ROCZNIKI POLSKIEGO TOWARZYSTWA MATEMATYCZNEGO Séria I: PRACE MATEMATYCZNE XXIV (1984)
J. P o p io l e k (Bialystok)
On integro-differential equations in para-normed space
Ugowski [ 6 ] proved the existence theorem of solutions of the first initial-boundary value problem for the system of parabolic integro- differential equations. The proof of this theorem is based on Schauder’s fixed point theorem and Friedman’s a priori estimates for a single parabolic equations ([3], p. 65 and p. 191).
Author [4] considered the first initial-boundary value problem for the following infinite system of parabolic integro-differential equations:
(1) £ а% (x, t)DÎ и "+ £ ЬТ (х, î)DXjum + cm (x, t)u " -D ,u "
i,j — 1 J i = l
= f m (x , t, u \ u2, ..., DXl u \ DXl u2, ..., DXnu \ DXf u2, ...
..., J u1 (y, t)nl (x, t ; dy ), j u 2 (y, t)\i2 (x, t ; dy), ...) (m = 1, 2, ...).
g , g ,
Method of proof of the existence of solutions is similar to the method applied by Ugowski in [ 6 ].
In this paper we are dealing with the existence of solutions of the first initial-boundary value problem for the system of type (Î). We take advantage of the Schauder-Tichonov fixed point theorem for a locally convex space [5].
Let G be a bounded open domain of Euclidean space En+1 of the variables (x, t) = (xl 9 x2, ..., x n, t), where n ^ 2, enclosed by two domains R 0 and R T lying on the planes t = 0 and t — T = const > 0 respectively, and by a surface S located in the strip 0 < t ^ T. By I we denote the parabolic boundary of domain G, i.e. E = R 0 u S .
Let us introduce the following norms ([3], p. 61)
. iG I , 0 4 1 I ,G î |G î \ U ( P ) - U ( Q )I
Mo = sup |u(P)|, « = Mo+ sup r -- ,
P e G
P,QeG
L « » \ l) JMf+a = Ыа + J \DxM\S> M2 + « = MÎ + « + Z l ^ x M S + \Dtu\S, i,j= 1
i= 1
where d(P, Q) — the parabolic distance of points P, Qe G; 0 < a < l
By Ck+a{G) (к = 0, 1, 2) we denote the set of all functions u(x, t) for which \u\f+a < oo.
Set и (x, t) = (u1 (x, t), u2 (x, t), ...), û(x, t) = (n 1 (x, t), Ü2 (x, t), ...), where um(x, t), й™(х, t)eCk+a{G) (m = 1 ,2 ,...; к = 0, 1, 2; 0 < a < 1). We define the following operations:
w + u = (w 1 +Ы1, m 2 + m 2, rju — ( щ 1, щ 2, ...)
(rj is a real number).
9Then the set Q°+a(G) is a linear space of all vector-functions u(x, t). In the C£°+a(G) we define the function
( 2 ) NlF+« = I ^ 2m Mm + \u l m[G k + a where constants Mm > 0 will be specified later on.
The function || ||fc+a is a para-norm, because it satisfies conditions (F l)~
(F4) ([1], p. 103). The space C^°+a(G) with the para-norm (2) is a locally convex space ([1], p. 313).
For every 0 < t ^ Tlet Bm (m = 1, 2, ...) be an operator defined on the set of all vector-functions u(x, t) regular in GT with values belonging to the set of all functions defined in GT\ I \ where Gr = G n {(x, t): 0 < t < t } and Г = I n |(x, t): 0 < t ^ t }.
We shall consider the problem
(3) L" и" s £ aTj (x, t) D1 u " + £ b? (x,t) Dx. u” +
i j = l i = l
+ cm(x, t)um- D t um = Bmu, (x, t)E~GT\ r , (4) um(x, t) = cpm(x, t), ( x , t ) e r (m = 1 ,2 ,...) .
By a solution of equation (3) we shall always understand a regular solution, i.e. continuous in the domain GT and possessing in Gr \ I z continuous derivatives appearing in Lmum.
The following assumptions are introduced (/, j = 1, 2, ..., n; m
= 1, 2, ...; 0 < т ^ T).
I. For any (x ,t ) e G and ÇeEn we have a” (x, t) — а™(х, t), П
а%(х, t)£i£j ^ K 0\Ç\2, where K 0 is a positive constant.
i,j= 1
II. The coefficients of Lm satisfy the uniform Holder condition with exponent a in G, and, moreover, a ^ e C 1^ 0(S) ([ 6 ], p. 256).
III. The sufrace S belongs to C 2 +an C 2- 0 ([ 6 ], P- 256).
IV. The functions (pm(x, t) are of class C1 +P(ZT) n С 2 +(Х(ГГ), a < fi < 1.
V. If <PeC?+f({Gx) n C ? +a{Gx) and Ф = (p on Г , then ВтФ = Ьт(рт on dR0.
VI. Operators Bm are continuous in the space C f +a(Gx), i.e. if u, use C f> +a(Gx) and lim || ms —и||?+в = 0, then lim \\Bmus —Bmu\\oT = 0.
s - * o o s-> oo
VII. Operators Bm map the space C*+a(Gr) into the space C 0 (GT) and furthermore the following inequality holds
(5) |B"u|g
where Mm > 0 is such that
( 6 ) К (a) Tr \Lm Фт|<>* + l^ m|i+ a « Mm( l - £ ( a ) r ')
for a constant K(a) depending only on a, K 0, K 2, К ъ and domain G ([ 6 ], p.
262); y = î ( l —a).
T heorem 1. I f assumptions I-VII are satisfied, then there exists a solution u(x, t) = (u 1 (x, t), u2(x, t), ...) of problem (3), (4); moreover, ueCf+p{Gx) n
п С 2 *+£(С'), 0 < £ < 1.
P ro o f. Denote by A the set of all vector-functions u e C f +a(Gx), such that |мт|?+а ^ M m and um(x, t) = q>m(x, t) on I х for every positive integer m.
Now for u e A consider the following problem
(7) L mvm = Bmu (x , t ) e G x\ r ,
( 8 ) vm{x,t) = (pm{x,t) (x , t ) e Z x {m = 1 , 2 ,...).
By virtue of assumptions I-VII and Lemma 1 ([ 6 ], p. 258) problem (7), ( 8 ) possesses a unique solution v(x, t) belonging to C 2 +e(Gx); Furthermore by (2.5) ([ 6 ], p. 262) u eC ? +ls(Gx).
Now, we define on A a transformation Z setting v — Zu. Applying a topological method based on the theorem of Schauder-Tichonov ([5]) we shall prove that Z has a fixed point.
At present we show that Z(A) cz A. In view of inequality (2.5) ([ 6 ], p.
262) we have
(9) И ? '+, « K(«r»(|B"a|g' + |L" 0 4 f ) + |<2>"|f+(1.
Assumption VII applied to (9) yields the relation
(Ю) \ v l ?+, « м т.
It also follows from (10) and assumption IV that veA. Note further that Z is continuous, i.e. lim ||ue — к||?+в = 0 implies lim ||Zus — Zu\ff+a = 0.
s oo .s oo