Optimized high-order splitting methods for some classes of parabolic equations

Linear differential equations in abstract spaces (34G10) One-parameter semigroups and linear evolution equations (47D06) Numerical solutions to abstract evolution equations (65J08) Numerical methods for initial value problems involving ordinary differential equations (65L05)

Abstract: We are concerned with the numerical solution obtained by splitting methods of certain parabolic partial differential equations. Splitting schemes of order higher than two with real coefficients necessarily involve negative coefficients. It has been demonstrated that this second-order barrier can be overcome by using splitting methods with complex-valued coefficients (with positive real parts). In this way, methods of orders 3 to 14 by using the Suzuki--Yoshida triple (and quadruple) jump composition procedure have been explicitly built. Here we reconsider this technique and show that it is inherently bounded to order 14 and clearly sub-optimal with respect to error constants. As an alternative, we solve directly the algebraic equations arising from the order conditions and construct methods of orders 6 and 8 that are the most accurate ones available at present time, even when low accuracies are desired. We also show that, in the general case, 14 is not an order barrier for splitting methods with complex coefficients with positive real part by building explicitly a method of order 16 as a composition of methods of order 8.

Review text

The subject of this paper is an application of the splitting method to evolution differential equations of the following form NEWLINE\[NEWLINE{du\over dt}(t)=Au(t)+Bu(t),\;\;u(0)=u_0, \leqno(1)NEWLINE\]NEWLINE where \(A\), \(B\) are operators defined over a Banach space \(X\). The primitive form of the splitting method consists in an approximation of the solution of the equation (1) by the solution of two independent equations NEWLINE\[NEWLINE{dv\over dt}(t)=Av(t)\;\;\text{and}\;\;{dw\over dt}(t)=Bw(t)\leqno (2)NEWLINE\]NEWLINE with properly chosen initial values, on a time interval \(t\in [0,h]\) with small \(h>0\). Splitting is an important tool for numerical methods, and also for the analysis of solutions of (1) in various applications. The operation of splitting is feasible under additional conditions. For example, if \(A\), \(B\), and \(A+B\) generate \({\mathcal C}^0\), a semigroup for \(t\geq 0\) over the space \(X\), then splitting is possible.NEWLINENEWLINEThe splitting method is well known from the early sixties of the twentieth century, and there exists a rich literature of this subject. Under certain conditions the method of the form (2) can reach the degree of approximation \(O(h)\) or \(O(h^2)\). This work is devoted to the question how to build the method of higher order of approximation. In fact, it is possible to make even a very high-order splitting, taking methods which may be written in a `semi-group language' as NEWLINE\[NEWLINEe^{ha_1A}e^{hb_1B}e^{ha_2A}e^{hb_2B}\cdots.NEWLINE\]NEWLINE The authors discuss the problem of the choice of the coefficients \(a_i\), \(b_i\). Because of the semigroup \({\mathcal C}^0\) properties, real positive coefficients are admitted. But the authors take into account also complex coefficients with positive real parts, and find new splittings of higher order. The paper contains certain propositions of sets of coefficients for a splitting of order 6 and 8. A rich bibliography is joined.

Recommendations

Cites work

Cited in

(56)

This page was built for publication: Optimized high-order splitting methods for some classes of parabolic equations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2840621)