Large Computational Problems Research Articles

A Vehicle Detection Algorithm Based on Compressive Sensing and Background Subtraction

An efficient vehicle detection method is a necessity for the intelligent transportation system under the complicated traffic environment at present. To solve the problems of large computation and the poor real-time in traditional vehicle detecting methods, this article proposes a real-time vehicle detecting algorithm integrated compressive sensing (CS) theories and background subtraction method. In addition, this paper undertakes the reconstruction of foreground image of vehicle based on the orthogonal matching pursuit (OMP) algorithm. Proved by the experimental result, the proposed vehicle detection algorithm could produce higher precision detection, smaller calculation and higher-quality reconstructed image compared to the traditional ones.

Guest Editorial: Field-Programmable Technology

Field-Programmable Technology (FPT) describes those electronic systems where the hardware as well as the software can be programmed on an application by application basis. Field Programmable Gate Arrays (FPGAs) represent the most common form of FPT, widely used in applications such as signal and image processing, telecommunications and computer networking. FPT continues to intrigue researchers; current research ranges from transistor-level programmable logic, through to high performance applications, and encompasses the design methodologies and design tools needed for such systems. In “Reconfigurable Blocks Based on Balanced Ternary,” Paul Beckett and Tayab Memon describe a new programmable logic block which takes advantage of some of the unique characteristics of Silicon-onInsulator in order to deliver balanced ternary (−1,0,+1) logic functions and memory cells. In “Using Data Contention in Dual-ported Memories for Security Applications”, Tim Guneysu investigates the problem of how to protect design IP in a technology where hardware programs are difficult to secure. His paper explores device-specific behaviour during write collisions in dual-port FPGA memories to support IP protection. For embedded systems that use floating point arithmetic, the choice has been fast hardware or slow software implementations. In “Improving Floating-Point Performance in Less Area: Fractured Floating Point Units (FFPUs),” Neil Hockert and Katherine Compton look at partial hardware support for floating point, which gives better performance than software floating point, and lower area than a full hardware unit. As logic circuits get larger, the time for physical design at logic level discourages broad exploration of high-level architectural design alternatives. In “Rapid Synthesis and Simulation of Computational Ciruits in an MPPA,” David Grant, Graeme Smecher, Guy Lemieux and Rosemary Francis present a tool flow (RVETool) for rapidly compiling computational circuits into a Massively Parallel Processor Array. In “Automated Mapping of the MapReduce Pattern onto Parallel Computing Platforms,” Qiang Lie, Tim Todman, Wayne Luk and George Constantinides explore using FPGAs effectively for large computational problems. Having identified that many applications use a “MapReduce” computational pattern, they develop a L. Shannon (*) School of Engineering Science, Simon Fraser University, Burnaby, Canada e-mail: lshannon@ensc.sfu.ca

Ender Wiggin Played Mafia Wars Too

Games hosted in social networks, gaming servers or mobile phones can be used as massively parallel, but surreptitious processing networks. These grids, and the activities of the unwitting players, can be used for good (such as solving large resource allocation problems or finding improvised explosive devices) or for evil (to allow criminals to gain unauthorized access to a system or break encryption algorithms). Surreptitious processing networks differ from botnets in that no rouge processing elements are present on the user side; instead, the computational payload is outsourced to humans who unknowingly provide partial solutions as they play the game. This approach makes the solution of large computational problems and the recruitment and exploitation of participants easy. This department is part of a special issue on social networking.

Strategy Alternatives for Homeland Air and Cruise Missile Defense

Air and cruise missile defense of the U.S. homeland is characterized by a requirement to protect a large number of critical assets nonuniformly dispersed over a vast area with relatively few defensive systems. In this article, we explore strategy alternatives to make the best use of existing defense resources and suggest this approach as a means of reducing risk while mitigating the cost of developing and acquiring new systems. We frame the issue as an attacker-defender problem with simultaneous moves. First, we outline and examine the relatively simple problem of defending comparatively few locations with two surveillance systems. Second, we present our analysis and findings for a more realistic scenario that includes a representative list of U.S. critical assets. Third, we investigate sensitivity to defensive strategic choices in the more realistic scenario. As part of this investigation, we describe two complementary computational methods that, under certain circumstances, allow one to reduce large computational problems to a more manageable size. Finally, we demonstrate that strategic choices can be an important supplement to material solutions and can, in some cases, be a more cost-effective alternative.

Distributed computing as a virtual supercomputer: Tools to run and manage large-scale BOINC simulations

Distributed computing (DC) projects tackle large computational problems by exploiting the donated processing power of thousands of volunteered computers, connected through the Internet. To efficiently employ the computational resources of one of world's largest DC efforts, GPUGRID, the project scientists require tools that handle hundreds of thousands of tasks which run asynchronously and generate gigabytes of data every day. We describe RBoinc, an interface that allows computational scientists to embed the DC methodology into the daily work-flow of high-throughput experiments. By extending the Berkeley Open Infrastructure for Network Computing (BOINC), the leading open-source middleware for current DC projects, with mechanisms to submit and manage large-scale distributed computations from individual workstations, RBoinc turns distributed grids into cost-effective virtual resources that can be employed by researchers in work-flows similar to conventional supercomputers. The GPUGRID project is currently using RBoinc for all of its in silico experiments based on molecular dynamics methods, including the determination of binding free energies and free energy profiles in all-atom models of biomolecules.

Calculation of left ventricular volume based on GVF model and optical flow field

Concerning the problems of difficult segmentation and large computation in calculating the Left Ventricular (LV) volume using tagged cardiac magnetic resonance image,a new method of calculating the LV volume was proposed. First an advanced snake model based on the gradient vector field (GVF-snake) was used to segment the contour of the left ventricular and an initial snake line of the contour was got,then optical flow method was used to track the contour of the LV in the successive images. Finally,the Simpson method was applied to calculate the volume of the LV. The comparison of the result by this method with the result got manually shows that the combination of the GVF-snake model and optical flow method is feasible to calculate the volume of the LV.

TOWARDS A SCALABLE SCIENTIFIC DATA GRID MODEL AND SERVICES

Scientific Data Grid mostly deals with large computational problems. It provides geographically distributed resources for large-scale data-intensive applications that generate large scientific data sets. This required the scientist in modern scientific computing communities involved in managing massive amounts of a very large data collections that are geographically distributed. Research in the area of grid has given various ideas and solutions to address these requirements. However, nowadays the number of participants (scientists and institutions) that are involved in this kind of environment is increasing tremendously. This situation has lead to a problem of scalability. In order to overcome this problem we need a data grid model that can scale well with the increasing number of users. Peer-to-peer (P2P) is one of the architectures that is a promising scale and dynamism environment. In this paper, we present a P2P model for Scientific Data Grid that utilizes the P2P services to address the scalability problem. By using this model, we study and propose various decentralized discovery strategies that intend to address the problem of scalability. We also investigate the impact of data replication that addresses the data distribution and reliability problem for our Scientific Data Grid model on the propose discovery strategies. For the purpose of this study, we have developed and used our own data grid simulation written using PARSEC. We illustrate our P2P Scientific Data Grid model and our data grid simulation used in this study. We then analyze the performance of the discovery strategies with and without the existence of replication strategies relative to their success rates, bandwidth consumption and average number of hop.

Ultra-fast Carrier Transport Simulation on the Grid. Quasi-Random Approach

The problem for simulation ultra-fast carrier transport in nano-electronics devices is a large scale computational problem and requires HPC and/or Grid computing resources. The most widely used techniques for modeling this carrier transport are Monte Carlo methods. In this work we consider a set of stochastic algorithms for solving quantum kinetic equations describing quantum effects during the femtosecond relaxation process due to electron-phonon interaction in one-band semiconductors or quantum wires. The algorithms are integrated in a Grid-enabled package named S tochas-tic A Lgorithms for U ltra-fast T ransport in s E miconductors (SALUTE). There are two main reasons for running this package on the computational Grid: (i) quantum problems are very computationally intensive; (ii) the inherently parallel nature of Monte Carlo applications makes efficient use of Grid resources. Grid (distributed) Monte Carlo applications require that the underlying random number streams in each subtask are independent in a statistical sense. The efficient application of quasi-Monte Carlo algorithms entails additional difficulties due to the possibility of job failures and the inhomogeneous nature of the Grid resource. In this paper we study the quasi-random approach in SALUTE and the performance of the corresponding algorithms on the grid, using the scrambled Halton, Sobol and Niederreiter sequences. A large number of tests have been performed on the EGEE and SEEGRID grid infrastructures using specially developed grid implementation scheme. Novel results for energy and density distribution, obtained in the inhomogeneous case with applied electric field are presented.

New image segmentation method based on gradient

To solve the problems of large computation,slow convergence,and easily falling into local extrimum of traditional image segmentation algorithms,a new image segmentation method based on color image gradient calculation was proposed. The experimental results on corn and Lena images,compared with level set method and watershed method,show that the new method can segment images better,faster and steadier,and suits for both grey images and color images.

Parallel workflows for data-driven structural equation modeling in functional neuroimaging

We present a computational framework suitable for a data-driven approach to structural equation modeling (SEM) and describe several workflows for modeling functional magnetic resonance imaging (fMRI) data within this framework. The Computational Neuroscience Applications Research Infrastructure (CNARI) employs a high-level scripting language called Swift, which is capable of spawning hundreds of thousands of simultaneous R processes (R Development Core Team, 2008), consisting of self-contained SEMs, on a high performance computing system (HPC). These self-contained R processing jobs are data objects generated by OpenMx, a plug-in for R, which can generate a single model object containing the matrices and algebraic information necessary to estimate parameters of the model. With such an infrastructure in place a structural modeler may begin to investigate exhaustive searches of the model space. Specific applications of the infrastructure, statistics related to model fit, and limitations are discussed in relation to exhaustive SEM. In particular, we discuss how workflow management techniques can help to solve large computational problems in neuroimaging.

Application resource requirement estimation in a parallel-pipeline model of execution

We propose a massively parallel framework termed a parallel-pipeline model of execution that can be employed on a homogeneous computational cluster. We show that speedups near-linear in the number of processors are achievable for applications involving reduction operations based on a novel, parallel-pipeline model of execution. As computational clusters become viable alternative platforms for solving large computational problems, the research community acknowledges that the cluster environment can be used effectively when adaptive resource management is employed. This requires the ability to estimate the resource requirements of applications before scheduling decisions are made. We propose a resource estimation model for applications executed in the parallel-pipeline model of execution. We develop a performance model that predicts the resource utilization (i.e., computation and communication complexity) for applications executing under the parallel-pipeline model on a homogeneous computational cluster. This performance prediction model can provide information to a cluster scheduler.

Towards Rapid Redesign: Pattern-based Redesign Planning for Large-Scale and Complex Redesign Problems

We have developed a decomposition-based rapid redesign methodology for large and complex computational redesign problems. While the overall methodology consists of two general steps: diagnosis and repair, in this paper we focus on the repair step in which decomposition patterns are utilized for redesign planning. Resulting from design diagnosis, a typical decomposition pattern solution to a given redesign problem indicates the portions of the design model necessary for recomputation as well as the interaction part within the model accountable for design change propagation. Following this, in this paper we suggest repair actions with an approach derived from an input pattern solution, to generate a redesign road map allowing for taking a shortcut in the redesign solution process. To do so, a two-stage redesign planning approach from recomputation strategy selection to redesign road map generation is proposed. An example problem concerning the redesign of a relief valve is used for illustration and validation.

Data discovery algorithm for scientific data grid environment

In modern scientific computing communities, scientists are involved in managing massive amounts of very large data collections in a geographically distributed environment. Research in the area of grid computing has given us various ideas and solutions to address these requirements. Data grid mostly deals with large computational problems and provides geographically distributed resources for large-scale data-intensive applications that generate large data sets. Peer-to-peer (P2P) networks have also become a major research topic over the last few years. In a distributed P2P system, a discovery algorithm is required to locate specific information, applications, or users within the system. In this research work, we present our scientific data grid as a large P2P-based distributed system model. By using this model, we study various discovery algorithms for locating data sets in a data grid system. The algorithms we studied are based on the P2P architecture. We investigate these algorithms using our Grid Simulator developed using PARSEC. In this paper, we illustrate our scientific data grid model and our Grid Simulator. We then analyze the performance of the discovery algorithms relative to their average number of hop, success rates and bandwidth consumption.

Collaboration Online: The Example of Distributed Computing

Distributed Computing is a new form of online collaboration; such projects divide a large computational problem into small tasks that are sent out over the Internet to be completed on personal computers. Millions of people all over the world participate voluntarily in such projects, providing computing resources that would otherwise cost millions of dollars. However, Distributed Computing only works if many people participate. The technical challenge is to slice a problem into thousands of tiny pieces that can be solved independently, and then to reassemble the solutions. The social problem is how to find all those widely dispersed computers and persuade their owners to participate. This article examines what makes a collaborative Distributed Computing project successful. We report on data from a quantitative survey and a qualitative study of participants on several online forums, and discuss and analyze Distributed Computing using Arquilla and Ronfeldt's (2001) five-level network organization framework.

Efficient rendering of multiblock curvilinear grids with complex boundaries

Domain decomposition is a popular technique for solving large computational problems that require data to be divided into smaller sub-domains. The exact manner of decomposition depends on the computational needs of the algorithm and often introduces irregular boundaries. Each subdomain forms a block of a larger grid and can be solved or rendered separately by different processing nodes. Rendering of each sub-domain can result in images which are then composited in a back-to-front or front-to-back manner. This scenario is useful when visualization is used concurrently with the simulation. However, the irregularity of boundaries may prohibit the correct image composition due to a visibility anomaly between the sub-domains. In this paper, we present an algorithm based on object-space partitioning to resolve this problem. To accelerate the partitioning process, two techniques are introduced. First, an image-space partition representation is employed for fast assignment of data points to correct partitions. Secondly, a k-d tree is used to subdivide the view-space adaptively according to the complexity of the surface. This view-space partition provides a trade-off between performance and accuracy of the rendered image. Large gains in performance can be achieved with only small losses of accuracy. Two examples of curvilinear grids of different complexity are used to demonstrate the effectiveness of this scheme. Copyright © 2005 John Wiley & Sons, Ltd.

High performance horizons [high performance computing

The field of high performance computing is undergoing a lot of changes. At the low end, supercomputers are becoming ever more affordable while, at the high end, it is becoming increasingly difficult to exploit the full potential of the most advanced hardware, given the sheer scale of the software challenges involved. There are also signs that the boundary between the worlds of the supercomputer and the desktop - traditionally very distinct - is starting to blur. Supercomputers continue to scale, with performance advancing at a rate that exceeds the rate of chip-level improvements, as characterized by Moore's law. These advances largely stem from the dramatic growth in cluster technology, which allows many processors to be closely coupled to solve large computational problems.

Computational challenges in the numerical treatment of large air pollution models

Air pollution, especially the reduction of the air pollution to some acceptable levels, is an important environmental problem, which will become even more important in the next 10–20 years. This problem can successfully be studied when high-resolution comprehensive models are developed and used on a routine basis. However, such models are very time-consuming, even when modern high-speed computers are available. Indeed, if an air pollution model is to be applied on a large space domain by using fine grids, then its discretization will always lead to huge computational problems. Assume, for example, that the space domain is discretized by using a (480 × 480) grid and that the number of chemical species studied by the model is 35. Then several systems of ordinary differential equations containing 8 064 000 equations have to be treated at every time-step (the number of time-steps being typically several thousand). If a three-dimensional version of the same air pollution model is to be used, then the figure above must be multiplied by the number of layers. It is extremely difficult to treat such large computational problems, even when the fastest computers that are available at present are used. There is an additional great difficulty, which is very often underestimated (or even neglected) when large application packages are moved from sequential computers to modern parallel machines. The high-speed computers have normally a very complicated memory architecture and, therefore, the task of producing an efficient code for the particular high-speed computer that is available is both extremely hard and time consuming. The use of standard parallelization tools in the solution of the problems sketched above is discussed in this paper. Numerical results obtained on different types of parallel computers are presented.

Developing SPPVM modules with visual basic

The development of true Parallel Machines of which literatures are scanty is still algorithmitic. Since the need for increasing power in computations is the wish of every programmer or designer, PVMs become more of a choice. PVMs enable large computational problems to be solved more cost effectively by using aggregated power and memory of many computers. This paper develops a SPPVM (Single Processor Parallel Virtual Machine) with Visual Basic. It employs the VB Shell command to cause a single program or instruction to execute in two different shells in same memory space. A quick sort program is implemented on two shells. Each shell sorts 500 elements. The machine used for the execution is: Intel (r) Celeron ™ Processor – Genuine Intel ~ 600 Mhz, with Total Physical Mem. = 30.45MB, Available Physical Mem. = 232KB, Total Virtual Mem. = 2GB, Available Virtual Mem. = 1.90GB, Page File space = 1.97GB at runtime. Result comparison with a strictly sequential version reveals that the SPPVM executes in less time than ½ x Tseq (thetheoretical value for two processes in parallel). The key benefit of this paper is to enable programmers explore parallel programming features on micro systems and develop their own SPPVM. Keywords: Shell function, Self-diagnostic, Multi processing, Multi-Platform, Parallel Processing. [Global Jnl Mathematical Sci Vol.2(1) 2003: 29-36]

Optorsim: A Grid Simulator for Studying Dynamic Data Replication Strategies

Computational grids process large, computationally intensive problems on small data sets. In contrast, data grids process large computational problems that in turn require evaluating, mining and producing large amounts of data. Replication, creating geographically disparate identical copies of data, is regarded as one of the major optimization techniques for reducing data access costs. In this paper, several replication algorithms are discussed. These algorithms were studied using the Grid simulator: OptorSim. OptorSim provides a modular framework within which optimization strategies can be studied under different Grid configurations. The goal is to explore the stability and transient behaviour of selected optimization techniques. We detail the design and implementation of OptorSim and analyze various replication algorithms based on different Grid workloads.

Necessity is the mother of invention: a simple grid computing system using commodity tools

Access to sufficient resources is a barrier to scientific progress for many researchers facing large computational problems. Gaining access to large-scale resources (i.e., university-wide or federally supported computer centers) can be difficult, given their limited availability, particular architectures, and request/review/approval cycles. Simultaneously, researchers often find themselves with access to workstations and older clusters overlooked by their owners in favor of newer hardware. Software to tie these resources into a coherent Grid, however, has been problematic. Here, we describe our experiences building a Grid computing system to conduct a large-scale simulation study using “borrowed” computing resources distributed over a wide area. Using standard software components, we have produced a Grid computing system capable of coupling several hundred processors spanning multiple continents and administrative domains. We believe that this system fills an important niche between a closely coupled local system and a heavyweight, highly customized wide area system.