Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “vertex”

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 397 records · Page 22
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Architecture-Preserving Provable Repair of Deep Neural Networks

Deep neural networks (DNNs) are becoming increasingly important components of software, and are considered the state-of-the-art solution for a number of problems, such as image recognition. However, DNNs are far from infallible, and incorrect behavior of DNNs can have disastrous real-world consequences. This paper addresses the problem of architecture-preserving V-polytope provable repair of DNNs. A V-polytope defines a convex bounded polytope using its vertex representation. V-polytope provable repair guarantees that the repaired DNN satisfies the given specification on the infinite set of points in the given V-polytope. An architecture-preserving repair only modifies the parameters of the DNN, without modifying its architecture. The repair has the flexibility to modify multiple layers of the DNN, and runs in polynomial time. It supports DNNs with activation functions that have some linear pieces, as well as fully-connected, convolutional, pooling and residual layers. To the best our knowledge, this is the first provable repair approach that has all of these features. We implement our approach in a tool called APRNN. Using MNIST, ImageNet, and ACAS Xu DNNs, we show that it has better efficiency, scalability, and generalization compared to PRDNN and REASSURE, prior provable repair methods that are not architecture preserving.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

cuAlign: Scalable Network Alignment on GPU Accelerators

Given two graphs, the objective of network alignment is to find the best one-to-one mapping of vertices in one graph (??) to vertices in the other (??), such that the number of overlaps is maximized. We say that edges(??, ??) ???and(??', ??') ??? are overlapped if ?? is mapped to ??' and ?? is mapped to??'. Network alignment is an important optimization problem with several applications in bioinformatics, computer vision and ontology matching. Since it is an NP-hard problem, efficient heuristics and scalable implementations are necessary. In this work, we introduce a new framework that combines the concepts of intra-network proximity using vertex embedding,Belief Propagation (BP) and approximate weighted matching, and provides qualitative improvements up to22%over state-of-the-art approaches. We also provide scalable implementations on GPU accelerators, demonstrating up to19×speedup for Belief Propagation and 3× speedup for approximate weighted matching relative to previous multithreaded implementation. A combination of combinatorial and algebraic kernels within the network alignment algorithm poses significant hurdles for parallelization. Load imbalance and irregular DRAM traffic limit achievable performance on GPUs. Our novel approach identifies and exploits unique structural proper-ties of the BP-based algorithm and employs code fusion to reduce data movement between different steps of the algorithm. Using a diverse set of inputs, we demonstrate qualitative improvements of our algorithms, and performance gains of our GPU-accelerated implementation. We believe that our work will enable algorithmic improvements and practical applications of network alignment.

Xiang, Lizhi↗
DOE OSTI2021

krowkee

krowkee is a toolkit for scalably and efficiently summarizing many data streams in distributed memory. krowkee is intended for applications where one needs to summarize huge loosely structured data, such as matrices or graphs, where individual components such rows/columns or vertex adjacency information are impractical to store and directly inspect. krowkee ingests these objects as data streams - unstructured, arbitrarily ordered lists of updates - and accumulates summaries thereof in the form of data sketches.

Dunton, AlecM.↗
DOE OSTI2022

ECP-ExaGraph/Submodular-b-matching

A b-MATCHING is a subset of edges M such that at most b(v) edges in M are incident on each vertex v, where b(v) is specified. We present a distributed-memory parallel algorithm, b-SUITOR, that computes a b-MATCHING with more than half the maximum weight in a graph with weights on the edges

Ferdous, S M↗
DOE OSTI2020

Sphynx

Sphynx is a software library for parallel graph partitioning and load balancing. It partitions a set of vertices in a graph such that the vertex parts are balanced, and the edge cut is approximately minimized. The primary use case is to partition a computational graph into smaller tasks that can be assigned to processors (cores) for efficient parallel computing. Sphynx is based on the Trilinos scientific computing toolkit so can run on both multi-CPU and multi-GPU systems. SAND2020-12183 M Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-NA0003525.

Boman, Erik↗
DOE OSTI2022

A resilient network recovery framework against cascading failures with deep graph learning

Because of the increasing importance and dependencies of infrastructure networks and the potential for massive cascading failures in real-world network systems, maintenance optimization to effectively reduce system performance loss caused by diverse disruptions is of significant interest among researchers and practitioners. In this work, a new recovery framework was developed to rapidly identify important system components for maintenance to improve network resilience against cascading failures. Here this work provides distinct advantages to determine an optimal maintenance priority by combining real-time network structure importance with other maintenance prioritization based on customer preference. This approach adopts structural graph embedding and deep reinforcement learning to extract real-time network topology information (such as minimum vertex cover) to update the maintenance priority during the recovery process. Based on the case studies on synthetic networks and a US airport network, the proposed recovery framework with real-time network topology awareness shows better performance than other maintenance prioritization strategies regarding resilience enhancement. This work improves the understanding of how the changing network structure influences maintenance effects. It also provides insights of the practical usefulness of advanced deep learning on helping optimal maintenance prioritization to effectively reduce the intensity and extent of cascading failures.

42 ENGINEERING↗
DOE OSTI2020

Jeankempite, Ca 5 (AsO 4 ) 2 (AsO 3 OH) 2 (H 2 O) 7 , a new arsenate mineral from the Mohawk Mine, Keweenaw County, Michigan, USA

Abstract Jeankempite, Ca 5 (AsO 4 ) 2 (AsO 3 OH) 2 (H 2 O) 7 , is a new mineral species (IMA2018-090) discovered amongst coatings of arsenate minerals on oxidised copper arsenides from the Mohawk No. 2 mine, Mohawk, Keweenaw County, Michigan, USA. The new mineral occurs as lamellar bundles of colourless to white plates up to 1 mm wide and is visually indistinguishable from guérinite, with which it forms intergrowths. Jeankempite is transparent to translucent with a waxy lustre and white streak, is non-fluorescent under longwave and shortwave ultraviolet illumination, has a Mohs hardness of ~1.5 and brittle tenacity with uneven fracture. Crystals are flattened on {01 $\bar{1}$ } and exhibit perfect cleavage on {01 $\bar{1}$ }. Optically, jeankempite is biaxial (+), α = 1.601(2), β = 1.607(2), γ = 1.619(2) (white light); 2V meas. = 72(2)° and 2V calc. = 71.0°. The empirical formula is (Ca 4.97 Na 0.013 Mg 0.017 )(As 3.99 S 0.01 ) 4 O 23 H 16 , based on 23 O and 16 H atoms per formula unit. Thermogravimetric analysis indicates that jeankempite undergoes four weight losses totalling 16.82%, close to the expected loss of 16.30%, corresponding to eight H 2 O. Jeankempite is triclinic, P $\bar{1}$ , a = 6.710(6), b = 14.901(14), c = 15.940(15) Å, α = 73.583(12)°, β = 81.984(12)°, γ = 82.754(12)°, V = 1507(2) Å 3 and Z = 3. The final structure was refined to R 1 = 0.0591 for 2781 reflections with I obs > 3σ I . The crystal structure of jeankempite is built from a network of edge- and vertex-sharing CaO 6 , CaO 7 and AsO 4 polyhedra, and we hypothesise that the new mineral has formed due to a topotactic reaction brought on by dehydration of preexisting guérinite.

Mineralogy↗
DOE OSTI2022

graphenv: a Python library for reinforcement learning on graph search spaces

Many important and challenging problems in combinatorial optimization (CO) can be expressed as graph search problems, in which graph vertices represent full or partial solutions and edges represent decisions that connect them. Graph structure not only introduces strong relational inductive biases for learning (Battaglia et al., 2018) - in this context, by providing a way to explicitly model the value of transitioning (along edges) between one search state (vertex) and the next - but lends itself to problems both with and without clearly defined algebraic structure. For example, classic CO problems on graphs such as the Traveling Salesman Problem (TSP) can be expressed as either pure graph search or integer programs. Other problems, however, such as molecular optimization, do no have concise algebraic formulations and yet are readily implemented as a graph search (V. et al., 2022; Zhou et al., 2019). Such "model-free" problems constitute a large fraction of modern reinforcement learning (RL) research owing to the fact that it is often much easier to write a forward simulation that expresses all of the state transitions and rewards, than to write down the precise mathematical expression of the full optimization problem. In the case of molecular optimization, for example, one can use domain knowledge alongside existing software libraries to model the effect of adding a single bond or atom to an existing but incomplete molecule, and let the RL algorithm build a model of how good a given decision is by "experiencing" the simulated environment many times through. In contrast, a model-based mathematical formulation that fully expresses all the chemical and physical constraints is intractable. In recent years, RL has emerged as an effective paradigm for optimizing searches over graphs and led to state-of-the-art heuristics for games like Go and chess, as well as for classical CO problems such as the TSP. This combination of graph search and RL, while powerful, requires non-trivial software to execute, especially when combining advanced state representations such as Graph Neural Networks (GNN) with scalable RL algorithms.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Open string stub as an auxiliary string field

Witten’s open string field theory with a generalized version of stubs is reformulated as a cubic string field theory using an auxiliary string field. The gauge symmetries and equations of motion as well as the associative algebra of the resulting theory are investigated. Integrating out either the original or auxiliary field is shown to recover the conventional cubic theory. Our analysis demonstrates that deformations due to the stubs can be described as a homotopy transfer purely in the context of strong deformation retract. We also discuss to what extent the vertex regions resulting from stubs provide a model for the elementary interactions of closed string field theory.

Physics↗
DOE OSTI2022

Transverse Single Spin Asymmetries of Heavy Flavor Electrons and Charged Pions in 200 GeV $p+p^{\uparrow}$ Collisions at Midrapidity

Transverse single spin asymmetries of particles produced in p+p^{\uparrow} p + p ↑ collisions provide insight on the partonic spin and momentum structure of hadrons; heavy flavor electrons provide access to initial state spin-momentum correlations of gluons in the proton, while charged pions provide access to initial and final state transverse spin effects. Charged particles are measured at midrapidity at PHENIX using the silicon vertex detector and central arm spectrometer, made of an electromagnetic calorimeter, a ring-imaging Cherenkov detector, and drift and pad chambers. Recent results for both electron and charged pion measurements from the 2015 running period will be presented.

Fitzgerald, Dillon↗
DOE OSTI2020

Graph Theory and IC Component Design Analysis

Graph analysis in large integrated circuit (IC) designs is an essential tool for verifying design logic and timing via dynamic timing analysis (DTA). IC designs resemble graphs with each logic gate as a vertex and the conductive connections between gates as edges. Using DTA digital statistical correlations, graph condensation, and graph partitioning, it is possible to identify high-entropy component centers and paths within an IC design. Identification of high-entropy component centers (HECC) enables focused DTA, effectively lowering the computational complexity of DTA on large integrated circuit graphs. In this paper, a devised methodology termed IC layout subgraph component center identification (CCI) is used to identify described. CCI lowers DTA computationally complexity by condensing IC graphs into reduced subgraphs in which dominant logic functions are verified.

42 ENGINEERING↗
DOE OSTI2021

Universal Monte Carlo Event Generator

With the Jefferson Lab 12 GeV physics program underway and plans for the future Electron-Ion Collider (EIC), the nuclear physics community is entering a new era of exploration of QCD phenomena involving extensive data taking and event-level processing. This brings with it the potential for unprecedented access to multidimensional particle momentum distributions (PMDs) that can be connected to various theoretical frameworks by unfolding the emergent quantum mechanical properties of QCD using the PMDs. In practice, the PMDs are rendered as discretized histograms (typically one- or two-dimensional projections), and detector effects must be taken into account to unfold the pure detector effect-free PMDs that can be connected with theory. One of the challenges in this new era is obtaining faithful reconstructions of the multidimensional PMDs that preserve all of the inherent particle correlations. In this LDRD project we developed a novel approach using machine learning (ML) that solves this challenge, by avoiding entirely the need to use histograms as the main numerical technique to obtain the detector effect-free PMDs. The new approach is, moreover, scalable to higher dimensional PMDs. The central idea involves training neural networks (NNs) to generate synthetic event-level data (momenta 4-vectors of final state particles) to preserve all correlations among the particles. This is achieved by converting the trial synthetic vertex-level events to detector-level events using detector simulators. The NNs are then tuned using a specialized distance metric between the synthetic detector events and the real detector events.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2022

Material point generation from an STL file

In this short note we describe a method for generating material points from a file in the stere olithography (STL) format commonly used in computer aided design (CAD), 3-D printing, and computer aided manufacturing applications. In this file format the surface of an object is described by triangles in three-dimensional space. The coordinates of the triangle vertexes are provided in the file. In this note we assume that the object of interest is inside of the enclosing set of triangles. Similar methods can also be used to generate material points outside of the enclosing triangles.

42 ENGINEERING↗
DOE OSTI2023

Search for Dark Matter using mono-Higgs and the ATLAS Pixel Detector (Final Report)

A large component of the mass-energy of the universe is composed of dark matter (DM), whose properties and interactions with known particles are not yet understood. Searches for DM at the Large Hadron Collider (LHC) provide important information, complementary to direct and indirect detection experiments, that is necessary to determine whether an observed signal indeed stems from DM. Furthermore, the discovery of the Higgs boson provides a unique avenue to search for DM because the potential interaction of the Higgs with DM would lead to the unique signature of a Higgs boson recoiling against DM. This process is typically referred to as mono-Higgs because DM does not interact strongly with most known particles and will therefore pass unseen through the detector, leading to a single detected Higgs boson and a large imbalance of momentum. Due to the strength of the interactions of the Higgs with Standard Model particles, it is unlikely for a Higgs boson to be radiated from initial state quarks. Therefore, the observation of this process would provide direct insight into the mechanism by which DM couples to known particles. The objective of this research program is to search for the dark matter particles produced in association with a Higgs boson at the LHC using the ATLAS (A Toroidal LHC Apparatus) detector, specifically when the Higgs boson decays to two bottom quarks. This research will benefit greatly from the development of innovative identification techniques of boosted Higgs bosons. The program includes the upgrade of the ATLAS Pixel readout system, which is critical to maintain high performance of tracking, vertexing, and boosted Higgs tagging in the planned higher luminosity run of the LHC.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2023

TPC Test-stands: An Overview & Future Prospects

The DUNE near detector design in Phase II will include a magnetized high-pressure gaseous-argon TPC (HPgTPC) surrounded by a calorimeter, called ND-GAr. ND-GAr is designed to provide fine-grained tracking, and due to its low detection threshold, it will be able to constrain one of the least-understood uncertainties for oscillation analysis: nuclear effects in argon at the neutrino interaction vertex. This talk will provide an overview of and an outlook about the ongoing R&D efforts in the context of the existing test stands that are evaluating the various readout system options for DUNE ND-GAr.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2023

Light long-lived particles at colliders

Searches for long-lived particles (LLPs) are a rapidly expanding frontier at the LHC and other collider experiments. Still, many gaps remain in the current search program, in particular for LLPs with masses at the GeV or sub-GeV scale and with very large decay lengths. In this talk, I will illustrate two different approaches to filling this gap by discussing two models of light LLPs and their associated collider signals. First, I will show that the dominant decay mode of vectorlike leptons can be a very long-lived pseudoscalar and a tau lepton and argue that the muon chambers of CMS or ATLAS are ideal places to search for this final state. Second, I will illustrate the excellent sensitivity of Belle II to light LLPs with meter-scale decay lengths using the example of displaced vertex signals from strongly interacting dark sectors.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2023

Vectorization of Dynamic Subgraphs via Generative Models (Final Report)

An important class of data analysis tasks stem from comparing subsets of connected records within massive sets of complex relational data. A common approach is to represent each set of connected records with a small graph, or set of data entities (graph vertices) and their relationships (graph edges), and efficient methods to gauge similarity for pairs of graphs are of high interest. This project concentrated on dynamic graphs, where each edge record has an associated timestamp denoting the time of observation. Pre-existing techniques for comparing dynamic graphs concentrate on either computing graph edit distance (number of vertex and edge deletion, addition, and timestamp modifications) or vectorizing the graph with counts of a limited set of dynamic graph motifs (tiny fundamental subgraphs) and computing distances between the vectors. These approaches are less able to see similarities in graphs that are fairly different in size but come from identical graph generation processes. The motif counting approach can be improved for graphs from the same process, but suffers from requiring many types of motifs meaning it is expensive. Moreover, many motifs are not present for small graphs, meaning realizing a a much larger graph came from the same process is difficult.

97 MATHEMATICS AND COMPUTING↗