Computing time investigations for variational multirate integration

Consider a mechanical system that contains slow and fast dynamics. Let it be possible, to split the potential energy into a slow and a fast potential and the configuration vector into slow and fast variables. For such systems, multirate schemes simulate the different parts using different time steps with the goal to save computing time. For the proposed multirate scheme, a time grid consisting of micro and macro nodes is used and the integrator is derived from a discrete variational principle. Variational integrators conserve properties like symplecticity and momentum maps and have good energy behavior. To solve the resulting system of coupled nonlinear equations, a Newton-Raphson iteration with an analytical Jacobian is used. It is demonstrated that the multirate approach leads to less computing time compared to singlerate simulation by means of three example systems, the Fermi-Pasta-Ulam problem, a triple spherical pendulum and a simple atomistic model, where the latter two are subject to constraints. Computing times are compared for different numbers of micro and macro nodes for dynamic simulations during a certain time interval. (© 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Limitations of computing time savings in variational multirate integrations

For systems that contain slow and fast dynamics, variational multirate integration schemes are used. These schemes split the system into parts which are simulated using a time grid consisting of micro and macro nodes. This leads to computing time savings, however not unlimited, for a certain number of micro steps per macro step the computing time is minimal. To find a relation between this minimum computing time and the number of variables in the system, the computing time for the Fermi-Pasta-Ulam problem (FPU) is measured for different numbers of masses and different numbers of micro steps. In addition, the numerical convergence of the variational multirate integration is shown for the FPU. (© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Variational multirate integration in multi-body dynamics

For systems that contain slow and fast dynamics, variational multirate integration schemes are used. These schemes split the system into parts which are simulated using two time grids consisting of micro and macro nodes. This formulation can be extended for multi-body systems. The rigid multi-body system is described by the so called director formulation and constraints describing the joints connecting the bodies. With the Lagrange multiplier method, the constraints are introduced into the equations of motion. A way to implement the null space method into the variational multirate framework is shown and the influence on the number of unknowns is investigated. (© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Multirate finite step methods for linear acoustics

Many processes contain phenomena on different time scales, leading to model equations with fast and small parts. There are several approaches to solve these equations, like additive Runge Kutta methods or multirate infinitesimal steps methods (MIS). Both methods make use of the additive splitting of the ODE in fast and small parts. The multiple infinitesimal step method integrates the slow part with a large macro stepsize, whereas the fast terms are solved with several smaller steps of a simpler method. The order conditions of a MIS method are derived under the assumption of the exact integration of the fast parts. We develop the multirate finite step methods (MFS). These methods are derived from the MIS methods, by taking a simple small scale integrator for the fast terms. This small scale integrator uses the same number of steps in each stage. With these assumptions, we derive the order conditions, such that the order is independent in the number of small steps. (© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Efficient simulation of a slow-fast dynamical system using multirate finite difference schemes

We consider a system of ordinary differential equations describing a slow-fast dynamical system, in particular, a predator-prey system that is highly susceptible to local time variations. This model exhibits coexistence of predatorprey dynamics in the case when the prey population grows much faster than that of the predators with a quite diversified time response. For particular parametric values their interactions show a stable relaxation oscillation in the positive octant. Such characteristics are di?cult to mimic using conventional time integrators that are used to solve systems of ordinary di?erential equations. To resolve this, we design and analyze multirate time integration methods to solve a mathematical model for a slow-fast dynamical system. Proposed methods are based on using extrapolation multirate discretisation algorithms. Through these methods, we reduce the integration time by integrating the slow sub-system with a larger step length than the fast sub-system. This allows us to efficiently solve multiscale ordinary differential equations. Besides theoretical results, we provide thorough numerical experiments which confirm that these multirate schemes outperform corresponding single-rate schemes substantially both in terms of computational work and CPU times.     

Structure preserving integrators for non-linear structural dynamics

A well known and major drawback of standard time integration schemes in the field of non-linear elastodynamics is their unstable behavior in the case of stiff material behaviour. Even second order accurate implicit time integration schemes are unable to resolve the problem under consideration effectively. To remedy this drawback, structure preserving integrators have been developed. Therefore, the goal of this paper is to compare recently developed integrators. In particular, an energy and momentum conserving scheme, based on a publication by Betsch & Steinmann [1], as well as a symplectic variational integrator, proposed by Lew et al. [4] and Wendlandt & Marsden [3], based on a mid-point evaluation of the discrete Lagrangian, are presented. Two representative numerical examples will outline the characteristics of the different approaches. In particular, a stiff non-linear spring pendulum and a finite element model of non-linear structural dynamics are considered. (© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Variational formulations for Vlasov–Poisson–Fokker–Planck systems

Time‐discrete variational schemes are introduced for both the Vlasov–Poisson–Fokker–Planck (VPFP) system and a natural regularization of the VPFP system. The time step in these variational schemes is governed by a certain Kantorovich functional (or scaled Wasserstein metric). The discrete variational schemes may be regarded as discretized versions of a gradient flow, or steepest descent, of the underlying free energy functionals for these systems. For the regularized VPFP system, convergence of the variational scheme is rigorously established. Copyright © 2000 John Wiley & Sons, Ltd.     

Multirate finite step methods with varying step sizes

Many partial differential equations consist of slow and fast scales. Often, the right hand side of semidiscretized PDEs can be split additively in corresponding fast and slow parts. Many methods utilise the additive splitting of these equations, like generalized additive Runge-Kutta (GARK) methods or multirate infinitesimal step methods. The latter one treat the slow part with macro step sizes, whereas the fast part is integrated a ODE solver. The corresponding order conditions assume the exact solution of the auxiliary ODE, i.e. assume an infinite number of small steps. We extend the MIS approach by fixing the number of steps, which leads to the multirate finite steps (MFS) method. The order conditions are derived, such that the order is independent in the number of small steps in each stage. Finally, we confirm the theoretical results by numerical experiments. (© 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Variational integrators of higher order for flexible multibody systems

In this paper we present a time stepping scheme which is based on a variational integrator. This higher-order time stepping scheme includes constraints and a viscoelastic material formulation. A variational integrator is structure-preserving which results from using a discrete variational principle. Therefore, a variational integrator always takes the form of discrete EULER-LAGRANGE equations or the equivalent position-momentum equations. In this framework, we consider the motion of a flexible rope with non-holonomic constraints by the LAGRANGE-multiplier technique. The time stepping scheme is derived from a space-time discretization of HAMILTON's principle. The space discretization is based on one-dimensional linear LAGRANGE polynomials, whereas the time discretization is based on higher-order polynomials and higher-order quadrature rules. (© 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Variational integrators for electric circuits

Variational integrators are symplectic-momentum preserving integrators that are based on a discrete variational formulation of the underlying system. So far, variational integrators have been mainly developed and used for a wide variety of mechanical systems. In this work, we develop a variational integrator for the simulation of electric circuits. An appropriate variational formulation is presented to model the circuit from which the equations of motion are derived. Finally, a corresponding time-discrete variational formulation provides an iteration scheme for the simulation of the electric circuit. In this way, a variational integrator is constructed that gains several advantages. A comparison to standard integration techniques shows that even for simple LCR circuits a better long-time energy behavior and frequency preservation can be obtained. (© 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

传递辛矩阵群收敛于辛Lie群

通过作用量变分原理,给出了Hamilton正则方程离散积分的传递辛矩阵表示,利用Hamilton正则方程给出了其对应的Lie代数．说明了当时间区段长度趋近于0时,离散系统积分的传递辛矩阵群收敛于连续时间Hamilton系统微分方程分析积分得到的辛Lie群．     

Numerical stability analysis of an acceleration scheme for step size constrained time integrators

Time integration schemes with a fixed time step, much smaller than the dominant slow time scales of the dynamics of the system, arise in the context of stiff ordinary differential equations or in multiscale computations, where a microscopic time-stepper is used to compute macroscopic behaviour. We discuss a method to accelerate such a time integrator by using extrapolation. This method extends the scheme developed by Sommeijer [Increasing the real stability boundary of explicit methods, Comput. Math. Appl. 19(6) (1990) 37–49], and uses similar ideas as the projective integration method. We analyse the stability properties of the method, and we illustrate its performance for a convection–diffusion problem.     

A multirate time stepping strategy for stiff ordinary differential equations

To solve ODE systems with different time scales which are localized over the components, multirate time stepping is examined. In this paper we introduce a self-adjusting multirate time stepping strategy, in which the step size for a particular component is determined by its own local temporal variation, instead of using a single step size for the whole system. We primarily consider implicit time stepping methods, suitable for stiff or mildly stiff ODEs. Numerical results with our multirate strategy are presented for several test problems. Comparisons with the corresponding single-rate schemes show that substantial gains in computational work and CPU times can be obtained. AMS subject classification (2000)  65L05, 65L06, 65L50     

Continuous and Discrete Clebsch Variational Principles

The Clebsch method provides a unifying approach for deriving variational principles for continuous and discrete dynamical systems where elements of a vector space are used to control dynamics on the cotangent bundle of a Lie group via a velocity map. This paper proves a reduction theorem which states that the canonical variables on the Lie group can be eliminated, if and only if the velocity map is a Lie algebra action, thereby producing the Euler–Poincaré (EP) equation for the vector space variables. In this case, the map from the canonical variables on the Lie group to the vector space is the standard momentum map defined using the diamond operator. We apply the Clebsch method in examples of the rotating rigid body and the incompressible Euler equations. Along the way, we explain how singular solutions of the EP equation for the diffeomorphism group (EPDiff) arise as momentum maps in the Clebsch approach. In the case of finite-dimensional Lie groups, the Clebsch variational principle is discretized to produce a variational integrator for the dynamical system. We obtain a discrete map from which the variables on the cotangent bundle of a Lie group may be eliminated to produce a discrete EP equation for elements of the vector space. We give an integrator for the rotating rigid body as an example. We also briefly discuss how to discretize infinite-dimensional Clebsch systems, so as to produce conservative numerical methods for fluid dynamics.      

Asynchronous variational Lie group integration for geometrically exact beam dynamics

For the elastodynamic simulation of a spatially discretized beam, asynchronous variational integrators (AVI) offer the possibility to use different time steps for every element [1]. They are symplectic and conserve discrete momentum maps and since the presented integrator for geometrically exact beam dynamics [2] is derived in the Lie group setting (SO(3) for the representation of rotational degrees of freedom), it intrinsically preserves the group structure without the need for constraints [3]. A decrease of computational cost is to be expected in situations, where the time steps have to be very low in certain parts of the beam but not everywhere, e.g. if some regions of the beam are moving faster than others. The implementation allows synchronous as well as asynchronous time stepping and shows very good energy behaviour, i.e. there is no drift of the total energy for conservative systems. (© 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

基于Lie群的刚体动力学建模及数值计算方法研究

基于Lie群和Lie代数之间的指数映射等价关系,推导了基于Lie群的自由刚体连续动力学方程.结合离散变分原理,推导了其Lie群离散变分积分子.通过证明可知连续和离散动力学系统都具有动量守恒性.对连续动力学方程进行同维化处理,使其变为常规非线性方程组的形式,利用Runge-Kutta法进行求解;基于Runge-Kutta基本理论,推导了直接用于Lie群的Runge-Kutta法,从而使Runge-Kutta法可用于求解变维非线性方程组;通过Lie代数变换,利用Kelly变换和Newton迭代对Lie群离散变分积分子进行求解.仿真对比结果表明,3种算法下的计算结果高度吻合,且能高精度地保持系统的结构守恒和动量守恒性.     

Optimal feedback control for constrained mechanical systems

An approach to minimize the control costs and ensuring a stable deviation control is the Riccati controller and we want to use it to control constrained dynamical systems (differential algebraic equations of Index 3). To describe their discrete dynamics, a constrained variational integrators [1] is used. Using a discrete version of the Lagrange-d’Alembert principle yields a forced constrained discrete Euler-Lagrange equation in a position-momentum form that depends on the current and future time steps [2]. The desired optimal trajectory (q_opt, p_opt) and according control input u_opt is determined solving the discrete mechanics and optimal control (DMOC) algorithm [3] based on the variational integrator. Then, during time stepping of the perturbed system, the discrete Riccati equation yields the optimal deviation control input u_R. Adding u_opt and u_R to the discrete Euler-Lagrange equation causes a structure preserving trajectory as both DMOC and Riccati equations are based on the same variational integrator. Furthermore, coordinate transformations are implemented (minimal, redundant and nullspace) enabling the choice of different coordinates in the feedback loop and in the optimal control problem. (© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Multi-symplectic variational integrators for the Gross–Pitaevskii equations in BEC

In this paper, a variational integrator is constructed for Gross–Pitaevskii equations in Bose–Einstein condensate. The discrete multi-symplectic geometric structure is derived. The discrete mass and energy conservation laws are proved. The numerical tests show the effectiveness of the variational integrator, and the performance of the proved discrete conservation law.     

TVD-based Finite Volume Methods for Sound-Advection-Buyancy Systems

Split-explicit Runge-Kutta methods provide an efficient integration procedure for hyperbolic systems with coupled slow and fast wave phenomena. They are generalized to multirate infinitesimal step methods (MIS) in order to develop an order to provide order conditions and to establish stability properties. The construction of MIS methods is based on an underlying Runge-Kutta method. This method is choosen to be total variation diminishing (TVD) to improve the stability properties of the method. Here, the maximum Courant number is improved by a factor of 4. (© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

A Moreau-type Variational Integrator

In this paper we derive a variational integrator for nonsmooth mechanical systems by discretizing the principle of virtual action with finite elements in time. After the discretization with local finite elements, the constitutive laws for the contact forces are introduced as in Moreau's time stepping scheme. This derivation shows exemplary how variational integrators for systems with frictional unilateral constraints can be derived. The long-time energy behavior of the presented scheme is compared with the behavior of Moreau's stepping scheme on an example system. (© 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)