set, acquired at a p$\bar{p}$ center-of-mass energy of √s = 1.96 TeV, contains primarily multijet events and represents approximately 1 fb -1 of data. The cross section limits for associated Higgs production and Technicolor pro- cesses are determined through a background subtraction method using data to estimate the background. This four-jet channel is potentially very powerful, but is extremely challenging due to the large multijet background from QCD processes. Background rejection is performed by utilizing b-tagging, pre-selection cuts, a multi-variate boosted decision tree discriminant, and the correlated information contained in the M(bb) and M(jj) dijet invariant masses. The search for V H (WH+ZH) processes yields a 95% confidence level observed upper limit of 20.4 pb on the VH cross section for a Higgs mass of 115 GeV/c 2 . Additionally, a 95% confidence level observed upper limit of 16.7 pb was set for a Higgs boson mass of 125 GeV/c 2 and 24.6 pb was set for a Higgs boson mass of 135 GeV/c 2 . The same data set was used to place limits on the Technicolor process ρTC → WπTC where the technirho mass was fixed to 240 GeV/c 2 . For a technipion mass of 115 GeV/c 2 we find a 95% confidence level observed upper limit on the cross section of 49 pb. The technipion masses of 125 GeV/c 2 and 140 GeV/c 2 , the 95% confidence level observed upper limits are 57 pb and 71 pb, respectively.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Coulomb correlation effects in LaFeAsO: An LDA + DMFT(QMC) study

Effects of Coulomb correlation on the LaFeAsO electronic structure are investigated by the LDA + DMFT(QMC) method (combination of the local density approximation with the dynamic mean-field theory; impurity solver is a quantum Monte Carlo algorithm). The calculation results show that LaFeAsO is in the regime of intermediate correlation strength with a significant part of the spectral density moved from the Fermi energy to the Hubbard bands and far from the edge of the metal-insulator transition. Correlations affect iron d-orbitals differently. The t{sub 2g} states (xz, yz and x{sup 2} - y{sup 2} orbitals) have a higher energy due to crystal field splitting and are nearly half-filled. Their spectral functions have a pseudogap with the Fermi level position on the higher subband slope. The lower energy e{sub g} set (xy and 3z{sup 2} - r{sup 2} orbitals) have occupancies significantly larger than 1/2 with typically metallic spectral functions.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Robustness of Deep Learning Classification to Adversarial Input on GPUs: Asynchronous Parallel Accumulation Is a Source of Vulnerability

The ability of machine learning (ML) classification models to resist small, targeted input perturbations—known as adversarial attacks—is a key measure of their safety and reliability. We show that floating-point non associativity (FPNA) coupled with asynchronous parallel programming on GPUs is sufficient to result in misclassification, without any perturbation to the input. Additionally, we show that this misclassification is particularly significant for inputs close to the decision boundary and that standard adversarial robustness results may be overestimated up to 4.6 when not considering machine-level details. We first study a linear classifier, before focusing on standard Graph Neural Network (GNN) architectures and datasets used in robustness assessments. We develop a novel black-box attack using Bayesian optimization to discover external workloads that can change the instruction scheduling which bias the output of reductions on GPUs and reliably lead to misclassification. Motivated by these results, we present a new learnable permutation (LP) gradient-based approach to learning floating-point operation orderings that lead to misclassifications. The LP approach provides a worst-case estimate in a computationally efficient manner, avoiding the need to run identical experiments tens of thousands of times over a potentially large set of possible GPU states or architectures. Finally, using instrumentation-based testing, we investigate parallel reduction ordering across different GPU architectures under external background workloads, when utilizing multi-GPU virtualization, and when applying power capping. Our results demonstrate that parallel reduction ordering varies significantly across architectures under the first two conditions, substantially increasing the search space required to fully test the effects of this parallel scheduler-based vulnerability. These results and the methods developed here can help to include machine-level considerations into adversarial robustness assessments, which can make a difference in safety and mission critical applications.

Shanmugavelu, Sanjif [Maxeler Technologies, a Groq↗

Unsupervised Power System Event Detection and Classification Using Unlabeled PMU Data

This paper proposes a novel data-driven power system event detection and classification method based on 5TB of actual PMU measurements collected from the US western interconnect. Firstly, a set of comprehensive power quality rules are proposed to pre-filter the raw data and extract the regions of interest (ROI). Six distinct event categories are defined and corresponding patterns are chosen as references. Meanwhile, detailed characteristics of patterns are summarized to enhance our understanding of the actual events. Then, the time-independent feature vectors are generated by extracting the statistical, temporal, and spectral features from the raw time-series data. Furthermore, an ensemble model is proposed to cluster the events by combining multiple K-means clustering models using a voting strategy. Besides, both system-level and PMU-level clustering models are developed. The accuracy and robustness of the event detection method are further improved through interactive evaluation of the two-level clustering results. This paper summarizes the actual characteristics of each event category and provides a reliable basis for accurate label generation. The experiments demonstrate the effectiveness of the proposed event detection and classification method.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Unsupervised Power System Event Detection and Classification Using Unlabeled PMU Data

This paper proposes a novel data-driven power system event detection and classification method based on 5TB of actual PMU measurements collected from the US western interconnect. Firstly, a set of comprehensive power quality rules are proposed to pre-filter the raw data and extract the regions of interest (ROI). Six distinct event categories are defined and corresponding patterns are chosen as references. Meanwhile, detailed characteristics of patterns are summarized to enhance our understanding of the actual events. Then, the time-independent feature vectors are generated by extracting the statistical, temporal, and spectral features from the raw time-series data. Furthermore, an ensemble model is proposed to cluster the events by combining multiple K-means clustering models using a voting strategy. Besides, both system-level and PMU-level clustering models are developed. The accuracy and robustness of the event detection method are further improved through interactive evaluation of the two-level clustering results. This paper summarizes the actual characteristics of each event category and provides a reliable basis for accurate label generation. The experiments demonstrate the effectiveness of the proposed event detection and classification method.

24 POWER TRANSMISSION AND DISTRIBUTION↗

TH-A-9A-01: Active Optical Flow Model: Predicting Voxel-Level Dose Prediction in Spine SBRT

Purpose: To predict voxel-level dose distribution and enable effective evaluation of cord dose sparing in spine SBRT. Methods: We present an active optical flow model (AOFM) to statistically describe cord dose variations and train a predictive model to represent correlations between AOFM and PTV contours. Thirty clinically accepted spine SBRT plans are evenly divided into training and testing datasets. The development of predictive model consists of 1) collecting a sequence of dose maps including PTV and OAR (spinal cord) as well as a set of associated PTV contours adjacent to OAR from the training dataset, 2) classifying data into five groups based on PTV's locations relative to OAR, two “Top”s, “Left”, “Right”, and “Bottom”, 3) randomly selecting a dose map as the reference in each group and applying rigid registration and optical flow deformation to match all other maps to the reference, 4) building AOFM by importing optical flow vectors and dose values into the principal component analysis (PCA), 5) applying another PCA to features of PTV and OAR contours to generate an active shape model (ASM), and 6) computing a linear regression model of correlations between AOFM and ASM.When predicting dose distribution of a new case in the testing dataset, the PTV is first assigned to a group based on its contour characteristics. Contour features are then transformed into ASM's principal coordinates of the selected group. Finally, voxel-level dose distribution is determined by mapping from the ASM space to the AOFM space using the predictive model. Results: The DVHs predicted by the AOFM-based model and those in clinical plans are comparable in training and testing datasets. At 2% volume the dose difference between predicted and clinical plans is 4.2±4.4% and 3.3±3.5% in the training and testing datasets, respectively. Conclusion: The AOFM is effective in predicting voxel-level dose distribution for spine SBRT. Partially supported by NIH/NCI under grant #R21CA161389 and a master research grant by Varian Medical System.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Exploratory Visual Analysis of Anomalous Runtime Behavior in Streaming High Performance Computing Applications

Online analysis of runtime behavior is essential for performance tuning in streaming scientific workflows. Integration of anomaly detection and visualization is necessary to support human-centered analysis, such as verification of candidate anomalies utilizing domain knowledge. In this work, we propose an efficient and scalable visual analytics system for online performance analysis of scientific workflows toward the exascale scenario. Here, our approach uses a call stack tree representation to encode the structural and temporal information of the function executions. Based on the call stack tree features (e.g., execution time of the root function or vector representation of the tree structure), we employ online anomaly detection approaches to identify candidate anomalous function executions. We also present a set of visualization tools for verification and exploration in a level-of-detailed manner. General information, such as distribution of execution times, are provided in an overview visualization. The detailed structure (e.g., function invocation relations) and the temporal information (e.g., message communication) of the execution call stack of interest are also visualized. The usability and efficiency of our methods are verified in a real-world HPC application.

97 MATHEMATICS AND COMPUTING↗

Derivation of the conditional moment closure equations for spray combustion

In this work we derive the fundamental equations for conditional moment closure (CMC) modelling of individual phases set in a two-phase flow. The derivation is based on the instantaneous transport equations for the single phase that involve a level set/indicator function technique for accounting for interfaces. Special emphasis is put on spray combustion with the CMC equations formulated for the gas phase. The CMC equations are to be viewed as an adjunct to existing methods for the modelling of the dynamics of sprays: they provide a refinement of the modelling of chemical reactions in the gas phase. The resulting CMC equations differ significantly from those already in use in the literature. They contain, of course, unclosed terms that need to be modelled. Investigation of the unclosed terms associated with evaporation at the droplet surface is well beyond the capabilities of laboratory measurement or direct numerical simulation. It is proposed that modelling of these terms be based on the well-established 'laws' of similarity between heat and mass transfer: an example is detailed for one example of the general modelling of the spray dynamics. Other unclosed terms are important throughout the gas phase. Models used for these terms in single-phase flows are reviewed and it is proposed that any modifications needed for these models be investigated by DNS of suitable model problems having good resolution of the flow and mixing in the inter-droplet space. It is proposed that a spray analogue of the scalar mixing layer that has been widely studied in single-phase flows be used as the model problem for such DNS studies and also for LES and RANS modelling. (author)

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗