We gratefully acknowledge support from
the Simons Foundation and member institutions.

Numerical Analysis

New submissions

[ total of 25 entries: 1-25 ]
[ showing up to 2000 entries per page: fewer | more ]

New submissions for Thu, 9 Jul 20

[1]  arXiv:2007.03729 [pdf, ps, other]
Title: Robust and effective eSIF preconditioning for general SPD matrices
Authors: Jianlin Xia
Subjects: Numerical Analysis (math.NA)

We propose an unconditionally robust and highly effective preconditioner for general symmetric positive definite (SPD) matrices based on structured incomplete factorization (SIF), called enhanced SIF (eSIF) preconditioner. The original SIF strategy proposed recently derives a structured preconditioner by applying block diagonal preprocessing to the matrix and then compressing appropriate scaled off-diagonal blocks. Here, we use an enhanced scaling-and-compression strategy to design the new eSIF preconditioner. Some subtle modifications are made, such as the use of two-sided block triangular preprocessing. A practical multilevel eSIF scheme is then designed. We give rigorous analysis for both the enhanced scaling-and-compression strategy and the multilevel eSIF preconditioner. The new eSIF framework has some significant advantages and overcomes some major limitations of the SIF strategy. (i) With the same tolerance for compressing the off-diagonal blocks, the eSIF preconditioner can approximate the original matrix to a much higher accuracy. (ii) The new preconditioner leads to much more significant reductions of condition numbers due to an accelerated magnification effect for the decay in the singular values of the scaled off-diagonal blocks. (iii) With the new preconditioner, the eigenvalues of the preconditioned matrix are much better clustered around $1$. (iv) The multilevel eSIF preconditioner is further unconditionally robust or is guaranteed to be positive definite without the need of extra stabilization, while the multilevel SIF preconditioner has a strict requirement in order to preserve positive definiteness. Comprehensive numerical tests are used to show the advantages of the eSIF preconditioner in accelerating the convergence of iterative solutions.

[2]  arXiv:2007.03801 [pdf, other]
Title: A variable timestepping algorithm for the unsteady Stokes/Darcy model
Subjects: Numerical Analysis (math.NA)

This report considers a variable step time discretization algorithm proposed by Dahlquist, Liniger and Nevanlinna and applies the algorithm to the unsteady Stokes/Darcy model. Although long-time forgotten and little explored, the algorithm performs advantages in variable timestep analysis of various fluid flow systems, including the coupled Stokes/Darcy model. The paper proves that the approximate solutions to the unsteady Stokes/Darcy model are unconditionally stable due to the G-stability of the algorithm. Also variable time stepping error analysis follows from the combination of G-stability and consistency of the algorithm. Numerical experiments further verify the theoretical results, demonstrating the accuracy and stability of the algorithm for time-dependent Stokes/Darcy model.

[3]  arXiv:2007.03846 [pdf, other]
Title: Fully discrete loosely coupled Robin-Robin scheme for incompressible fluid-structure interaction: stability and error analysis
Subjects: Numerical Analysis (math.NA)

We consider a fully discrete loosely coupled scheme for incompressible fluid-structure interaction based on the time semi-discrete splitting method introduced in {\emph{[Burman, Durst \& Guzm\'an, arXiv:1911.06760]}}. The splittling method uses a Robin-Robin type coupling that allows for a segregated solution of the solid and the fluid systems, without inner iterations. For the discretisation in space we consider piecewise affine continuous finite elements for all the fields and ensure the inf-sup condition by using a Brezzi-Pitk\"aranta type pressure stabilization. The interfacial fluid-stresses are evaluated in a variationally consistent fashion, that is shown to admit an equivalent Lagrange multiplier formulation. We prove that the method is unconditionally stable and robust with respect to the amount of added-mass in the system. Furthermore, we provide an error estimate that shows the error in the natural energy norm for the system is $\mathcal O\big(\sqrt{T}(\sqrt{\Delta t} + h)\big)$ where $T$ is the final time, $\Delta t$ the time-step length and $h$ the space discretization parameter.

[4]  arXiv:2007.04012 [pdf, other]
Title: A pressure-robust discretization of Oseen's equation using stabilization in the vorticity equation
Subjects: Numerical Analysis (math.NA)

Discretization of Navier-Stokes' equations using pressure-robust finite element methods is considered for the high Reynolds number regime. To counter oscillations due to dominating convection we add a stabilization based on a bulk term in the form of a residual-based least squares stabilization of the vorticity equation supplemented by a penalty term on (certain components of) the gradient jump over the elements faces. Since the stabilization is based on the vorticity equation, it is independent of the pressure gradients, which makes it pressure-robust. Thus, we prove pressure-independent error estimates in the linearized case, known as Oseen's problem. In fact, we prove an $O(h^{k+\frac12})$ error estimate in the $L^2$-norm that is known to be the best that can be expected for this type of problem. Numerical examples are provided that, in addition to confirming the theoretical results, show that the present method compares favorably to the classical residual-based SUPG stabilization.

[5]  arXiv:2007.04147 [pdf, other]
Title: Error estimates of hybridizable interior penalty methods using a variable penalty for highly anisotropic diffusion problems
Subjects: Numerical Analysis (math.NA)

In this paper, we derive improved a priori error estimates for families of hybridizable interior penalty discontinuous Galerkin (H-IP) methods using a variable penalty for second-order elliptic problems. The strategy is to use a penalization function of the form $\mathcal{O}(1/h^{1+\delta})$, where $h$ denotes the mesh size and $\delta$ is a user-dependent parameter. We then quantify its direct impact on the convergence analysis, namely, the (strong) consistency, discrete coercivity and boundedness (with $h^{\delta}$-dependency), and we derive updated error estimates for both discrete energy- and $L^{2}$-norms. All theoretical results are supported by numerical evidence.

[6]  arXiv:2007.04170 [pdf, other]
Title: The Multivariate Theory of Functional Connections: Theory, Proofs, and Application in Partial Differential Equations
Comments: 33 pages, 6 figures, 14 tables
Subjects: Numerical Analysis (math.NA); Analysis of PDEs (math.AP)

This article presents a reformulation of the Theory of Functional Connections: a general methodology for functional interpolation that can embed a set of user-specified linear constraints. The reformulation presented in this paper exploits the underlying functional structure presented in the seminal paper on the Theory of Functional Connections to ease the derivation of these interpolating functionals -- called constrained expressions -- and provides rigorous terminology that lends itself to straightforward derivations of mathematical proofs regarding the properties of these constrained expressions. Furthermore, the extension of the technique to and proofs in $n$-dimensions is immediate through a recursive application of the univariate formulation. In all, the results of this reformulation are compared to prior work to highlight the novelty and mathematical convenience of using this approach. Finally, the methodology presented in this paper is applied to two partial differential equations with different boundary conditions, and, when data is available, the results are compared to state-of-the-art methods.

[7]  arXiv:2007.04225 [pdf, other]
Title: Commutator-free Lie group methods with minimum storage requirements and reuse of exponentials
Authors: Alexei Bazavov
Comments: 40 pages, 10 figures
Subjects: Numerical Analysis (math.NA); Computational Physics (physics.comp-ph)

A new format for commutator-free Lie group methods is proposed based on explicit classical Runge-Kutta schemes. In this format exponentials are reused at every stage and the storage is required only for two quantities: the right hand side of the differential equation evaluated at a given Runge-Kutta stage and the function value updated at the same stage. The next stage of the scheme is able to overwrite these values. The result is proven for a 3-stage third order method and a conjecture for higher order methods is formulated. Five numerical examples are provided in support of the conjecture.

[8]  arXiv:2007.04244 [pdf, other]
Title: NPLIC: A Machine Learning Approach to Piecewise Linear Interface Construction
Subjects: Numerical Analysis (math.NA); Machine Learning (cs.LG); Computational Physics (physics.comp-ph); Fluid Dynamics (physics.flu-dyn)

Volume of fluid (VOF) methods are extensively used to track fluid interfaces in numerical simulations, and many VOF algorithms require that the interface be reconstructed geometrically. For this purpose, the Piecewise Linear Interface Construction (PLIC) technique is most frequently used, which for reasons of geometric complexity can be slow and difficult to implement. Here, we propose an alternative neural network-based method called NPLIC to perform PLIC calculations. The model is trained on a large synthetic dataset of PLIC solutions for square, cubic, triangular, and tetrahedral meshes. We show that this data-driven approach results in accurate calculations at a fraction of the usual computational cost.

Cross-lists for Thu, 9 Jul 20

[9]  arXiv:2007.03793 (cross-list from math.AP) [pdf, other]
Title: Approximation of surface diffusion flow: a second order variational Cahn--Hilliard model with degenerate mobilities
Subjects: Analysis of PDEs (math.AP); Numerical Analysis (math.NA)

This paper tackles the approximation of surface diffusion flow using a Cahn--Hilliard-type model. We introduce and analyze a new second order variational phase field model which associates the classical Cahn--Hilliard energy with two degenerate mobilities. This association allows to gain an order of approximation of the sharp limit. In a second part, we propose some simple and efficient numerical schemes to approximate the solutions, and we provide numerical 2D and 3D experiments that illustrate the interest of our model in comparison with other Cahn--Hilliard models.

[10]  arXiv:2007.03800 (cross-list from cs.LG) [pdf, ps, other]
Title: Efficient and Parallel Separable Dictionary Learning
Subjects: Machine Learning (cs.LG); Image and Video Processing (eess.IV); Numerical Analysis (math.NA); Machine Learning (stat.ML)

Separable, or Kronecker product, dictionaries provide natural decompositions for 2D signals, such as images. In this paper, we describe an algorithm to learn such dictionaries which is highly parallelizable and which reaches sparse representations competitive with the previous state of the art dictionary learning algorithms from the literature. We highlight the performance of the proposed method to sparsely represent image data and for image denoising applications.

[11]  arXiv:2007.03830 (cross-list from math.OC) [pdf, ps, other]
Title: Computational Semi-Discrete Optimal Transport with General Storage Fees
Authors: Mohit Bansil
Comments: 23 pages, comments welcome!
Subjects: Optimization and Control (math.OC); Analysis of PDEs (math.AP); Numerical Analysis (math.NA)

We propose and analyze a modified damped Newton algorithm to solve the semi-discrete optimal transport with storage fees. We prove global linear convergence for a wide range of storage fee functions, the main assumption being that each warehouse's storage costs are independent. We show that if $F$ is an arbitrary storage fee function that satisfies this independence condition then $F$ can be perturbed into a new storage fee function so that our algorithm converges. We also show that the optimizers are stable under these perturbations. Furthermore, our results come with quantitative rates.

[12]  arXiv:2007.04019 (cross-list from physics.soc-ph) [pdf, other]
Title: Hyperbolic models for the spread of epidemics on networks: kinetic description and numerical methods
Subjects: Physics and Society (physics.soc-ph); Numerical Analysis (math.NA)

We consider the development of hyperbolic transport models for the propagation in space of an epidemic phenomenon described by a classical compartmental dynamics. The model is based on a kinetic description at discrete velocities of the spatial movement and interactions of a population of susceptible, infected and recovered individuals. Thanks to this, the unphysical feature of instantaneous diffusive effects, which is typical of parabolic models, is removed. In particular, we formally show how such reaction-diffusion models are recovered in an appropriate diffusive limit. The kinetic transport model is therefore considered within a spatial network, characterizing different places such as villages, cities, countries, etc. The transmission conditions in the nodes are analyzed and defined. Finally, the model is solved numerically on the network through a finite-volume IMEX method able to maintain the consistency with the diffusive limit without restrictions due to the scaling parameters. Several numerical tests for simple epidemic network structures are reported and confirm the ability of the model to correctly describe the spread of an epidemic.

Replacements for Thu, 9 Jul 20

[13]  arXiv:1902.07029 (replaced) [pdf, other]
Title: A level-set multigrid technique for nonlinear diffusion in the numerical simulation of marble degradation under chemical pollutants
Subjects: Numerical Analysis (math.NA)
[14]  arXiv:1906.01687 (replaced) [pdf, other]
Title: Stochastic Gradients for Large-Scale Tensor Decomposition
Subjects: Numerical Analysis (math.NA); Machine Learning (cs.LG); Machine Learning (stat.ML)
[15]  arXiv:1910.06524 (replaced) [pdf, ps, other]
Title: Adjoint-based exact Hessian computation
Subjects: Numerical Analysis (math.NA)
[16]  arXiv:1910.14067 (replaced) [pdf, other]
Title: Spectral properties of kernel matrices in the flat limit
Comments: 40 pages, 8 pages
Subjects: Numerical Analysis (math.NA); Spectral Theory (math.SP); Statistics Theory (math.ST)
[17]  arXiv:1911.06792 (replaced) [pdf, other]
Title: On differentiable local bounds preserving stabilization for Euler equations
Comments: arXiv admin note: text overlap with arXiv:1912.11487
Subjects: Numerical Analysis (math.NA)
[18]  arXiv:1912.13291 (replaced) [pdf, other]
Title: A doubly stochastic block Gauss-Seidel algorithm for solving linear equations
Authors: Kui Du, Xiaohui Sun
Comments: 14 pages, 2 figures. More general results are given in this version
Subjects: Numerical Analysis (math.NA)
[19]  arXiv:2001.04179 (replaced) [pdf, other]
Title: Randomized extended block Kaczmarz for solving least squares
Comments: 20 pages, 3 figures, more general results are presented
Subjects: Numerical Analysis (math.NA)
[20]  arXiv:2003.08706 (replaced) [pdf, other]
Title: THB-spline approximations for turbine blade design with local B-spline approximations
Subjects: Numerical Analysis (math.NA)
[21]  arXiv:2005.02938 (replaced) [pdf, other]
Title: A Residual Based A Posteriori Error Estimators for AFC Schemes for Convection-Diffusion Equations
Authors: Abhinav Jha
Subjects: Numerical Analysis (math.NA)
[22]  arXiv:2007.02910 (replaced) [pdf, other]
Title: A Weighted Randomized Kaczmarz Method for Solving Linear Systems
Subjects: Numerical Analysis (math.NA); Optimization and Control (math.OC)
[23]  arXiv:2006.07441 (replaced) [pdf, ps, other]
Title: On the optimal constants in the two-sided Stechkin inequalities
Subjects: Classical Analysis and ODEs (math.CA); Functional Analysis (math.FA); Numerical Analysis (math.NA)
[24]  arXiv:2006.13844 (replaced) [pdf, ps, other]
Title: $\mathcal{H}_2$ optimal structure-preserving model order reduction of second-order systems by iterative rational Krylov algorithm
Comments: 17 pages 12 figures
Subjects: Optimization and Control (math.OC); Systems and Control (eess.SY); Metric Geometry (math.MG); Numerical Analysis (math.NA)
[25]  arXiv:2007.03576 (replaced) [pdf, other]
Title: A Task-based Multi-shift QR/QZ Algorithm with Aggressive Early Deflation
Authors: Mirko Myllykoski
Comments: 34 pages, 19 figures, 8 tables. Minor corrections to text and figures. Submitted to TOMS
Subjects: Mathematical Software (cs.MS); Distributed, Parallel, and Cluster Computing (cs.DC); Numerical Analysis (math.NA)
[ total of 25 entries: 1-25 ]
[ showing up to 2000 entries per page: fewer | more ]

Disable MathJax (What is MathJax?)

Links to: arXiv, form interface, find, math, recent, 2007, contact, help  (Access key information)