The subsurface flow is usually subject to uncertain porous media structures. However, in most cases we only have partial knowledge about the porous media properties. A common approach is to model the uncertain parameters as random fields, then the expectation of Quantity of Interest(QoI) can be evaluated by the Monte Carlo method. In this study, we develop a full multigrid-multilevel Monte Carlo (FMG-MLMC) method to speed up the evaluation of random parameters effects on single-phase porous flows. In general, MLMC method applies a series of discretization with increasing resolution and computes the QoI on each of them, the success of which lies in the effective variance reduction. We exploit the similar hierarchies of MLMC and multigrid methods, and obtain the solution on coarse mesh Qcl as a byproduct of the multigrid solution on fine mesh Qfl on each level l. In the cases considered in this thesis, the computational saving is 20% theoretically. In addition, a comparison of Monte Carlo and Quasi-Monte Carlo (QMC) methods reveals a smaller estimator variance and faster convergence rate of the latter method in this study.

In this paper, we consider domain decomposition preconditioners for a system of linear algebraic equations arising from the <i>p</i>-version of the fem. We propose several multi-level preconditioners for the Dirichlet problems in the sub-domains in two and three dimensions. It is proved that the condition number of the preconditioned system is bounded by a constant independent of the polynomial degree. The proof uses interpretations of the <i>p</i>-version element stiffness matrix and mass matrix on [-1,1] as <i>h</i>-version stiffness matrix and weighted mass matrix. The analysis requires wavelet methods.

We deal with the numerical solution of the Navier-Stokes equations describing a motion of viscous compressible flows. The governing equations are discretized with the aid of discontinuous Galerkin finite element method which is based on a discontinuous piecewise polynomial approximation. The discretizations leads to a large nonlinear algebraic system. In order to solve this system efficiently, we develop the so-called p-multigrid solution strategy which employ as a projec- tion and a restriction operators the L2 -projection in the spaces of polynomial functions on each element separately. The p-multigrid technique is studied, deve- loped and implemented in the code ADGFEM. The computational performance of the method is presented.

The development of adaptive mesh refinement capabilities in the field of computational fluid dynamics is an essential tool for enabling the simulation of larger and more complex physical problems. While such techniques have been known for a long time, most simulations do not make use of them because of the lack of a robust implementation. In this work, we present recent progresses that have been made to develop adaptive mesh refinement features in Nek5000, a code based on the spectral element method. These developments are driven by the algorithmic challenges posed by future exascale supercomputers. First, we perform the study of the strong scaling of Nek5000 on three petascale machines in order to assess the scalability of the code and identify the current bottlenecks. It is found that strong scaling limit ranges between 5, 000 and 220, 000 degrees of freedom per core depending on the machine and the case. The need for synchronized and low latency communication for efficient computational fluid dynamics simulation is also confirmed. Additionally, we present how Hypre, a library for linear algebra, is used to develop a new and efficient code for performing the setup step required prior to the use of an algebraic multigrid solver for preconditioning the pressure equation in Nek5000. Finally, the main objective of this work is to develop new methods for estimating the error on a numerical solution of the Navier–Stokes equations via the resolution of an adjoint problem. These new estimators are compared to existing ones, which are based on the decay of the spectral coefficients. Then, the estimators are combined with newly implemented capabilities in Nek5000 for automatic grid refinement and adaptive mesh adaptation is carried out. The applications considered so far are steady and two-dimensional, namely the lid-driven cavity at Re = 7, 500 and the flow past a cylinder at Re = 40. The use of adaptive mesh refinement techniques makes mesh generation easier and it is shown that a similar accuracy as with a static mesh can be reached with a significant reduction in the number of degrees of freedom. / <p>QC 20171114</p>

The implementation of a hybrid spectral/finite-element discretization on the unsteady, incompressible, Navier-Stokes equations with a semi-implicit time-stepping method, an explicit treatment of the advective terms, and an implicit treatment of the pressure and viscous terms leads to an algorithm capable of calculating 3D flows over complex 2D geometries. This also results in multiple Fourier mode linear systems which must be solved at every timestep, which naturally leads to two parallelization approaches: Fourier space partitioning, where each processor individually and simultaneously solves a linear system, and physical space partitioning, where all processors collectively solve each linear system, sequentially advancing through Fourier modes. These two parallelization approaches are compared based upon computational cost using multiple solvers: direct sparse LU, smoothed aggregation AMG, and single-level ILUT preconditioned GMRES; and on two supercomputers of different memory architecture(distributed and shared memory). This study revealed Fourier space partitioning outperforms physical space partitioning in all problems analyzed, and scales more efficiently as well. These differences were more dramatic on the distributed memory platform than the shared memory platform. Another study compares the previously mentioned solvers along with one additional solver, pointwise AMG, in Fourier space partitioning without parallelization to better understand computational scaling for problems with large meshes. It was found that the direct sparse LU solver performed well in terms of computational time, scaled linearly, but had very high memory usage which scaled in a super-linear manner. The single-level ILUT preconditioned GMRES solver required the least amount of memory, which also scaled linearly, but only had acceptable performance in terms of computational time for coarse meshes. Both AMG methods scaled linearly in both memory usage and time, and were comparable to the direct sparse LU solver in terms of computational time. The results of these studies are particularly useful for implementation of this algorithm on challenging and complex flows, especially direct numerical and large-eddy simulations. Reducing computational cost allows the analysis and understanding of more flows of practical interest.

Accelerating physics-based simulations has been an evergreen topic across different scientific communities. This dissertation is devoted to this subject addressing bottlenecks in state-of-the-art approaches to the simulation of fluids of large-scale scenes, viscous threads, magnetic fluids, and the simulation of fibers and thin structures. The contributions within the thesis are rooted in mathematical modeling and numerical simulation as well as in machine learning. The first part deals with the simulation of incompressible flow in a multigrid fashion. For the variational viscous equation, geometric multigrid is inefficient. An Unsmoothed Aggregation Algebraic Multigrid method is devised with a multi-color Gauss-Seidel smoother, which consistently solves this equation in a few iterations for various material parameters. This framework is 2.0 to 14.6 times faster compared to the state-of-the-art adaptive octree solver in commercial software for the large-scale simulation of both non-viscous and viscous flow. In the second part, a new physical model is devised to accelerate the macroscopic simulation of magnetic fluids. Previous work is based on the classical Smoothed-Particle Hydrodynamics (SPH) method and a Kelvin force model. Unfortunately, this model results in a force pointing outwards causing significant levitation problems limiting the application of more advanced SPH frameworks such as Divergence-Free SPH (DFSPH) or Implicit Incompressible SPH (IISPH). This shortcoming has been addressed with this new current loop magnetic force model resulting in more stable and fast simulations of magnetic fluids using DFSPH and IISPH. Following a different trajectory, the third part of this thesis aims for the acceleration of iterative solvers widely used to accurately simulate physical systems. We speedup the simulation for rod dynamics with Graph Networks by predicting the initial guesses to reduce the number of iterations for the constraint projection part of a Position-based Dynamics solver. Compared to existing methods, this approach guarantees long-term stability and therefore leads to more accurate solutions.

No description available.

The accurate and efficient numerical simulation of geophysical flows is of great interest in numerical weather prediction and climate modeling as well as in numerous critical areas and industries, such as agriculture, construction, tourism, transportation, weather-related disaster management, and sustainable energy technologies. Oceanic and atmospheric flows display an enormous range of temporal and spatial scales, from seconds to decades and from centimeters to thousands of kilometers, respectively. Scale interactions, both spatial and temporal, are the dominant feature of all aspects of general circulation models in geophysical fluid dynamics. In this thesis, to decrease the cost for these geophysical flow computations, several types of multiscale methods were systematically developed and tested for a variety of physical settings including barotropic and stratified wind-driven large scale ocean circulation models, decaying and forced two-dimensional turbulence simulations, as well as several benchmark incompressible flow problems in two and three dimensions. The new models proposed here are based on two classes of modern multiscale methods: (i) interpolation based approaches in the context of the multigrid/multiresolution methodologies, and (ii) deconvolution based spatial filtering approaches in the context of large eddy simulation techniques. In the first case, we developed a coarse-grid projection method that uses simple interpolation schemes to go between the two components of the problem, in which the solution algorithms have different levels of complexity. In the second case, the use of approximate deconvolution closure modeling strategies was implemented for large eddy simulations of large-scale turbulent geophysical flows. The numerical assessment of these approaches showed that both the coarse-grid projection and approximate deconvolution methods could represent viable tools for computing more realistic turbulent geophysical flows that provide significant increases in accuracy and computational efficiency over conventional methods. / Ph. D.

Esta tesis tiene el objetivo de estudiar aplicaciones de la transformada wavelet discreta (DWT) al álgebra lineal numérica. Se hace un estudio de las distintas variantes de paralelización de la DWT y se propone una nueva variante paralela, en memoria distribuida, con distribuciones de datos orientadas a bloques de matrices, como la 2DBC de ScaLAPACK. La idea es que la DWT en muchos casos es una operación intermedia y debe ajustarse a las distribuciones de datos que se estén usando. Se define y demuestra una forma de calcular exactamente la cantidad de elementos que debe comunicar cada procesador para que se puedan calcular de forma independiente todo los coeficientes wavelet en una cantidad de niveles determinada. Finalmente se propone una variante específica, más eficiente, para el cálculo de la DWT-2D cuando se aplica como paso previo a la resolución de un sistema de ecuaciones distribuido 2DBC, considerando una permutación de las filas y columnas del sistema que minimiza las comunicaciones. Otro de los aportes de esta tesis es el de considerar como un caso típico, el cálculo de la DWT-2D no estándar en matrices dispersas, proponemos algoritmos para realizar esta operación sin necesidad de construir explícitamente la matriz wavelet. Además tenemos en cuenta el fenómeno de rellenado (fill-in) que ocurre al aplicar la DWT a una matriz dispersa. Para ello exploramos con los métodos de reordenamiento clásicos de grado mínimo y de reducción a banda. De forma adicional sugerimos como pueden influir esos reordenamientos a la convergencia de los métodos multimalla ya que ocurre una redistribución de la norma de la matriz hacia los niveles inferiores de la representación multi-escala, lo que garantizaría una mejor compresión. El campo de aplicación de la transformada wavelet que se propone es la resolución de grandes sistemas de ecuaciones lineales. / Acevedo Martínez, L. (2009). Computación paralela de la transformada Wavelet; Aplicaciones de la transformada Wavelet al Álgebra Lineal Numérica [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/7107

Este trabalho propõe uma nova abordagem, baseada em wavelets, para o método Multigrid Algébrico (WAMG). Nesta nova abordagem, a Transformada Discreta Wavelet é aplicada na matriz de coeficientes do sistema linear gerando uma aproximação dessa matriz em cada nível do processo de multiresolução. As vantagens da nova abordagem, que incluem maior facilidade de paralelização e menor tempo de montagem, são apresentadas com detalhes e uma análise quantitativa de convergência do método WAMG é realizada a partir da sua aplicação em problemas testes. O WAMG também é testado como pré- condicionador para métodos iterativos no subespaço de Krylov na análise magnetostática e magnetodinâmica (regime permanente senoidal) pelo Método dos Elementos Finitos, e em matrizes esparsas extraidas das coleções Matrix Market e da Universidade da Flórida. São apresentados resultados numéricos comparando o WAMG com o Multigrid Algébrico tradicional e com os pré-condicionadores baseados em decomposições incompletas de Cholesky e LU. / In this work we propose a wavelet-based algebraic multigrid method (WAMG) as a linear system solver as well as a prediconditioner for Krylov subspace methods. It is a new approach for the Algebraic Multigrid method (AMG), which considers the use of Discrete Wavelet Transform (DWT) in the construction of a hierarchy of matrices. The two-dimensional DWT is applied to produce an approximation of the matrix in each level of the wavelets multiresolution decomposition process. The main advantages of this new approach are presented and a quantitative analysis of its convergence is shown after its application in some test problems. The WAMG also is tested as a preconditioner for Krylov subspace methods in problems with sparse matrices, in nonlinear magnetic field problems and in 3D time-harmonic Electromagnetic Edge-based Finite Element Analysis. Numerical results are presented comparing the WAMG with the standard Algebraic Multigrid method and with the preconditioners based on the incomplete Cholesky and LU decompositions.