检索结果-内蒙古大学图书馆

On reduced input-output dynamic mode decomposition

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Benner, Peter Himpe, Christian Mitchell, Tim Computational Methods in Systems and Control Theory Max Planck Institute for Dynamics of Complex Technical Systems Sandtorstr. 1 Magdeburg39106 Germany

The identification of reduced-order models from high-dimensional data is a challenging task, and even more so if the identified system should not only be suitable for a certain data set, but generally approximate the input-output behavior of the data source. In this work, we consider the input-output dynamic mode decomposition method for system identification. We compare excitation approaches for the data-driven identification process and describe an optimization-based stabilization strategy for the identified *** Codes 93B30, 90C99 Copyright © 2017, The Authors. All rights reserved.

关键词： Clustering algorithms

Low-rank computation of posterior covariance matrices in Bayesian inverse problems

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Benner, Peter Qiu, Yue Stoll, Martin Computational Methods in Systems and Control Theory Group Max Planck Institute for Dynamics of Complex Technical Systems Sandtorstr. 1 Magdeburg39106 Germany Numerical Linear Algebra for Dynamical Systems Group Max Planck Institute for Dynamics of Complex Technical Systems Sandtorstr. 1 Magdeburg39106 Germany

We consider the problem of estimating the uncertainty in statistical inverse problems using Bayesian inference. When the probability density of the noise and the prior are Gaussian, the solution of such a statistical inverse problem is also Gaussian. Therefore, the underlying solution is characterized by the mean and covariance matrix of the posterior probability density. However, the covariance matrix of the posterior probability density is full and large. Hence, the computation of such a matrix is impossible for large dimensional parameter spaces. It is shown that for many ill-posed problems, the Hessian matrix of the data misfit part has low numerical rank and it is therefore possible to perform a low-rank approach to approximate the posterior covariance matrix. For such a low-rank approximation, one needs to solve a forward partial differential equation (PDE) and the adjoint PDE in both space and time. This in turn gives O(nxnt) complexity for both, computation and storage, where nx is the dimension of the spatial domain and nt is the dimension of the time domain. Such computations and storage demand are infeasible for large problems. To overcome this obstacle, we develop a new approach that utilizes a recently developed low-rank in time algorithm together with the low-rank Hessian method. We reduce both the computational complexity and storage requirement from O(nxnt) to O(nx + nt). We use numerical experiments to illustrate the advantages of our approach. Copyright © 2017, The Authors. All rights reserved.

关键词： Constrained optimization

Improving the Performance of Numerical Algorithms for the Bethe-Salpeter Eigenvalue Problem

学校读者我要写书评

暂无评论

PAMM 2018年第1期18卷

作者： Peter Benner Andreas Marek Carolin Penke Max Planck Institute for Dynamics of Complex Technical Systems Computational Methods in Systems and Control Theory Sandtorstr. 1 39106 Magdeburg Max Planck Computing and Data Facility Gießenbachstraße 2 85748 Garching

The Bethe-Salpeter eigenvalue problem arises in the computation of the electronic structure of many-body physical systems. The resulting matrix is complex, admits a certain block structure and can become extremely large. This raises the need for structure-preserving algorithms running in parallel on high performance compute clusters. In this paper we examine how a recently proposed direct method given in the BSEPACK library can be improved using eigenvalue solvers from the ELPA library. For large matrices a runtime reduction of up to 80% is achieved.

关键词：

Linear Low-Rank Parameter-Dependent Fluid-Structure Interaction Discretization in 2D

学校读者我要写书评

暂无评论

PAMM 2018年第1期18卷

作者： Roman Weinhandl Peter Benner Thomas Richter Max Planck Institute for Dynamics of Complex Technical Systems Computational Methods in Systems and Control Theory Sandtorstr. 1 39106 Magdeburg Otto von Guericke University Magdeburg Faculty of Mathematics Universitätsplatz 2 39106 Magdeburg

Fluid-structure interaction problems involve material parameters such as the shear modulus of the solid and the dynamic fluid viscosity. In order to examine the behaviors of various solids and adapt the problems to different fluid configurations there is a need to vary such material parameters flexibly. This can be done by a parameter-dependent fluid-structure interaction discretization which yields a system matrix that has block diagonal structure. As also discussed in [2], the resulting equation is equivalent to a matrix equation which allows for a low-rank approach where the iterate is represented by a tensor. A low-rank GMRES variant similar to what was discussed in [1] can then be applied to such parameter-dependent systems.

关键词：

Fast Low-Rank Empirical Cross Gramians

学校读者我要写书评

暂无评论

PAMM 2018年第1期17卷

作者： Christian Himpe Tobias Leibner Stephan Rave Jens Saak Computational Methods in Systems and Control Theory Max Planck Institute for Dynamics of Complex Technical Systems Sandtorstr. 1 39106 Magdeburg Germany Applied Mathematics University of Münster Einsteinstr. 62 48149 Münster Germany

The cross Gramian matrix encodes the input-output coherence of linear control systems and is used in projection-based model reduction. The empirical cross Gramian is a data-driven variant of the cross Gramian which also extends to nonlinear systems. A drawback of the empirical cross Gramian for large-scale systems is its full order and dense structure; yet, it may be computed column-wise. Using the hierarchical approximate proper orthogonal decomposition (HAPOD), this partial computability can be exploited to obtain a truncated projection for model order reduction without assembling a full cross Gramian. (© 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)

关键词：

Numerical solution to low rank perturbed lyapunov equations by the sign function method

学校读者我要写书评

暂无评论

Numerical solution to low rank perturbed lyapunov equations ...

Joint 87th Annual Meeting on the International Association of Applied Mathematics and Mechanics, GAMM 2016 and Deutsche Mathematiker‐Vereinigung, DMV 2016

作者： Benner, Peter Denißen, Jonas Max Planck Institute for Dynamics of Complex Technical Systems Computational Methods in Systems and Control Theory Sandtorstr. 1 Magdeburg39106 Germany

We investigate the numerical solution to a low rank perturbed Lyapunov equation AT X + XA = W via the sign function method (SFM). The sign function method has been proposed to solve Lyapunov equations, see e.g. [1], but here we focus on a framework where the matrix A has a special structure, i.e. A = B + UCVT, where B is a blockdiagonal matrix and UCVT is a low rank perturbation. We show that this structure can be kept throughout the sign function iteration but the rank of the perturbation doubles per iteration. Therefore, we apply a low rank approximation to the perturbation in order to keep its numerical rank small. We compare the standard SFM with its structure preserving variant presented in this paper by means of numerical examples from viscously damped mechanical systems. © 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

关键词： Matrix algebra

Adjoint-based optimal boundary control of a stefan problem system fully coupled with navier-stokes equations

学校读者我要写书评

暂无评论

Adjoint-based optimal boundary control of a stefan problem s...

Joint 87th Annual Meeting on the International Association of Applied Mathematics and Mechanics, GAMM 2016 and Deutsche Mathematiker‐Vereinigung, DMV 2016

作者： Baran, Björn Heiland, Jan Computational Methods in Systems and Control Theory Max Planck Institute for Dynamics of Complex Technical Systems Sandtorstr. 1 Magdeburg39106 Germany

Free boundary and moving boundary problems, that can be used to model crystal growth or the solidification and melting of pure materials, receive growing attention in science and technology. The optimal control of these problems appear even more interesting since certain desired shapes of the boundaries improve, e.g., the material quality in the case of crystal growth. We consider the so called two-phase Stefan problem that models a solid and a liquid phase separated by a moving interface. In the work presented, we take a sharp interface model approach and define a quadratic tracking-type cost functional that penalizes the deviation of the interface from the desired state at a final time as well as the control costs. Following the "optimize-then-discretize" paradigm, we formulate a first order optimality system using the formal Lagrange approach and derive the adjoint PDE system that provides the needed gradient of the cost functional. By means of an example setup of a container with an in-and outflow of water and a cooling unit at the bottom, we illustrate how the derived formulations can be used to achieve a desired interface between the solid and the fluid phase by controlling the flow at the inlet. © 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

关键词： Crystal growth

An inexact Newton-Krylov method for stochastic eigenvalue problems

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Benner, Peter Onwunta, Akwum Stoll, Martin Computational Methods in Systems and Control Theory Max Planck Institute for Dynamics of Complex Technical Systems Sandtorstrasse 1 Magdeburg39106 Germany Numerical Linear Algebra for Dynamical Systems Group Max Planck Institute for Dynamics of Complex Technical Systems Sandtorstrasse 1 Magdeburg39106 Germany Technische Universität Chemnitz Faculty of Mathematics Professorship Scientific Computing Chemnitz09107 Germany

This paper aims at the efficient numerical solution of stochastic eigenvalue problems. Such problems often lead to prohibitively high dimensional systems with tensor product structure when discretized with the stochastic Galerkin method. Here, we exploit this inherent tensor product structure to develop a globalized low-rank inexact Newton method with which we tackle the stochastic eigenproblem. We illustrate the effectiveness of our solver with numerical experiments. Copyright © 2017, The Authors. All rights reserved.

关键词： Galerkin methods

Solving optimal control problems governed by random Navier-Stokes equations using low-rank methods

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Benner, Peter Dolgov, Sergey Onwunta, Akwum Stoll, Martin University of Bath Department of Mathematical Sciences Claverton Down BathBA2 7AY United Kingdom Computational Methods in Systems and Control Theory Max Planck Institute for Dynamics of Complex Technical Systems Sandtorstr. 1 Magdeburg39106 Germany

Many problems in computational science and engineering are simultaneously characterized by the following challenging issues: uncertainty, nonlinearity, nonstationarity and high dimensionality. Existing numerical techniques for such models would typically require considerable computational and storage resources. This is the case, for instance, for an optimization problem governed by time-dependent Navier-Stokes equations with uncertain inputs. In particular, the stochastic Galerkin finite element method often leads to a prohibitively high dimensional saddle-point system with tensor product structure. In this paper, we approximate the solution by the low-rank Tensor Train decomposition, and present a numerically efficient algorithm to solve the optimality equations directly in the low-rank representation. We show that the solution of the vorticity minimization problem with a distributed control admits a representation with ranks that depend modestly on model and discretization parameters even for high Reynolds numbers. For lower Reynolds numbers this is also the case for a boundary control. This opens the way for a reduced-order modeling of the stochastic optimal flow control with a moderate cost at all stages. Copyright © 2017, The Authors. All rights reserved.

关键词： Stochastic systems