On the perturbation algorithm for the semidiscrete scheme for the evolution equation and estimation of the approximate solution error using semigroups (Q519129)

From MaRDI portal
scientific article
Language Label Description Also known as
English
On the perturbation algorithm for the semidiscrete scheme for the evolution equation and estimation of the approximate solution error using semigroups
scientific article

    Statements

    On the perturbation algorithm for the semidiscrete scheme for the evolution equation and estimation of the approximate solution error using semigroups (English)
    0 references
    0 references
    0 references
    4 April 2017
    0 references
    Let be given the initial value problem of the standard form for evolution equation in Banach space \(X\) \[ {{du(t)}\over{dt}}+Au(t)=f(t),\;\;t\in[0.T],\leqno(1) \] \[ u(0)=u_0 \] with the operator \(A\) such that \(-A\) generates a strongly continuous semigroup \(exp(-tA)\), and \(f\) is a continuously differentiable function. In connection with this problem, the authors consider the three-level finite difference scheme, `perturbed' by the second-order finite difference, of the following form: \[ {{u(t_k)-u(t_{k-1})}\over{\tau}}+Au(t_k)+{\tau\over2} ({u(t_k)-2u(t_{k-1})+{u(t_{k-2})}\over{\tau^2}})=\leqno(2) \] \[ =f(t_k)-{\tau}^2R_k(\tau,u),\;\;R_k(\tau,u)\in X. \] Replacing in (2) the term \({\tau\over 2}\) by the parameter \(\epsilon\), then taking the (formal) development \[ u_k=\sum_{j=0}^\infty \epsilon^j u_k^{(j)} \] and gathering terms of equal `power' \(\epsilon^j\), they deduce the following sequence of finite difference schemes: \[ {{u_k^{(0)}-u_{k-1}^{(0)}}\over \tau}+Au_k^{(0)}=f(t_k) \] \[ {{u_k^{(1)}-u_{k-1}^{(1)}}\over \tau}+Au_k^{(1)}= -{{u_k^{(0)}-2u_{k-1}^{(0)}-u_{k-2}^{(0)}}\over \tau} \] \[ \cdots \] Let \[ v_k=u_k^{(0)}+{\tau\over 2}u_k^{(1)} \] \[ \cdots \] The main theorem of this paper says that under certain `natural' conditions on the operator \(A\) and on the function \(f\), the vector \(v_k\) approximates the solution \(u(t_k)\) with order \(O(\tau^2ln^2{1\over\tau})\). All above implies that if for the problem (1) the two level finite difference scheme of the Crank-Nicolson type is applied: \[ {{u_k-u_{k-1}}\over\tau}+A{{u_k+u_{k-1}}\over2}=f(t_{k+{1\over2}}), \] the similar error estimation can be obtained. In addition, the authors generalize the result to four levels finite difference schemes. It is to be emphasized that this paper contains complete proofs to the main results.
    0 references
    initial value problem
    0 references
    differential equation with linear operator in Banach space
    0 references
    finite difference schemes
    0 references
    reduction to two-level schemes
    0 references
    Crank-Nicolson scheme
    0 references
    error bounds
    0 references
    convergence
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references