Gerd Heber

Scientific data management in the coming decade

Scientific instruments and computer simulations are creating vast data stores that require new scientific methods to analyze and organize the data. Data volumes are approximately doubling each year. Since these new instruments have extraordinary precision, the data quality is also rapidly improving. Analyzing this data to find the subtle effects missed by previous studies requires algorithms that can simultaneously deal with huge datasets and that can find very subtle effects --- finding both needles in the haystack and finding very small haystacks that were undetected in previous measurements.

A geometric approach to modeling microstructurally small fatigue crack formation: I. Probabilistic simulation of constituent particle cracking in AA 7075-T651

Microstructurally small fatigue crack (MSFC) formation includes stages of incubation, nucleation and microstructurally small propagation. In AA 7075-T651, the fracture of Al7Cu2Fe constituent particles is the major incubation source. In experiments, it has been observed that only a small percentage of these Fe-bearing particles crack in a highly stressed volume. The work presented here addresses the identification of the particles prone to cracking and the prediction of particle cracking frequency, given a distribution of particles and crystallographic texture in such a volume. Three-dimensional elasto-viscoplastic finite element analyses are performed to develop a response surface for the tensile stress in the particle as a function of the strain level surrounding the particle, parent grain orientation and particle aspect ratio. A technique for estimating particle strength from fracture toughness, particle size and intrinsic flaw size is developed. Particle cracking is then determined by comparing particle stress and strength. The frequency of particle cracking is then predicted from sampling measured distributions of grain orientation, particle aspect ratio and size. Good agreement is found between the predicted frequency of particle cracking and two preliminary validation experiments. An estimate of particle cracking frequency is important for simulating the next stages of MSFC formation: inserting all particles into a microstructural model for these stages is computationally intractable and physically unnecessary.

Efficient query processing on unstructured tetrahedral meshes

Modern scientific applications such as fluid dynamics and earthquake modeling heavily depend on massive volumes of data produced by computer simulations. Such applications require new data management capabilities in order to scale to terabyte-scale data volumes. The most common way to discretize the application domain is to decompose it into pyramids, forming an unstructured tetrahedral mesh. Modern simulations generate meshes of high resolution and precision, to be queried by a visualization or analysis tool. Tetrahedral meshes are extremely flexible and therefore vital to accurately model complex geometries, but also are difficult to index. To reduce query execution time, applications either use only subsets of the data or rely on different (less flexible) structures, thereby trading accuracy for speed.This paper presents efficient indexing techniques for common spatial (point and range) on tetrahedral meshes. Because the prevailing multidimensional indexing techniques attempt to approximate the tetrahedra using simpler shapes (primarily rectangles) the query performance deteriorates significantly as a function of the meshs geometric complexity. We develop Directed Local Search (DLS), an efficient indexing algorithm based on mesh topology information that is practically insensitive to the geometric properties of meshes. We show how DLS can be easily and efficiently implemented within modern DBMS without requiring new exotic index structures and complex preprocessing. Finally, we present a new data layout approach for tetrahedral mesh datasets that provides better performance for scientific applications.compared to the traditional space filling curves. In our PostgreSQL implementation DLS reduces the number of disk page accesses by 26% to 4x, and improves the overall query execution time by 25% to 4.

On the Effects of Modeling As-Manufactured Geometry: Toward Digital Twin

A simple, nonstandardized material test specimen, which fails along one of two different likely crack paths, is considered herein. The result of deviations in geometry on the order of tenths of a millimeter, this ambiguity in crack path motivates the consideration of as-manufactured component geometry in the design, assessment, and certification of structural systems. Herein, finite element models of as-manufactured specimens are generated and subsequently analyzed to resolve the crack-path ambiguity. The consequence and benefit of such a “personalized” methodology is the prediction of a crack path for each specimen based on its as-manufactured geometry, rather than a distribution of possible specimen geometries or nominal geometry. The consideration of as-manufactured characteristics is central to the Digital Twin concept. Therefore, this work is also intended to motivate its development.

Load adaptive algorithms and implementations for the 2D discrete wavelet transform on fine-grain multithreaded architectures

In this paper we present a load adaptive parallel algorithm and implementation to compute 2D Discrete Wavelet Transform (DWT) on multithreading machines. In a 2D DWT computation, the problem sizes reduces at every decomposition level and the lengths of the emerging computation paths also vary. The parallel algorithm proposed in this paper dynamically scales itself to the varying problem size. Experimental results are reported based on the implementations of the proposed algorithm on a 2D node multithreading emulation platform, EARTH-MANNA. We show that multithreading implementations of the proposed algorithm are at least 2 times faster than the MPI based message passing implementations reported in the literature. We further show that the proposed algorithm and implementations scale linearly with respect to problem and machine sizes.

A comparison of finite element and atomistic modelling of fracture

Are the cohesive laws of interfaces sufficient for modelling fracture in polycrystals using the cohesive zone model? We examine this question by comparing a fully atomistic simulation of a silicon polycrystal with a finite element simulation with a similar overall geometry. The cohesive laws used in the finite element simulation are measured atomistically. We describe in detail how to convert the output of atomistic grain boundary fracture simulations into the piecewise linear form needed by a cohesive zone model. We discuss the effects of grain boundary microparameters (the choice of section of the interface, the translations of the grains relative to one another and the cutting plane of each lattice orientation) on the cohesive laws and polycrystal fracture. We find that the atomistic simulations fracture at lower levels of external stress, indicating that the initiation of fracture in the atomistic simulations is likely dominated by irregular atomic structures at external faces, internal edges, corners and junctions of grains. Thus, the cohesive properties of interfaces alone are not likely to be sufficient for modelling the fracture of polycrystals using continuum methods.

Parallel FEM Simulation of Crack Propagation - Challenges, Status, and Perspectives

Understanding how fractures develop in materials is crucial to many disciplines, e.g., aeronautical engineering, material sciences, and geophysics. Fast and accurate computer simulation of crack propagation in realistic 3D structures would be a valuable tool for engineers and scien tists exploring the fracture process in materials. In the following, we will describe a next generation crack propagation simulation softw are that aims to make this potential a reality.

Ordering Unstructured Meshes for Sparse Matrix Computations on Leading Parallel Systems

Computer sim ulationsof realistic applications usually require solving a set of non-linear partial differential equations (PDEs) over a finite region. The process of obtaining numerical solutions to the governing PDEs involves solving large sparse linear or eigen systems over the unstructured meshes that model the underlying physical objects. These systems are often solved iterativ ely, where the sparse matrix-vector multiply (SPMV) is the most expensive operation within each iteration. In this paper, we focus on the efficiency of SPMV using various ordering/partitioning algorithms. We examine different implementations using three leading programming paradigms and architectures. Results show that ordering greatly improves performance, and that cache reuse can be more important than reducing communication. However, a multithreaded implementation indicates that ordering and partitioning are not required on the Tera MTA to obtain an efficient and scalable SPMV.

Developing a Communication Intensive Application on the EARTH Multithreaded Architecture

This paper reports a study of sparse matrix vector multiplication on a parallel distributed memory machine called EARTH, which supports a fine-grain multithreaded program execution model on off-the-shelf processors. Such sparse computations, when parallelized without graph partitioning, have a high communication to computation ratio, and are well known to have limited scalability on traditional distributed memory machines. EARTH offers a number of features which should make it a promising architecture for this class of applications, including local synchronizations, low communication overheads, ability to overlap communication and computation, and low context-switching costs. On the NAS CG benchmark Class A inputs, we achieve linear speedups on the 20-node MANNA platform, and an absolute speedup of 79 on 120 nodes on a simulated extension. The speedup improves to 90 on 120 nodes for Class B. This is achieved without inspector/executor, graph partitioning, or any communication minimization phase, which means that similar results can be expected for adaptive problems.

Landing CG on EARTH: A Case Study of Fine-Grained Multithreading on an Evolutionary Path

We report on our work in developing a fine-grained multithreaded solution for the communication-intensive Conjugate Gradient (CG) problem. In our recent work, we developed a simple yet efficient program for sparse matrix-vector multiply on a multi-threaded system. This paper presents an effective mechanism for the reduction-broadcast phase, which is integrated with the sparse MVM, resulting in a scalable implementation of the complete CG application. Three major observations from our experiments on the EARTH multithreaded testbed are: (1) The scalability of our CG implementation is impressive, e.g., absolute speedup is 90 on 120 processors for the NAS CG class B input. (2) Our dataflow-style reduction-broadcast network based on fine-grain multithreading is twice as fast as a serial reduction scheme on the same system. (3) By slowing down the network by a factor of 2, no notable degradation of overall CG performance was observed.