Developing Mango Graph Studio and its Applications for Bioinformatics and Systems Biology (SBIR Phase I Grant Final Technical Report)
Explore the source record for details and available documents.
SEARCH · Engineering Papers
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.
Explore the source record for details and available documents.
A new Viterbi decoder for convolutional codes with constraint lengths up to 15, called the Big Viterbi Decoder, is under development for the Deep Space Network. It will be demonstrated by decoding data from the Galileo spacecraft, which has a rate 1/4, constraint-length 15 convolutional encoder on board. Here, the mathematical theory underlying the design of the very-large-scale-integrated (VLSI) chips that are being used to build this decoder is explained. The deBruijn graph B sub n describes the topology of a fully parallel, rate 1/v, constraint length n+2 Viterbi decoder, and it is shown that B sub n can be built by appropriately wiring together (i.e., connecting together with extra edges) many isomorphic copies of a fixed graph called a B sub n building block. The efficiency of such a building block is defined as the fraction of the edges in B sub n that are present in the copies of the building block. It is shown, among other things, that for any alpha less than 1, there exists a graph G which is a B sub n building block of efficiency greater than alpha for all sufficiently large n. These results are illustrated by describing a special hierarchical family of deBruijn building blocks, which has led to the design of the gate-array chips being used in the Big Viterbi Decoder.
Big data processing systems such as Spark are employed in an increasing number of diverse applications-such as machine learning, graph computation, and scientific computing-each with dynamic and different resource needs. These applications increasingly run on heterogeneous hardware, e.g., with out-of-core accelerators. However, big data platforms do not factor in the multi-dimensional heterogeneity of applications and hardware. This leads to a fundamental mismatch between the application and hardware characteristics, and the resource scheduling adopted in big data platforms. For example, Hadoop and Spark consider only data locality when assigning tasks to nodes, and typically disregard the hardware capabilities and suitability to specific application requirements. In this paper, we present RUPAM, a heterogeneity-aware task scheduling system for big data platforms, which considers both task-level resource characteristics and underlying hardware characteristics, as well as preserves data locality. RUPAM adopts a simple yet effective heuristic to decide the dominant scheduling factor (e.g., CPU, memory, or I/O), given a task in a particular stage. Our experiments show that RUPAM is able to improve the performance of representative applications by up to 62.3% compared to the standard Spark scheduler.
This slide deck provides an overview of the data and resources available in the Sherlock Data Warehouse. Sherlock was developed and is currently maintained by the Aviation Systems Division at NASA Ames Research Center. Sherlock contains a valuable collection of flight, air traffic management, and weather data. But Sherlock is not just a data archive. Sherlock also includes tools and resources to access, download, and visualize data, as well as resources to process the data. This overview summarizes Sherlock data sources, demonstrates data analytics and visualization with MicroStrategy, illustrates disparate data integration using the ATM Knowledge graph, and presents a machine learning use case using the Big Data system.
The poster discusses data analysis of the WAVgraph database and applied machine learning methods for it. The database is a long-term project that seeks to be a comprehensive repository of information on cyber threats and is updated regularly. It was previously unanalyzed and unexplored. The goal was to learn more about it and its contents in order to have a better understanding and enable better use. The data analysis and discovery enabled further exploration through natural language processing, similarity, and clustering methods. The poster shows some of the insights from the analysis and explains the methods used for the machine learning applications.
Finding actionable insights from data has always been difficult. As the scale and forms of data increase tremendously, the task of finding value becomes even more challenging. Data scientists at Oak Ridge National Laboratory are leveraging unique leadership infrastructure (e.g. Urika-XA and Urika-GD appliances) to develop scalable algorithms for semantic, logical and statistical reasoning with unstructured Big Data. We present the deployment of such a framework called ORiGAMI (Oak Ridge Graph Analytics for Medical Innovations) on the National Library of Medicine s SEMANTIC Medline (archive of medical knowledge since 1994). Medline contains over 70 million knowledge nuggets published in 23.5 million papers in medical literature with thousands more added daily. ORiGAMI is available as an open-science medical hypothesis generation tool - both as a web-service and an application programming interface (API) at http://hypothesis.ornl.gov . Since becoming an online service, ORIGAMI has enabled clinical subject-matter experts to: (i) discover the relationship between beta-blocker treatment and diabetic retinopathy; (ii) hypothesize that xylene is an environmental cancer-causing carcinogen and (iii) aid doctors with diagnosis of challenging cases when rare diseases manifest with common symptoms. In 2015, ORiGAMI was featured in the Historical Clinical Pathological Conference in Baltimore as a demonstration of artificial intelligence to medicine, IEEE/ACM Supercomputing and recognized as a Centennial Showcase Exhibit at the Radiological Society of North America (RSNA) Conference in Chicago. The final paper will describe the workflow built for the Cray Urika-XA and Urika-GD appliances that is able to reason with the knowledge of every published medical paper every time a clinical researcher uses the tool.
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%.
Graph theory applications within the physical sciences have a long history. However, it has never moved to the forefront of analytical techniques employed by the computational chemistry community. Though much of the earliest work leveraged percolation theory as a backdrop for understanding critical phenomena and phase transitions within statistical mechanical simulations, modern data science has much to contribute. Here in this paper, we focus on the realm of intermolecular networks, where vertices represent molecules or particles, and an edge represents an intermolecular or interparticle interaction. These interactions are widely understood to dictate many of the physical properties of soft matter, and for liquids in particular, understanding this network is essential. Consider the sheer volume of ~70 years of literature associated with the hydrogen bond network of water, a topic that remains an active area of research even today. The aim of this work is twofold: first, to put into context prior algorithms and analyses that employed “connectedness theory” or lattice-based models to understand the intermolecular networks of chemical systems; and second to discuss a more general strategy for analyzing the multiscale structural and dynamic properties of chemical systems—intermolecular network theory. This approach encompasses foundational methods such as percolation theory, but also utilizes contemporary graph theoretical analyses that have evolved alongside the development of the World Wide Web, cloud computing, and big data. It is the realm of intermolecular network theory to determine the relationships between the essential physics of a system and the topological properties of the network and to derive new techniques in graph theory that can be related to the underlying physico-chemical or reactive properties of a system. Examples are used throughout this chapter that demonstrate applications to a wide variety of complex chemical systems, including ion association within electrolytes, mass transport and the properties of liquid–liquid phase boundaries, and multicomponent solutions.
In this era of data-driven decisions and discovery where Big Data is producing Bigger Data, data scientists at the Oak Ridge National Laboratory are leveraging unique leadership infrastructure (e.g., Urika XA and Urika GD appliances) to develop scalable algorithms for semantic, logical and statistical reasoning with Big Data (i.e., data stored in databases as well as unstructured data in documents). ORiGAMI is a next-generation knowledge-discovery framework that is: (a) knowledge nurturing (i.e., evolves seamlessly with newer knowledge and data), (b) smart and curious (i.e. using information-foraging and reasoning algorithms to digest content) and (c) synergistic (i.e., interfaces computers with what they do best to help subject-matter-experts do their best. ORiGAMI has been demonstrated using the National Library of Medicine's SEMANTIC MEDLINE (archive of medical knowledge since 1994).
In this era of data-driven decisions and discovery where Big Data is producing Bigger Data, data scientists at the Oak Ridge National Laboratory are leveraging unique leadership infrastructure (e.g., Urika XA and Urika GD appliances) to develop scalable algorithms for semantic, logical and statistical reasoning with Big Data (i.e., data stored in databases as well as unstructured data in documents). ORiGAMI is a next-generation knowledge-discovery framework that is: (a) knowledge nurturing (i.e., evolves seamlessly with newer knowledge and data), (b) smart and curious (i.e. using information-foraging and reasoning algorithms to digest content) and (c) synergistic (i.e., interfaces computers with what they do best to help subject-matter-experts do their best. ORiGAMI has been demonstrated using the National Library of Medicine's SEMANTIC MEDLINE (archive of medical knowledge since 1994).
We propose a new algorithm to improve the strong scalability of right-looking sparse LU factorization on distributed memory systems. Our 3D algorithm for sparse LU uses a three-dimensional MPI process grid, exploits elimination tree parallelism, and trades off increased memory for reduced per-process communication. We also analyze the asymptotic improvements for planar graphs (e.g., those arising from 2D grid or mesh discretizations) and certain non-planar graphs (specifically for 3D grids and meshes). For a planar graph with $n$ vertices, our algorithm reduces communication volume asymptotically in $n$ by a factor of $\mathscr{O}$ $\Big(\sqrt{log \ n}\Big)$ and latency by a factor of $\mathscr{O}$ $(log \ n)$. For nonplanar cases, our algorithm can reduce the per-process communication volume by 3× and latency by $\mathscr{O}$ $\Big(n^\frac13\Big)$ times. In all cases, the memory needed to achieve these gains is a constant factor. We implemented our algorithm by extending the 2D data structure used in SuperLU_DIST. Our new 3D code achieves empirical speedups up to 27× for planar graphs and up to 3.3× for non-planar graphs over the baseline 2D SuperLU_DIST when run on 24,000 cores of a Cray XC30. We extend the 3D algorithm for heterogeneous architectures by adding the Highly Asynchronous Lazy Offload (Halo) algorithm for co-processor offload. On 4096 nodes of a Cray XK7 with 32,768 CPU cores and 4096 Nvidia K20x GPUs, the 3D algorithm achieves empirical speedups up to 24× for planar graphs and 3.5× for non-planar graphs over the baseline 2D SuperLU_DIST with co-processor acceleration.
Practicaldataminingrarelyfalls exactlyinto the supervisedlearning scenario. Rather, the growing amount of unlabeled data poses a big challenge to large-scale semi-supervised learning (SSL). We note that the computationalintensivenessofgraph-based SSLarises largely from the manifold or graph regularization, which in turn lead to large models that are dificult to handle. To alleviate this, we proposed the prototype vector machine (PVM), a highlyscalable,graph-based algorithm for large-scale SSL. Our key innovation is the use of"prototypes vectors" for effcient approximation on both the graph-based regularizer and model representation. The choice of prototypes are grounded upon two important criteria: they not only perform effective low-rank approximation of the kernel matrix, but also span a model suffering the minimum information loss compared with the complete model. We demonstrate encouraging performance and appealing scaling properties of the PVM on a number of machine learning benchmark data sets.
The increasing interest in big data applications is ushering in a large effort in seeking new, efficient, and adapted data models to reduce complexity, while preserving maximal intrinsic information. Graph-based models have recently been getting a lot of attention on account of their intuitive and direct connection to the data. The cost of these models, however, is to some extent giving up geometric insight as well as algebraic flexibility. Topology, as an intermediate analysis medium, focuses on coarse structures of an object/signal in general. It affords a formalism of transitioning from a local to a global description of an object, while providing significant information, which respects the local structure of measurements. It may also support a global visualization (e.g., data variation trends) and enhances the understanding of the underlying phenomenon. Throughout this article, measurements may be considered proximal in various ways, depending on the specific application, and the selected proximity metrics are subsequently used in the construction of a graph structure, thus highlighting various groups of neighboring data. As described later, these groups, often based on proximity criteria, are represented by n-simplices, where n + 1 corresponds to the number of points in an associated group. Collectively, they form what is (and will be throughout) referred to as a simplicial complex, a special representation of a topological space. We find the key advantage of computational topology is its inherently algebraic structure on the basic elements of the resulting graph-like structure.
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.
Abstract Solving multi‐objective combinatorial optimization problems can quickly become computationally challenging when applied to large networks generated from big data. Present‐day first‐rate Mixed Integer Programming (MIP) solvers have parallelism built‐in to take advantage of multicore architectures; but specialized network optimization algorithms that can often solve graph problems more efficiently than a general MIP solver are typically programmed serially. Thus, these network algorithm implementations do not take full advantage of modern multicore computer capabilities. Many parallel computing languages include a fork/join framework into their concurrency packages that allow for simple conversion of recursive serial algorithms to operate in parallel without having to use complicated low‐level message‐passing interfaces. Fork/join is particularly well suited to divide and conquer algorithms such as Non‐Inferior Set Estimation (NISE) method for computing the supported Pareto front of a multi‐objective optimization problem. This work develops a simple general parallel NISE (pNISE) implementation using the Java 7 concurrency tools, and then offers results from a specific implementation of solving a bi‐objective shortest path problem. The results indicate that this method is capable of sizeable speed‐ups on both small‐scale personal computers as well as large‐scale shared memory supercomputers. Finally, other network problems suitable to this method are discussed, including multi‐objective variants of the minimum spanning tree problem, transportation problem, assignment problem, maximum flow problem, and the minimum‐cost flow problem. Copyright © 2015 John Wiley & Sons, Ltd.
A prototype hardware Big Viterbi Decoder (BVD) was completed for an experiment with the Galileo Spacecraft. Searches for new convolutional codes, studies of Viterbi decoder hardware designs and architectures, mathematical formulations, and decompositions of the deBruijn graph into identical and hierarchical subgraphs, and very large scale integration (VLSI) chip design are just a few examples of tasks completed for this project. The BVD bit error rates (BER), measured from hardware and software simulations, are plotted as a function of bit signal to noise ratio E sub b/N sub 0 on the additive white Gaussian noise channel. Using the constraint length 15, rate 1/4, experimental convolutional code for the Galileo mission, the BVD gains 1.5 dB over the NASA standard (7,1/2) Maximum Likelihood Convolution Decoder (MCD) at a BER of 0.005. At this BER, the same gain results when the (255,233) NASA standard Reed-Solomon decoder is used, which yields a word error rate of 2.1 x 10(exp -8) and a BER of 1.4 x 10(exp -9). The (15, 1/6) code to be used by the Cometary Rendezvous Asteroid Flyby (CRAF)/Cassini Missions yields 1.7 dB of coding gain. These gains are measured with respect to symbols input to the BVD and increase with decreasing BER. Also, 8-bit input symbol quantization makes the BVD resistant to demodulated signal-level variations which may cause higher bandwidth than the NASA (7,1/2) code, these gains are offset by about 0.1 dB of expected additional receiver losses. Coding gains of several decibels are possible by compressing all spacecraft data.
The explosive growth in Earth observational data in the recent decade demands a better method of interoperability across heterogeneous systems. The Earth science data system community has mastered the art in storing large volume of observational data, but it is still unclear how this traditional method scale over time as we are entering the age of Big Data. Indexed search solutions such as Apache Solr (Smiley and Pugh, 2011) provides fast, scalable search via keyword or phases without any reasoning or inference. The modern search solutions such as Googles Knowledge Graph (Singhal, 2012) and Microsoft Bing, all utilize semantic reasoning to improve its accuracy in searches. The Earth science user community is demanding for an intelligent solution to help them finding the right data for their researches. The Ontological System for Context Artifacts and Resources (OSCAR) (Huang et al., 2012), was created in response to the DARPA Adaptive Vehicle Make (AVM) programs need for an intelligent context models management system to empower its terrain simulation subsystem. The core component of OSCAR is the Environmental Context Ontology (ECO) is built using the Semantic Web for Earth and Environmental Terminology (SWEET) (Raskin and Pan, 2005). This paper presents the current data archival methodology within a NASA Earth science data centers and discuss using semantic web to improve the way we capture and serve data to our users.
Belief propagation (BP) on LDPC codes is an iterative decoding algorithm that performs information transfer on the Tanner graph, which represents the code. In each iteration, the algorithm exchanges information (LLR) between variable nodes and check nodes through the edges of the graph. LLR values represent the probability that a given bit in a transmitted codeword equals 0 or 1, given a received word. In the hardware part, recent advancements in intelligent technologies, such as artificial intelligence, big data analytics, autonomous vehicles, and speech/image recognition, have heightened the demand for faster calculations and reduced energy consumption.