Ruth Cheng | Mississippi State University (original) (raw)

Papers by Ruth Cheng

This report describes a particle tracking computer program named PT123. The development of PT123 ... more This report describes a particle tracking computer program named PT123. The development of PT123 was supported in part by the Civil Works Basic Research project entitled "Efficient Resolution of Complex Transport Phenomena Using Eulerian-Lagrangian Techniques" and in part by the System-Wide Water Resources Program (SWWRP). Given velocities, PT123 can track massless particles in 1-, 2-, and 3-D unstructured or converted structured meshes. The elements used to construct PT123 meshes are line elements in 1-D, triangular and/or quadrilateral elements in 2-D, and tetrahedral, triangular prism, and/or hexahedral elements in 3-D. One adaptive (embedded 4th-and 5th-order) and three non-adaptive (1st-, 2nd-, and 4th-order) Runge-Kutta (RK) methods are included in PT123 to solve the ordinary differential equations describing the motion of particles. The adaptive RK method allows the user to control tracking accuracy with specified error tolerances. The nonadaptive RK methods provide the user options to balance computational efficiency and accuracy by using lower order schemes for smooth velocity fields and higher order schemes for complex velocity fields. Both elementby-element (EBE) and non-element-by-element (NEBE) tracking approaches are incorporated into PT123. Both node-and element-based velocity can be used for particle tracking. PT123 can execute forward and backward tracking and output tracking history at a specified frequency. It tracks particles along the closed boundary and stops tracking when a particle encounters the open boundary through which particles enter or exit the computational domain. The start and end times of tracking are flexible as long as their corresponding velocities can be computed via temporal interpolation using the given velocities. This report is the first report of the series describing the development and application of PT123. It details the governing equation and numerical approaching associated with PT123 Version 1.0. Six test examples in multiple dimensions are used for verification and demonstration. The structure and the input guide of the computer program are given in the appendices. DISCLAIMER: The contents of this report are not to be used for advertising, publication, or promotional purposes. Citation of trade names does not constitute an official endorsement or approval of the use of such commercial products. All product names and trademarks cited are the property of their respective owners. The findings of this report are not to be construed as an official Department of the Army position unless so designated by other authorized documents.

Abstract : This issue of The Resource contains the following feature articles: "An HPC-Enabl... more Abstract : This issue of The Resource contains the following feature articles: "An HPC-Enabled Virtual Proving Ground for Seismic Unattended Ground Sensor Networks," by Dr. Mark L. Moran; "SC2002 'From Terabytes to Insights,'" by Ginny Miller; "PET Highlights," by Dr. Wayne Mastin; "Building on Previous Strategies to Create a Synthetic Application of Benchmarking," by Dr. Paul M. Bennett; "Scientific Visualization Center Technology Update," by Paul Adams; "Army Science Conference -- 'Transformational Science & Technology for the Army...a race for speed and precision,'" by Paul Adams; "Technology Enhancements in the ERDC MSRC Computational Environment," by Jay Cliburn; "ERDC MSRC Prepares to Assist Users When Cray X1 Arrives," by Robert Alter, Dr. Paul M. Bennett, Dr. Ruth Cheng, Robert Hunter, Carrie Mahood, Dr. Thomas Oppe, and Dr. Fred Tracy; and "Job Shadowing in Computer-Related Fields," by Rose J. Dykes. The newsletter also contains photographs of visitors and a list of acronyms.

Explicit time-stepping methods are commonly used to solve transient, time- dependent problems. Fo... more Explicit time-stepping methods are commonly used to solve transient, time- dependent problems. For problems involving small-scale features, adaptive unstruc- tured mesh refinement can be effective. Unfortunately, using such dynamic strate- gies on parallel computers can result in difficult load balancing issues because of the relatively high cost of the mesh partitioning to the time-step computation. We present a partitioning strategy that addresses this problem and demonstrate the ef- fectiveness of the proposed method in modeling of a two-dimensional Raleigh-Taylor instability.

Lecture Notes in Computer Science, 2002

P article trackingmethods are a versatile computational technique central to thesim ulationof a w... more P article trackingmethods are a versatile computational technique central to thesim ulationof a wide range of scientiŞcapplications. In this pap er w e presen t a new parallel approach for the dynamic partitioning of particle-mesh computational systems. The approach uses a framework, the \in-element" particle tracking method, based on the assumption that particle trajectories are computed by problemdata localized to individual elements. The parallel eŞciency of suchparticle-mesh systems dep endson the partitioning of b oththe mesh elements and the particles this distribution can change dramatically b ecause of movement of the particles and adaptive reŞnement of the mesh. T oaddress this problem we i n troduce a combined load function that is a function of b oth the particle and mesh element distributions. We presen t exp eriment results that detail the p erformance of this parallel load balancing approach for a three-dimensional particle-mesh test problem on an unstructured, adaptive mesh.

Lecture Notes in Computer Science, 2003

ABSTRACT Particle tracking methods are a versatile computational technique central to the simulat... more ABSTRACT Particle tracking methods are a versatile computational technique central to the simulation of a wide range of scientific applications. In this paper we present an a posteriori error estimator for adaptive mesh refinement (AMR) using particle tracking methods. The approach uses a parallel computing framework, the “in-element” particle tracking method, based on the assumption that particle trajectories are computed by problem data localized to individual elements. Adaptive mesh refinement is used to control the mesh discretization errors along computed characteristics of the particle trajectories. Traditional a posteriori error estimators for AMR methods inherit flaws from the discrete solution of time-marching partial differential equations (PDEs)-particularly for advection/convection-dominated transport applications. To address this problem we introduce a new a posteriori error estimator based on particle tracking methods.We present experimental results that detail the performance of a parallel implementation of this particle method approach for a two-dimensional, time-marching convection-diffusion benchmark problem on an unstructured, adaptive mesh.

Parallel Algorithms and Applications, 2004

ABSTRACT Particle tracking methods are central to a wide spectrum of scientific computing applica... more ABSTRACT Particle tracking methods are central to a wide spectrum of scientific computing applications. To support such applications, this paper presents a compact software architecture that can be used to interface parallel particle tracking software to computational mesh management systems. A detailed description is presented of the in-element particle tracking framework supported by this software architecture—a framework that encompasses most particle tracking applications. The use of this parallel software architecture is illustrated through the implementation of two differential equation solvers, the forward Euler and an implicit trapezoidal method, on a distributed, unstructured, computational mesh. A design goal of this software effort has been to interface to software libraries such as Scalable Unstructured Mesh Algorithms and Applications (SUMAA3d) in addition to application codes (e.g. FEMWATER). This goal of portability is achieved through a software architecture that specifies a lightweight functional interface that maintains the full functionality required by particle–mesh methods. The use of this approach in parallel programming environments written in C and Fortran is demonstrated.

IEEE transactions on cybernetics, Jan 22, 2015

Developers often spend valuable time navigating and seeking relevant code in software maintenance... more Developers often spend valuable time navigating and seeking relevant code in software maintenance. Currently, there is a lack of theoretical foundations to guide tool design and evaluation to best shape the code base to developers. This paper contributes a unified code navigation theory in light of the optimal food-foraging principles. We further develop a novel framework for automatically assessing the foraging mechanisms in the context of program investigation. We use the framework to examine to what extent the clustering of software entities affects code foraging. Our quantitative analysis of long-lived open-source projects suggests that clustering enriches the software environment and improves foraging efficiency. Our qualitative inquiry reveals concrete insights into real developer's behavior. Our research opens the avenue toward building a new set of ecologically valid code navigation tools.

2012 Third International Workshop on Recommendation Systems for Software Engineering (RSSE), 2012

Merging parallel versions of source code is a common and essential activity during the lifespan o... more Merging parallel versions of source code is a common and essential activity during the lifespan of largescale software systems. When a non-trivial number of conflicts is detected, there is a need to support the maintainer in investigating and resolving these conflicts. In this paper, we contribute a cost-benefit approach to ranking the conflicting software entities by leveraging both structural and semantic information of the source code. We present a study by applying our approach to a legacy system developed by computational scientists. The study not only demonstrates the feasibility of our approach, but also sheds light on the future development of conflict resolution recommenders.

Lecture Notes in Computer Science, 2005

A watershed software application is designed to model a coupled system of multiple physics on mul... more A watershed software application is designed to model a coupled system of multiple physics on multiple domains. Tremendous computational resources are required to integrate the system equations on large spatial domains with multiple temporal scales among them. Supported by the Department of Defense Common High Performance Computing Software Initiative, the parallel WASH123D software development aims to efficiently simulate one aspect (i.e., soil and land) of the battlespace environment. Currently, the coupled two-dimensional overland and three-dimensional subsurface flows have been completed. Different numerical approaches are implemented to solve different components of the coupled system. The parallelization of such a complex system is developed on an IT-based approach-modular, hierarchical model construction, portable, scalable, and embedded parallel computational tools development and integration. Experimental results are presented to demonstrate the successful implementation of the parallel algorithms. Detailed profiling is also provided to show the imposed lightweight communication overhead.

World Environmental and Water Resources Congress 2007, 2007

World Water & Environmental Resources Congress 2003, 2003

This paper presents the results of the implementation of interfacing parallel particle tracking s... more This paper presents the results of the implementation of interfacing parallel particle tracking software with the FEMWATER chemical transport code. FEMWATER is a well-known three-dimensional unstructured finite element code that models subsurface flow and transport. Tracy et al. parallelized FEMWATER using METIS as the partitioner for distributed memory systems. On distributed memory systems, domain decomposition or element-by-element strategies are usually adopted for finite element parallelization, especially for unstructured mesh applications. Eulerian-Lagrangian methods (ELM) are employed to solve a variety of partial differential equations (PDEs), particularly advection/convection-dominated PDEs. The transport module in FEMWATER uses ELM in addition to the traditional finite element method. To parallelize the chemical transport code, a parallel particle tracking algorithm is required to solve for advection in the Lagrangian step. Cheng and Plassmann have developed a library of parallel particle tracking algorithms for generic scientific computing applications. Their developments were based on a framework of parallel in-element particle tracking. What this means is that only element-based operations are specified and particles are partitioned and assigned to processors by their element location within the mesh. The design goal of the particle tracking library was to provide an efficient means for legacy codes such as FEMWATER to rapidly implement parallel particle tracking capability. By using FEMWATER as a candidate code, this paper demonstrates how the use of generic parallel particle tracking libraries speeds up the development of portable parallel versions of legacy application codes.

Developments in Water Science, 2004

... A generic reaction-based biogeochemical simulator Yilin Fang ~, Steve B. Yabusaki ~, and Gour... more ... A generic reaction-based biogeochemical simulator Yilin Fang ~, Steve B. Yabusaki ~, and Gour-Tsyh Yeh u ~Pacific Northwest National Laboratory, PO Box 999 ... In: Lichtner, PC, Steefel, CT, Oelkers, EH (Eds.), Reactive Transport in Porous Media, Reviews in Mineralology 34. ...

This paper presents a study of the accuracy and performance issues involved in parallel, in-eleme... more This paper presents a study of the accuracy and performance issues involved in parallel, in-element particle tracking methods. Eulerian-Lagrangian methods (ELM) are employed to solve a variety of partial differential equations (PDEs), and is well suited to solution of advection/convection-dominated PDEs. The La- grangian step employed by these methods is used to more accurately handle the advection/convection term. A necessary feature of these methods is the use of particle tracking techniques for computing advection accurately in the Lagrangian step. The technique, "in-element" particle tracking, traces fictitious particles on an element-by-element basis. Building on the serial version of this method, in this paper we a parallel in-element particle tracking algorithm. The approach used can be generalized to other methods based on particle tracking—including applications from visualization, N-body problems, and reacting and multi-phase flows.

This report presents the development of a numerical model simulating water flow and contaminant a... more This report presents the development of a numerical model simulating water flow and contaminant and sediment transport in watershed systems of one-dimensional river/stream network, two-dimensional overland regime, and three-dimensional subsurface media. The model is composed of two modules: flow and transport. Three options are provided in modeling the flow module in river/ stream network and overland regime: the kinematic wave approach, diffusion wave approach, and dynamic wave approach. The kinematic and diffusion wave approaches are known to be numerically robust in terms of numerical convergency and stability; i.e., they can generate convergent and stable simulations over a wide range of ground surface slopes in the entire watershed. The question is the accuracy of these simulations. The kinematic wave approach usually produces accurate solutions only over the region of steep slopes. The diffusion wave approach normally gives accurate solutions over the region of mild to steep s...

Procedia Computer Science, 2012

Many transient simulations spend a significant portion of the overall runtime solving a linear sy... more Many transient simulations spend a significant portion of the overall runtime solving a linear system. A wide variety of preconditioned linear solvers have been developed to quickly and accurately solve different types of linear systems, each having options to customize the preconditioned solver for a given linear system. Transient simulations may produce significantly different linear systems as the simulation progresses due to special events occurring that make the linear systems more difficult to solve or move the model closer to a state of equilibrium with easier to solve linear systems. Machine learning algorithms provide the ability to dynamically select the preconditioned linear solver for each linear system produced by a simulation. We test both single-label and multi-label classifiers, demonstrating that multi-label classifiers achieve the best performance due to associating multiple fast linear solvers with each tested linear system. For more difficult simulations, these classifiers produce significant speedups, while for less difficult simulations these classifiers achieve performance similar to the fastest single preconditioned linear solvers. We test classifiers generated using limited attribute sets, demonstrating that we can minimize overhead while still obtaining fast, accurate simulations.

Procedia Computer Science, 2010

The Galerkin least-squares finite element method for solving the Reynolds-averaged incompressible... more The Galerkin least-squares finite element method for solving the Reynolds-averaged incompressible turbulent 3-D Navier-Stokes equations is employed to simulate a navigation lock filling system in the numerical code Adaptive Hydraulics (ADH). The linear system is solved at each nonlinear iteration within every time-step using biconjugate gradient stabilized (BiCGstab) in combination with block-Jacobi (bjacobi) preconditioners, as it failed to solve the linear system because of dramatic changes in flow velocity and pressure early in the simulation. To overcome this problem, we used the Portable Extensible Toolkit for Scientific Computation (PETSc), a numerical library that provides multiple types of linear solvers. PETSc has been incorporated into the ADH code. The ADH-PETSc interface helps to systematically investigate the best linear solver for an ADH simulation. We found that a variant, known as enhanced BiCGstab(l) in combination with the additive Schwarz method (ASM), made it possible to simulate the John Day lock filling system. The BiCGstab(l) solver improved the rate of convergence because of a more reliable update strategy for the residuals. In addition, the simulation was run with various numbers of processors. The result shows good scaling of solution time as the number of processors increases

Journal of Computational Acoustics, 2012

Through finite difference time domain (FDTD) numerical simulation, we have studied the possible o... more Through finite difference time domain (FDTD) numerical simulation, we have studied the possible observation settings to improve the cost effectiveness in time-reversal (TR) source relocation in a two-dimensional (2D) urban setting under a number of typical scenarios. All scenario studies were based on the FDTD computation of the acoustic wave field resulted from an impulse source, propagated through an artificial village composed of 15 buildings and a set of sources and receivers, a typical urban setting has been extensively analyzed in previous studies. The FDTD numerical modeling code can be executed on an off-the-shelf graphic processor unit (GPU) that increases the speed of the time-reversal calculations by a factor of 200. With this approach the computational results lead to some significant conclusions. In general, using only one non-line-of-sight (NLOS) single receiver is not enough to do a quality work to re-locate the source via time-reversal. This is particularly true when...

International Journal for Numerical Methods in Engineering, 2003

We examined, through comparison among the full-coupling (FC), operator-splitting (OS), and predic... more We examined, through comparison among the full-coupling (FC), operator-splitting (OS), and predictorcorrector (PC) techniques, the e ectiveness of using the PC technique to solve depth-averaged reactive transport equations in the shallow water domain. Our investigation has led to three major conclusions. Firstly, both the OS and PC techniques can e ciently solve reactive transport equations because the advection-di usion transport equations are solved outside the non-linear iteration loop and the reaction equations are solved node by node. However, these two techniques may risk sacriÿcing computational accuracy. Secondly, the OS or PC technique incorporated with the Lagrangian-Eulerian (LE) approach can handle boundary sources more precisely than alternatively with the conventional Eulerian (CE) approach. Thirdly, with the LE approach incorporated, the numerical results from the three techniques agreed highly with one another except when di usion became signiÿcant. In this case, the PC technique's result still matched well with the FC technique's result, but di erences between the OS and FC techniques' results arose as di usion increased. Based on this study, we recommend to apply as a ÿrst step the PC technique to solving reactive transport equations with respect to both computational e ciency and accuracy. Copyright ? 2002 John Wiley & Sons, Ltd.