PROJECTION METHODS FOR EQUATIONS OF THE 2ND KIND

被引:12
作者
SLOAN, IH
NOUSSAIR, E
BURN, BJ
机构
[1] School of Mathematics, University of New South Wales, Sydney
关键词
D O I
10.1016/0022-247X(79)90179-3
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
A class of projection methods, differing from the classical projection methods, is studied for the equation y = f + Ky, where K is a compact linear operator in a Banach space E, and f ε{lunate} E, In these methods K is approximated by a finite-rank operator Kn, which is constructed with the aid of certain projection operators, and which satisfies Knz = Kz for all z belonging to a chosen subspace Un ⊂ E. Under certain conditions, it is shown that the convergence of the approximate solution is faster than that of any classical projection method based on the subspace Un. In an example, Un is taken to consist of piecewise constant functions, and the projections are so chosen that the method becomes equivalent to a single iteration of a classical method, the collocation method; in this case the error (in the supremum norm) is O( 1 n2), compared with O( 1 n) for the collocation method. © 1979.
引用
收藏
页码:84 / 103
页数:20
相关论文
共 7 条
[1]  
ATKINSON KE, 1976, SURVEY NUMERICAL MET
[3]  
Krasnosel'skii M. A., 1972, APPROXIMATE SOLUTION
[4]  
Sloan I. H., 1976, Journal of the Australian Mathematical Society, Series B (Applied Mathematics), V19, P422, DOI 10.1017/S0334270000001296
[5]  
SLOAN IH, 1976, NUMER MATH, V25, P231, DOI 10.1007/BF01399412
[6]   NEW APPROACH TO NUMERICAL-SOLUTION OF INTEGRAL-EQUATIONS [J].
SLOAN, IH ;
BURN, BJ ;
DATYNER, N .
JOURNAL OF COMPUTATIONAL PHYSICS, 1975, 18 (01) :92-105
[7]  
WAIT R, 1974, NUMERICAL SOLUTION I, pCH23