词条 | Draft:Local Linearization Method |
释义 |
Local Linearization MethodIn numerical analysis, the Local Linearization (LL) method is a general strategy for designing numerical integrators for differential equations based on a local (piecewise) linearization of the given equation on consecutive time intervals. The numerical integrators are then iteratively defined as the solution of the resulting piecewise linear equation at the end of each consecutive interval. The LL method has been developed for a variety of equations such as the ordinary, delayed, random and stochastic differential equations. The LL integrators are key component in the implementation of inference methods for the estimation of unknown parameters and unobserved variables of differential equations given time series of (potentially noisy) observations. The LL schemes are ideals to deal with complex models in a variety of fields as neuroscience, finance, forestry management, control engineering, mathematical statistics, etc. BackgroundDifferential equations have became an important mathematical tool for describing the time evolution of several phenomenon, e.g., rotation of the planets around the sun, the dynamic of assets prices in the market, the fire of neurons, the propagation of epidemics, etc. However, since the exact solutions of these equations are usually unknown, numerical approximations to them obtained by numerical integrators are necessary. Currently, many applications in engineering and applied sciences focused in dynamical studies demand the developing of efficient numerical integrators that preserve, as much as possible, the dynamics of these equations. With this main motivation, the Local Linearization integrators have been developed. High Order Local Linearization MethodHigh Order Local Linearization (HOLL) method is a generalization of the Local Linearization method oriented to obtain high order integrators for differential equations that preserve the stability{{dn|date=November 2017}} and dynamics of the linear equations. The integrators are obtained by splitting, on consecutive time intervals, the solution x of the original equation in two parts: the solution z of the locally linearized equation plus an high order approximation of the residual . Local Linearization schemeA Local Linearization (LL) scheme is the final recursive algorithm that allows the numerical implementation of a discretization derived from the LL or HOLL method for a class of differential equations. LL methods for ODEsConsider the d-dimensional Ordinary Differential Equation (ODE) with initial condition , where is a differentiable function. Let be a time discretization of the time interval with maximum stepsize h such that and . After the local linearization of the equation (1) at the time step the variation of constants formula yields where results from the linear approximation, and is the residual of the linear approximation. Here, and denote the partial derivatives of f with respect to the variables x and t, respectively, and . Local Linear discretizationFor a time discretization , the Local Linear discretization of the ODE (1) at each point is defined by the recursive expression The Local Linear discretization (4.3) converges with order 2 to the solution of nonlinear ODEs, but it match the solution of the linear ODEs. The recursion (4.3) is also known as Exponential Euler discretization. High Order Local Linear discretizationsFor a time discretization a High Order Local Linear (HOLL) discretization of the ODE (1) at each point is defined by the recursive expression where is an order (>2) approximation to the residual r The HOLL discretization (4.4) converges with order to the solution of nonlinear ODEs, but it match the solution of the linear ODEs. HOLL discretizations can be derived in two ways: 1) (quadrature-based) by approximating the integral representation (4.2) of r; and 2) (integrator-based) by using a numerical integrator for the differential representation of r defined by for all , where The quadrature-based discretization is often called Exponential Propagation Iterative or Exponential Rosenbrock, whereas the integrator-based one is called Locally Linearized discretization. HOLL discretizations are, for instance, the followings:
which is obtained by solving (4.5) via a s-stage Runge–Kutta (RK) scheme with coefficients .
which results from the approximation of in (4.2) by its order-p truncated Taylor expansion.
which results from the interpolation of in (4.2) by a polynomial of degree p on , where denotes the j-th backward difference of .
which results from the interpolation of in (4.2) by a polynomial of degree p on ,
which results from the interpolation of in (4.2) by a Hermite polynomial of degree p on . Local Linearization schemesAll numerical implementation of the LL (or of a HOLL) discretization involves approximations to integrals of the form where A is an d d matrix. Every numerical implementation of a Local Linear discretization of any order is generically called Local Linearization scheme. Computing integrals involving matrix exponentialAmong a number of algorithms to compute the integrals , those based on rational Padé and Krylov subspaces approximations for exponential matrix are preferred. For this, a central role is playing by the expression where are d-dimensional vectors, , , , being the d-dimensional identity matrix. If denotes the (p; q)-Padé approximation of and k is the smallest integer number such that If denotes the (m; p; q; k) Krylov-Padé approximation of , where is the dimension of the Krylov subspace. Order 2 LL schemeswhere the matrices , L and r are defined as and with . For large systems of ODEs Order 3 LL-Taylor schemeswhere for autonomous ODEs the matrices and are defined as . Here, denotes the second derivative of f with respect to x, and p + q > 2. For large systems of ODEs Order 4 LL-RK schemeswhere and with and p + q > 3. For large systems of ODEs, the vector in the above scheme is replaced by Locally Linearized Runge-Kutta of Dormand & Princewhere s = 6 is the number of stages, with , and are the Runge-Kutta coefficients of Dormand and Prince and p + q > 4. For large systems of ODEs, the vector in the above scheme is replaced by Stability and dynamicswith , and , and its approximation by various schemes. This system has two stable stationary points and one unstable stationary point in the region . LL methods for DDEsConsider the d-dimensional Delay Differential Equation (DDE) with m constant delays and initial condition for all where f is a differentiable function, is the segment function defined as for all is a given function, and Local Linear discretizationFor a time discretization , the Local Linear discretization of the DDE (5.1) at each point is defined by the recursive expression where is the segment function defined as and is a suitable approximation to for all such that Here, are constant matrices and are constant vectors. denote, respectively, the partial derivatives of f with respect to the variables t and x, and . The Local Linear discretization (5.2) converges to the solution of (5.1) with order if approximates with order for all . Local Linearization schemesDepending of the approximations and of the algorithm to compute different Local Linearizations schemes can be defined. Every numerical implementation of a Local Linear discretization is generically called Local Linearization scheme. Order 2 Polynomial LL schemeswhere the matrices and are defined as and , and . Here, the matrices , , and are defined as in (5.2), but replacing by and where with , is the Local Linear Approximation to the solution of (5.1) defined through the LL scheme (5.3) for all and by for . For large systems of DDEs with and . LL methods for RDEsConsider the d-dimensional Random Differential Equation (RDE) with initial condition where is a k-dimensional separable finite continuous stochastic process, and f is a differentiable function. Suppose that a realization (path) of is given. Local Linear discretizationFor a time discretization , the Local Linear discretization of the RDE (6.1) at each point is defined by the recursive expression where and is an approximation to the process for all Here, and denote the partial derivatives of with respect to and , respectively. Local Linearization schemesDepending of the approximations to the process and of the algorithm to compute different Local Linearizations schemes can be defined. Every numerical implementation of the Local Linear discretization is generically called Local Linearization scheme. LL schemeswhere the matrices are defined as , , and p+q>1. For large systems of RDEs, The convergence rate of both schemes is , where is the exponent of the Holder condition of . Figure 3 presents the phase portrait of the RDE and its approximation by two numerical schemes, where denotes a fractional Brownian process with Hurst exponent H=0.45. Strong LL methods for SDEsConsider the d-dimensional Stochastic Differential Equation (SDE) with initial condition , where the drift coefficient and the diffusion coefficient are differentiable functions, and is an m-dimensional standard Wiener process. Local Linear discretizationFor a time discretization , the order- (=1,1.5) Strong Local Linear discretization of the solution of the SDE (7.1) is defined by the recursive relation where and Here, denote the partial derivatives of with respect to the variables and t, respectively, and the Hessian matrix of with respect to . The strong Local Linear discretization converges with order (=1,1.5) to the solution of (7.1). High Order Local Linear discretizationsAfter the local linearization of the drift term of (7.1) at , the equation for the residual is given by for all , where High Order Local Linear discretization of the SDE (7.1) at each point is then defined by the recursive expression where is a strong approximation to the residual of order higher than 1.5. The strong HOLL discretization converges with order to the solution of (7.1). Local Linearization schemesDepending on the way of computing , and different numerical schemes could be obtained. Every numerical implementation of a strong Local Linear discretization of any order is generically called Strong Local Linearization scheme. Order 1 SLL schemeswhere the matrices , and are defined as in (4.6), is a i.i.d. zero mean Gaussian random variable with variance , and p+q>1. For large systems of SDEs, in the above scheme is replaced by . Order 1.5 SLL schemeswhere the matrices , and are defined as , is a i.i.d. zero mean Gaussian random variable with variance and covariance and p+q>1. For large systems of SDEs, in the above scheme is replaced by . Order 2 SLL-Taylor schemeswhere , , and are defined as in the order-1 SLL schemes, and is order 2 approximation to the multiple Stratonovish integral . Order 2 SLL-RK schemesFor SDEs with a single Wiener noise (m=1) where with . Here, for low dimensional SDEs, and for large systems of SDEs, where , , , and are defined as in the order-2 SLL-Taylor schemes, p+q>1 and . Stability and dynamicsBy construction, the strong LL and HOLL discretizations inherit the stability and dynamics of the linear SDEs, but it is not the case of the strong LL schemes in general. LL schemes (7.2)-(7.5) with are A-stable, including stiff and highly oscillatory linear equations. Moreover, for linear SDEs with random attractors, these schemes also have a random attractor that converges in probability to the exact one as the stepsize decreases and preserve the ergodicity of these equations for any stepsize. These schemes also reproduce essential dynamical properties of simple and coupled harmonic oscillators such as the linear growth of energy along the paths, the oscillatory behavior around 0, the symplectic structure of Hamiltonian oscillators, and the mean of the paths. For nonlinear SDEs with small noise (i.e., (7.1) with ), the paths of these SLL schemes are basically the nonrandom paths of the LL scheme (4.6) for ODEs plus a small disturbance related to the small noise. In this situation, the dynamical properties of that deterministic scheme, such as the linearization preserving and the preservation of the exact solution dynamics around hyperbolic equilibrium points and periodic orbits, become relevant for the paths of the SLL scheme. For instance, Fig 4 shows the evolution of domains in the phase plane and the energy of the stochastic oscillator and their approximations by two numerical schemes. Weak LL methods for SDEsConsider the d-dimensional stochastic differential equation with initial condition , where the drift coefficient and the diffusion coefficient are differentiable functions, and is an m-dimensional standard Wiener process. Local Linear discretizationFor a time discretization , the order- Weak Local Linear discretization of the solution of the SDE (8.1) is defined by the recursive relation where with and is a zero mean stochastic process with variance matrix Here, , denote the partial derivatives of with respect to the variables and t, respectively, the Hessian matrix of with respect to , and . The weak Local Linear discretization converges with order (=1,2) to the solution of (8.1). Local Linearization schemesDepending on the way of computing and different numerical schemes could be obtained. Every numerical implementation of the Weak Local Linear discretization is generically called Weak Local Linearization scheme. Order 1 WLL schemewhere, for SDEs with autonomous diffusion coefficients, , and are the submatrices defined by the partitioned matrix , with and is a sequence of d-dimensional independent two-points distributed random vectors satisfying . Order 2 WLL schemewhere , and are the submatrices defined by the partitioned matrix with and Stability and dynamicscomputed by various schemes. Historical notesBelow is a time line of the main developments of the LL method. - Pope D.A. (1963) introduces the LL discretization for ODEs and the LL scheme based on Taylor expantion. doi :10.1145/366707.367592- Ozaki T. (1985) introduces the LL method for the integration and estimation of SDEs. The term "Local Linearization" (LL) is used for first time. [https://doi.org/10.1016/S0169-7161(85)05004-0 doi:10.1016/S0169-7161(85)05004-0]- Biscay R. et al. (1996) reformulate the strong LL method for SDEs. doi:10.1007/BF00052324- Shoji I. and Ozaki T. (1997) reformulate the weak LL method for SDEs. doi: 10.1111/1467-9892.00064- Hochbruck M. et al. (1998) introduce the LL scheme for ODEs based on Krylov subspace approximation. doi:10.1137/S1064827595295337- Jimenez J.C. (2002) introduces the LL scheme for ODEs and SDEs based on rational Padé approximation. doi:10.1016/S0893-9659(02)00041-1- Carbonell F.M. et al. (2005) introduce the LL method for RDEs. doi: 10.1007/s10543-005-2645-9- Jimenez J.C. et al. (2006) introduce the LL method for DDEs. doi: 10.1137/040607356- de la Cruz H. et al. (2006, 2007) and Tokman (2006) introduce the two classes of HOLL integrators for ODEs: the integrator-based and the quadrature-based. 10.1007/11758501\\_22, 10.1016/j.amc.2006.06.096 and 10.1016/j.jcp.2005.08.032- de la Cruz H. et al. (2010) introduce the strong HOLL method for SDEs. doi:10.1007/s10543-010-0272-6Reference
|surname=Carbonell F. |surname2=Jimenez J.C. |surname3=Pedroso L.M. |year=2008 |title=Computing multiple integrals involving matrix exponentials, |journal=J. Comput. Appl. Math. |volume=213 |page=300–305 |doi=10.1016/j.cam.2007.01.007 }}
|surname=de la Cruz H. |surname2=Biscay R.J. |surname3=Carbonell F. |surname4=Jimenez J.C. |surname5=Ozaki T. |year=2006 |title=Local Linearization-Runge Kutta (LLRK) methods for solving ordinary differential equations, |journal=Lecture Note in Computer Sciences, Springer-Verlag. |volume=3991 |page=132–139 |doi=10.1007/11758501_22 |series=Lecture Notes in Computer Science |isbn=978-3-540-34379-0 }}
|surname=de la Cruz H. |surname2=Biscay R.J. |surname3=Carbonell F. |surname4=Ozaki T. |surname5=Jimenez J.C. |year=2007 |title=A higher order Local Linearization method for solving ordinary differential equations, |journal=Appl. Math. Comput. |volume=185 |page=197–212 |doi=10.1016/j.amc.2006.06.096 }}
|surname=de la Cruz H. |surname2= Biscay R.J. |surname3=Jimenez J.C. |surname4=Carbonell F. |year=2013 |title=Local Linearization - Runge Kutta Methods: a class of A-stable explicit integrators for dynamical systems, |journal=Math. Comput. Modelling. |volume=57 |issue= 3–4 |page=720–740 |doi=10.1016/j.mcm.2012.08.011 }}
|surname=de la Cruz H. |surname2=Biscay R.J. |surname3=Jimenez J.C. |surname4=Carbonell F. |surname5=Ozaki T. |year=2010 |title=High Order Local Linearization methods: an approach for constructing A-stable high order explicit schemes for stochastic differential equations with additive noise, |journal=BIT Numer. Math. |volume=50 |issue=3 |page=509–539 |doi=10.1007/s10543-010-0272-6 }}
|surname=de la Cruz H. |surname2=Jimenez J.C. |surname3=Zubelli J.P. |year=2017 |title=Locally Linearized methods for the simulation of stochastic oscillators driven by random forces, |journal=BIT Numer. Math. |volume=57 |page=123–151 |doi=10.1007/s10543-016-0620-2 }}
|surname=M. Hochbruck. |surname2=A. Ostermann. |year=2011 |title=Exponential multistep methods of Adams-type, |journal=BIT Numer. Math. |volume=51 |issue=4 |page=889–908 |doi=10.1007/s10543-011-0332-6 }}
|surname=Jimenez J.C. |year=2009 |title=Local Linearization methods for the numerical integration of ordinary differential equations: An overview, |journal=ICTP Technical Report. |volume=035 |page=357–373 |url=http://publications.ictp.it }}
|surname=Jimenez J.C. |surname2=Biscay R. |surname3=Mora C. |surname4=Rodriguez L.M. |year=2002 |title=Dynamic properties of the Local Linearization method for initial-value problems, |journal=Appl. Math. Comput. |volume=126 |page=63–68 |doi=10.1016/S0096-3003(00)00100-4 }}
|surname=Jimenez J.C. |surname2=Carbonell F. |year=2009 |title=Rate of convergence of local linearization schemes for random differential equations, |journal=BIT Numer. Math. |volume=49 |issue=2 |page=357–373 |doi=10.1007/s10543-009-0225-0 }}
|surname=Jimenez J.C. |surname2=Carbonell F. |year=2015 |title=Convergence rate of weak Local Linearization schemes for stochastic differential equations with additive noise, |journal=J. Comput. Appl. Math. |volume=279 |page=106–122 |doi=10.1016/j.cam.2014.10.021 }}
|surname=Jimenez J.C. |surname2=de la Cruz H. |year=2012 |title=Convergence rate of strong Local Linearization schemes for stochastic differential equations with additive noise, |journal=BIT Numer. Math. |volume=52 |issue=2 |page=357–382 |doi=10.1007/s10543-011-0360-2 }}
|surname=Jimenez J.C. |surname2=Pedroso L. |surname3=Carbonell F. |surname4=Hernandez V. |year=2006 |title=Local linearization method for numerical integration of delay differential equations, |journal=SIAM J. Numer. Analysis. |volume=44 |issue=6 |page=2584–2609 |doi=10.1137/040607356 }}
|surname=Jimenez J.C. |surname2=Sotolongo A. |surname3=Sanchez-Bornot J.M. |year=2014 |title=Locally Linearized Runge Kutta method of Dormand and Prince, |journal=Appl. Math. Comput. |volume=247 |page=589–606 |doi=10.1016/j.amc.2014.09.001 }}
|
随便看 |
|
开放百科全书收录14589846条英语、德语、日语等多语种百科知识,基本涵盖了大多数领域的百科知识,是一部内容自由、开放的电子版国际百科全书。