ROCZNIKI POLSKIEGO TOWARZYSTWA MATEMATYCZNEGO Seria I: PRACE MATEMATYCZNE X I (1967)
ANNALES SOCIETATIS MATHEMATICAE POLONAE Series I: COMMENTATIONES MATHEMATICAE X I (1967)
E.
Śl iw iń sk i(Kraków)
On the Neumann problem in an ^-dimensional hall-space
In the present paper we give a solution of the problem of Neumann in the ^-dimensional half-space. The assumptions concerning the density of f(s) are more general than in the paper [1] (see [1], p. 185-192).
Let
qdenote the distance of the points X = {xx, ..., xn) and
ń
— ((? ^, • • •, i) and let d$ — ds-^ ... ds^ ^.
Let f(8 ) be a continuous function defined in the whole (n—^ -di
mensional Eucliden space E n_u T = (<x, ..., <n_1} 1), \T\n = p2 + . .. + + < »_i+ l)n/2 and dT = . .. dtn_x. We shall prove that, under some assumptions on this function, the solution of Neumann problem in half- space is given by the formula
(1) * ( * ) = -
(2 f l ) CLnSS
[(й/j ^l)2 • • • H-f{S )d S
{ ^ n — 1 Sn — l)2 T"#'»]2_L/»2-|ł»-l Ln- 1
1
ss
(2— n ) a
f{S)dS П-2
_ i
where an = S S dT
E „ _ 1
\T\ That means that if xn > 0 then the function
u(X) satisfies the conditions 1° u(X) e<72, 2° Au(X) = 0, 3° if X ^ (x\, ...
...,ж°_!,<)) then we have limn* (-^) — /(ж?, ..., a?$Li)-
We begin with two definitions and two lemmas. We shall consider two classes of functions, К and L.
De f in it io n
1. К is the class of functions f ( X ) defined in E n_x and satisfying the following conditions: 1°/ ( X) is continuous in E n_1, 2° there exist a positive constant rx and a function co(r), where v = (s2- f ... +sL_1)1/2
>7*!, such that if r is outside C(Kr ) = {S: sj| + . ..
+ ^ _ i = ?T},then
\f{S)\ < co(r), 3° co(r) is a nonnegative nondecreasing function such that oo
the integral f r ~ 2(o(r)dr is finite.
1 8 0 E. Ś l i w i ń s k i
Definition
2. L is the class of functions defined for S e E n_i and satisfying the following conditions: 1° f(8 ) is continuous in E n_x, 2° there exists a nonnegatiye nondecreasing, continuous function Q(r) defined for r ^ 0 such tllQjtl f
01? 6V6ry р
8)
1Г of J)OHlfS (
00 у j • • • j 00_j)y (^/ij •••? Уть_l) ^ -^ть_
1and r = \ X Y I the inequality
!/(®1} •••, ®n_x)—/(y i, •••, y»_i)l < Q(r)
00
is satisfied, 3° the integral f r~2 Q(r)dr is finite, 4° there exists a continu-
1ous nonnegatiye function rp(oc) defined for
00> 0 such that
99(
0) = 0 and Q{a-b)
< 9o(a) Q(b) for a > 0, b > 0.
Lemma 1.
I f f ( 8 ) e K , then the integrals
A = < S S ^ T W d S ’ Jo = ss
Ет.— Л
(X i-S iffiS )
M + 2j
dS
converge uniformly in any cylinder 4-.
..+ # » -1< В 2, у < xn < A, where rj, A and В are arbitrary positive numbers, and i = 1, .. . , n —1, j — 0,1.
Proof. Let r — («Х + . . . + sn-i)1/2- We choose two numbers r0 and В such that r
0^ 2B and the inequality
(2)
f(JS) < eo(r) if r ^ r 0,
is satisfied. If r > r0, then we have
Q =
l{x1—s1)2+
. .. + {xn_ 1~ s n^ l)2+ x 2nf 12^ r —B ^ r - \ r = |r , and, in view of (
2), we obtain
SS
C(Kr)f(S) г dS < w SS
C(Kr)rn+2j dS, ™(r) ЛС?
C being a positive number. Let us introduce spherical coordinates (T) 6q = rcos^x ... cos
9?n_ 1sin
9?„_1, .. . , sn_l = rsiiupx in the right-hand integral which is transformed into the following
SS C4E
r)
oj(r) r w( r ) r n 2
= - p * * -
r
0)(r)
*■ = °>J тг+и *>
R
00
Cx being a positive constant. We have assumed that J r 2oj(r)dr <
00. в
Then for every e > 0 there exists
гх(ё)such that if R >
r x(e),then
% h SS
0(^ri)
№ и 4-2 j dS < e.
Neumann problem in an n-dimensional half-space 181
In order to prove the uniform convergence of the integral J 2 we consider the inequality \Xi —s*| < r-\- (ж2 + -.. - f x \ _ i)1/2 < r-\-B < |3r. Hence if r > r0 we obtain
\J'2\ = SS
W+2 jdS «SS C(KR) ~Zn+2T dS w(r)rJ
and, by introducing the transformation (T) we obtain finally
\-K\ sj С^П-> f
R
co(r)
dr < e.
Lemma 2. I f
/($ ) is continuous in
B n__xand
f ( S ) e K r\ L,then the function
(3) « д а = — o o — sr-
ап лп-1
p WQ dx
converges to f { x \ , a?$Li) when {xx, ... , xn) -> {x\ , ..., a£_i, 0).
Proof. Lemma 1 imphes the continuity of u(X ) in the half-space xn > 0. Moreover, we have
We transform the integral (3) and obtain substituting Sj — х г = ^ х п,
Sn—i X n _ j = tn_j X n,
?
* ( v = ~ - S S
E n.— 1
f (^1+
^\Xn , ..., xn_ x tn_ xхъ
- - jrj" dT.
Then
g(X) = v ( X ) - f ( x 0 1,...,x ° n^ )
[/ (*«1 ~b > • • • ) ^n-l + tn—l<^n) / (*«1 ) • • • > *«7l—l) ]
Ern—1
- S S
and
«~ss
dT
1 f* i2([(a?x—Жх+^1Жи)2 + >• • + {xn- i — +hi-i^n)2]^2)
Eyi,—i
|Tf dT
< 1 O O ^ ([([а?
1— a??| + 1<х|Д?п)2 + --. + (|а?п-
1—Д?п-х1 + l<n-i|a?n)2]1/2) Rr£
^ „ O O |T|W
E™
1 8 2 E . Ś l i w i ń s k i
Let e be an arbitrary positive number. If the coordinates of a point X satisfy the inequalities \хг — ж$| < e, \xn_i~Xn_i\ < s, 0 < xn < e, then we have
*£?([(\x1 —x\\ + |Ь1 Ж »)2~К • • 4” (\®n-i—®n~il 4~ Ihi-il^n)2]1^2)
< Q(ne\T\) < (p(ne) Q(\T\).
In view of the convergence of the integral
^ S S
i
Q(\T\)
I T\n dT we obtain the estimation
\g(X)\ <
cp(ne) an
^(\T\)
\T\n dT = Сг <p(ns)
<*n
where Сг is a positive constant. In fact, if we apply the transformation (T), then we have
J = C2f
0
о ( ( г Ч 1 ) ' у - 2
(r2 +1)"'2 C2 = const,
f Q((r3 + l ) ll2)rn- 2 ^ r a((r2+ l ) ,/2) ( r 4 l ) < n- 2>'2
J ( i ^ + i r d r < J Р + т р dr r O ^ + v n dr
J r2+ l
D 1
r Q(r)
^ < p ( 2 ) j - y - d r ( B > 1).
R
So we conclude that g{X) -> 0 if X -> (x\, 0).
Th e o r e m.
I f f ( X ) is continuous and belongs to К r\ L for S ^ E n_1 then the function u( X) defined by formula (1) is a solutio7i of the problem of Neumann in the half space xn > 0.
This follows from Lemma 1 (since the function Q2~n is harmonic).
R eferences
[1] О. Л. С о б о л е в , Уравнения математической физики, Москва 1954.