The modern Galerkin method for integral equations of the mixed type Lechosław Hącia Institute of Mathematics, Faculty of Electrical Engineering, Poznań University of Technology 60 - 965 Poznań, ul. Piotrowo 3A, e-mail: [email protected] We can see that integral equations of the mixed type t u ( x, t ) f ( x, t ) k ( x, t , y, s)u ( y, s)dyds . (1) 0 M or shortly u = f + Ku (2) play very important role in epidemiology, mechanics, electromagnetics and technology. Presented equations arise also in the heat conduction theory [5,7} and the mathematical modelling of the spatio-temporal development of an epidemic [1,2]. The spread of the disease in the given population can be described by the following mixed integral equations. Some initial-boundary problems for a number differential partial equations in physics are reducible to the considered integral equations. In this paper projection method of the Galerkin type is proposed which leads to a system of Volterra integral equations. Consider equation (1) in space-time, where f is given function in domain D M [0, T ] (M – a compact subset of m-dimensional Euclidean space) and u is unknown function in D. Given kernel k is defined in domain {( x, t, y, s): x, y M,0 s t T} . Numerical methods for studied equations were presented in [1, 3-7]. Convergence and error’s estimate of the method will be presented The modern method of the Galerkin type In this section we propose a projection method for equation (1) leading to solve a system of Volterra linear integral equations. Approximate solution of (1) we seak in the form u n x, t a t x - j (3) j j 1 for x, t D , D M 0, T , where: - n is an orthonormal and complete basis in w L M ; a is a solution to system of the following Volterra integral equations 2 j j a j t f j t n t k t , s a s ds , jk j 1, 2, ..., n k (4) k 1 0 with f j t f x, t x dx , j 1, 2, ..., n j M k jk t , s k x, t, y, s x y dydx , j k , j 1, 2, ..., n . k MM Lemma If f L2 D and k L2 , x, t , y, s : 0 s t T ; x, y M , then function (3) is a unique solution in the space L2 D of the equation t u n x, t f n x, t k n x, t , y, s u n y, s dyds , (5) 0M with f n x, t n f t x , k k 1 k (6) k n x, t , y , s n n k t, s x y . jk j (7) k j 1 k 1 Proof. Let us notice that u n x, t n u t x , k (8) k k 1 where u k t f k t n t k t , s y u y, s dyds , jk j k 1, 2, ..., n . n (9) j 1 0 M By orthonormality of k we obtain the Volterra system of integral equations u k t f k t n t k t , s u s ds , jk j k 1, 2, ..., n . (10) uk in space L2 0, T such j 1 0 From assumptions and Volterra theory it follows, this system has unique solution that u k t a k t for every k 1, 2, ..., n . Remark Equations (5) we can rewrite in the operator form un f n K nun , where (11) K n is Volterra-Fredholm integral operator of form (2) determined by the kernel k n (7). Theorem If f L2 D and k L2 , then sequence u n defined by formula (3) converges in the space unique solution of equation (1) and the estimate error c un u L2 D 1 c holds with c I K 1 Proof is based on the equality ff n n L2 D and u L2 D n L2 D to n kn k L2 . un u f n f K n K un K un u . References [1] Brunner H., On the numerical solution of nonlinear solution of Volterra equations by collocation methods, SIAM, J Numer.Anal., 27(1990). 987-1000. [2] Diekmann O., Threshold and traveling for the geografical spread of infection. J. Math Biology 6, (1978), 109-130 [3] Hącia L., Projection methods for integral equations in epidemic, Mathematical Modelling and Analysis, 7,2 (2002), 229-240. [4] Hącia L., Computational methods for Volterra-Fredholm integral equations, Computational Methods in Science and Technology, 8,2 (2002), 13-26. [5] Hącia L., Iterative-collocation method for integral equations of heat conduction problems, Lecture Notes in Computer Science, 2006 (in print). [6] Kauthen J.P., Continuous time collocation method for Volterra-Fredholm integral equations, Numer. Math., 56 (1989), 409-424. [7] Piskorek A., Integral equations - theory and applications, PWN, Warsaw 1997 (in Polish).
© Copyright 2026 Paperzz