Runge-KuttaMethods-RichardPalais.pdfVIP

Appendix H Runge-Kutta Methods In this appendix we will analyze the conditions on the coe铿僣ients of an explicit Runge-Kutta Method that are necessary and su铿僣ient to guarantee convergence with accuracy of order 肀 . In particular, we will establish the connection between these conditions and the set of rooted trees with no more than 肀 nodes. As a consequence, we will be able to show that there are 肀?stage methods of order 肀 for 肀 鈮 4 but not for 肀 4. We begin by brie铿倅 considering more general one-step methods, 肀睕+1 = 戆 (肀№睕 , 肀睕 , 肀? 鈩?, for approximating solutions of the scalar ODE 鈥 = 肀 (肀? 肀 (肀?). 肀 The local truncation error at 肀№睕 is the quantity 砑栱睕 de铿乶ed by 肀 (肀№睕+1 ) := 戆 (肀№睕 , 肀?肀№睕 ), 肀? 鈩? + 砑栱睕 . From our discussion following the convergence analysis of Euler鈥檚 Method in the body of the text, we can show that a 0-stable one-step method will converge to a solution of the ODE 肀 鈭 戆俄眱 +1 [肀№睖 , 肀№睖 + 肀 ] with global order of accuracy 肀 if 鈭ｍ紪肀 鈭 鈮 戆垛剮肀 +1 for some 戆 0 depending only on max肀♀垐(肀№睖 ,肀№睖 +肀 ) 鈭ｍ宝 (肀 +1) (肀?鈭? One approach to constructing methods satisfying such an esti- mate is to de铿乶e them using Taylor鈥檚 Theorem with Remainder by letting 戆 (肀№睕 , 肀睕 , 肀? 鈩? = 鈭戫眱 肀 (肀? (肀№睕 )/肀? 鈩庬睗 be the Taylor polyno- 肀滮0 mial of degree 肀 for 肀 (肀? centered at 肀№睕 and evaluated at 肀№睕+1 . To implement this idea, we must be able to express 肀 (肀? (肀№睕 ) in terms of 肀 and its derivatives evaluated at (肀№睕 , 肀睕 ). The resulting one-step methods are known as Taylor Methods. Taylor Methods are an option 263 264 H. Runge-Kutta Methods if the vector 铿乪ld that de铿乶es the ODE is given in a form that