ROCZNIKI POLSKIEGO TOWARZTSTWA MATE MAT YCZNE G O Séria I: PRACE MATEM AT Y CZNE X V II (1974)
J an Mtjsiâlek (Krakôw)
The mixed boundary conditions oî Dirichlet-Neumann type for the Laplace equation in the positive quadrant of the plane
In this paper we solve the Laplace equation in the region D — {(xx ? > 0, x 2 > 0} with Dirichlet condition an one half-line ah the boundary of D and with the Neumann condition on the another one.
In the paper [2] we solved only the Neumann problem for this region.
The Green function by means of which we solve the problem, is constructed by aid of the symmetric images.
1. Let X = X xx(xx, x 2), Y = Y ( y x, y 2) denote two different points in D.
Let X _n = ( — x x, x 2), X x_x = (xx, — x 2) be symmetric images of X in the axis Oy2 and Oyx, respectively. The point X _ x_x = ( —xx, — x 2) is the symmetric image of X _ n in the axis Oyx or of X x_x in the axis Oy2.
Let
rxT =
Y Xn =
[ ( Уг - о е х)2 + (у2- х 2)2Т * , r z h = Y X _n = [{Ух + осх)2 + (у2- х 2)2Т * , ГГЛ = * £1 - 1 = [(У1-®1)* + (Уа + ®2)2Г*, r l L i = X X -i-i — [(2/1 + ^i)2 + (Уг + ^г)2] *and let denote l x the half-line x 2 = 0, xx > 0 and l 2 the half-line xx — 0, æ2 > 0.
Theorem 1 . The Green function harmonic in D and satisfying the boundary conditions
(1) G(X, Y) II о II о for Y e l
(2) dG(X, Y)
= 0 for Y e l is of form
дух 2/1 = 0
(3) G(X, Y) = —logru —logr_ 11 + logr1_1-flogr._1-_1.
P ro o f. The function G(X, Y) of form logrlx-f .ff(X, Г),
where
H (X , Y) = - l o g r .n + lo g ^ .i + lo g r .j.!
is harmonie with respect to Y. The function G (X, Z) satisfies condition (1). For Y e l xwe get
iogru = lo g r^ j and logr_n = lo g r ^ ^ . For Y e l 2 we have by (3)
dG(X, Y)
\x\
f y i 2xx
У1 = 0
2а?! 2a?i 2a?i
+ (У2- ^ ) 2 ocl + (y2~ x 2)2 x\ + {y2 + x2f X?x + {y2 + X2Ÿ = 0.
2. Using some lemmas dealing with convergence of certain integrals we shall prove that the function
(4)
OO OÛ
и (хг, х 2) = ~ - f /1(2/1) dyx- f f 2(y2)G(X, Y)
2tz J d y2 y2=0 2tcJ Vi=0^2/27
where G(X, Y) is defined by (3) solves in D the Laplace equation and satisfies the mixed boundary conditions
(5) lim^(a?i, a?2) = f x(x°x) as (a?i, a?2) ->■ (a?J, o),®; > o , (6)
Indeed
lim ---- = /du 2(^2)
da?i 2 as (a?i, a?2) -► (0, x2). & V ©
= 2a?2{[(2/i-a?i)2 + a?5] x + [(2/i + ®i)2+ ^ ] x}.
a/2=°
Hence by (3) the function и {xx, a?2) defined by (4) takes the form 4
(4a) u (x x, x 2) = £ Ii(X ),
i = l
where
OO CC Г
I x(X) = -± fx (y x m y i-o o 1f + 4 T 1dy1,
7Z J 0 CC c00
I 2(X) = — / i W ^ i + ^ + ^ r 1# ! ,
7Г J 0
1 00
1 з(Х) = “
2^ J /
2Ы ^ [ ж 2! + (
2/
2- ж
2)
2]Й
2/2,
0 OO
I
4(X) = - ^ i J /2(2/2)l°g[a
?2+ (2
/2+ #
2)
2]d2/2.
№ dy*
Let a > 0 and let
Wx (a, A, B) = < —В , B} x <№, -4.), TF2(a, -4, B) = {a, A} x < —В, B>.
Lemma 1. Let the function f x{yx) be Lebesgue integrable and let for every rectangle Wx(a, A, B) exist a constant N (W X) > 0 such that the integral
/ \fi{yi)Vy\dyx
N
converges ; then the integrals
r dk
Ц (Х ) = (т г)-ч J
о *
and
!(X) = (7i)~1x 2 J /1(y1) ^ r [ ( y i + ®i)2 + ^ r 1^ 1»
w/iere & = О, 1, 2 ; i = 1 , 2 ; s = 0 ,1 , .. ., 7 exist and are almost uniformly convergent in the half-plane — oo < x x < + oo, x 2> 0.
P ro o f. The proof will be carried only for 1ЦХ), h = 0, i = 1, s = 0; the proof in remaing case is similar because those integrals admit an analogous majorant.
Let <TX = (yx — x x)2-\-ot%. By the inequality of the triangle: \ yx < dx
< 2y x. For y x > N (W X) and for each (xx, x 2)e Wx(a , A, B). Hence / l/i(2/i)I {.{Vi + ^ ] l dyx
N
OO
< j \МУ1)\Х21(У1-Х1)2 + 4ТЧ(У1-Хх)2 + Х1]-Ыух
N oo
f lfi(yi)lyr*d yi-
N
Lemma 2. Let f z{yz) be Lebesgue integrable and let for each rectangle
^2(a, A, B) exist a constant N (W 2) such that the integral
N
/
converges. Then the integrals
\ M y ^ o g y 2dy2
I f dk
Ц (Х ) f2(y2)-^gr{log[x2x + (y2-hx2f]} d y 2,
0 i
OO £
If(Z ) = - i - J f 2(y2) - ^ { l o g [ x î + (y2- x 2)2]}dy2,
n *
where к = 0 , 1 , 2 ; г = 1, 2 ; р = 0 ,1 , ..., 7 exist and are almost uniformly convergent in the half-plane x x > 0, — oo < x 2 < + oo.
P ro o f. We will prove thesis for the integrals I\{X) and 1\{X) for i — 1 ; к = 0 ; p = 1.
Let d2 — x\-\-(y2 — x 2)2. By the triangle inequality \ y 2< d < % y 2 for y 2 > N (W 2) and for each (xx, x2)e W 2{a, A, B).
Hence
OO
l f ( X ) = f lf 2(y 2)llo g[x21 + (y2- x 2)2]d y 2
N
OO
< 2
j
\f2(y2)\logy2dy2 for N (W 2).N
For the integral I\(X) we obtain ^
OO
I f( X ) = 2
j
\f2{y2)\oc1 [x\ + (y2 + x 2)2Y 1dy2Noo oo
< 2
J |/
2Ы1«1Й + 2/2Г
1%2<
2J
|/2(2/2)1 l°g2/2^2
for N (W 2) sufficiently large and for each (a?x, x 2)e W2{a, A , B). This majorant applies also to the remaining integrals.
Theoeem 2. TJnder the assumptions of Lemmas 1 and 2 the function u (x1, x 2) defined by (4) or (4a) is harmonic in D.
P ro o f. By Lemmas 1 and 2 the integrals If (X), Ц (Х ), I f (X), I f (X) converge almost uniformly, whence we can interchange the integration with the differentiation.
Therefore
*» OO
СО Г
Au{x u x2) = — fi{ y i)f i(y i)A [(y 1 ± x 2)* + 4 ]~ 1dy1±
7C J
0 00
00 c
± — /2(2/2)^ {l0g[^ + (^2 ± ^ 2)2] } ^ 2 = <>•
7C J 0
The Laplacjan A under the sign of the integral vanishes, since the involved
function is harmonic. 7
We shall prove now that the boundary conditions (5) and (6) are satisfied.
Lem m a 3. Let the function f 1{y1) satisfy the assumptions of Lemma 1 and let it be continuous at x% moreover, for each Ô > 0 let
f ^2I/i(2/i) - /iK)I[(2/i —^i)2 + ^ r ^ 2 / i- ^ 0
as xx x\, x2 -> 0 ;
i i W Л Ю to » ^2) -> К j o).
P roo f. Let
0 for y x < 0, Л Ы =
/1(2/1) f o r y ^ O . Hence
+ 00
/ £ /* —
i 1{ X ) = — f A y J K v i - ^ + x Y l t y x.
7t J It is known ([1], p. 270) that
x« +00
(8)
+00 +00
— 71 J
f
[(2 /i - % ) 2 + ^ ] _ 1 # i = — fЯ J ds1
+
6*2 == 1 .Whence +00
T /*
(9) — I fi{x\)i{yi~ X i? + ®lYl dyl = Л К ) . 7C J
— 00
Let ns write I x{X) as
+ OO
СО г
Ii(X)=J(X) + -± fMmyi-Xif+rtT'ày,
TZ J where
+ 0 0
со г
J ( X ) = - 1 [ Л ы - Л Ю З И л - ^ + ^ Г 1# !-
7C J By (8) and (9)
i . m - A K ) = j ( X ) .
Let e > 0 ; continuity of f x at xx implies existence of <5 > 0 such that
\yx-x\\ < 6implies
Now where
I/i(2/i) - /iK)I < £/2 - J ( X ) = J x(X) + J 2(X ),
J x{X) = — 7Г J
Г
[/i(2/i) - /iK )][(2/i- ^ i)2+ ^]
I3/X—
CO c
J 2(X) = —, [ Л Ы - Л Ю Ш ^ - ^ + ^ Г 1^ ! .
7C , J
Thus applying (9) we obtain
^ «/
|î/x - a : J | < ( 5
£ 00o Г+ oo „ e
< 7 7 — [(У 1-Я 1) +®?] = — .
2 7Г J 2
—00
It follows that J 1(X )-+ 0 as xx -> a?2 -> 0+.
By our assumption aq being sufficiently close to x\ and x2 being sufficiently close to zero, we have
|J2(X )| < £/2, whence
| J|< e.
Lem m a 4 . Let the functions /1(2/1) and /2(2/2) satisfy the assumptions of Lemmas 1, 2, 3 ; then
(a) I 2(X )-> 0 as 0+ , x2> 0,
(b) \1 §{Х} -\-1 ^Х)~\ — 0 us a?2 —> 0+, a?2 —> 0. P ro o f, ad (a) The integral J2(X) satisfies the inequality
00
I2( X ) < — f |/i(2/1)l(2/î + ^ + ^ r 1# i
7C J 0 /£» 00
TC J
о
The last integral is independent of x2.
ad (b) By Lemma 2 the integrals I 3(X) and I4(X) converge uniformly for х г > a > 0, —
00
< a?2 < + 00, a being an arbitrary positive number, whence those integrals are continuous functions and for хг = 0 we havei , ( X ) + i t ( X) = ~ f
/ .M P o gtâ + sD -io gtâ+ ÿî)] =
0.
0
Lemmas 3 and 4 imply that the boundary condition is satisfied. Let us prove now the boundary condition (6). By Lemmas 1 and 2
du дхг
4
У ^ Х ) ,
1
where
SAX) = f /^Уг){Уг-^хШУ
1- ^
1 )2+ (^Т
2Луи
1 ' 0
OO
®z(X) = ~ f ^
=~ J
М У 2 ) [ ^ 1 + ( У 2 - 0 0 2)2Т 1^У2,1 0
OO
= - ^ —“ = — — f Л (2/i)(У1 + æi) {.(Ух + x i f + ^]~uOC-i CO *) 2dyx,
0
d l (X) x r°
^4W = —^ ---- = - ^ J - J / а Ы Й + ^ а + ^а)8] - 1^ -
1 о
Lemma 5. Let the function /1(2/1) satisfy the assumption of Lemma 1, let /2(2/2) be continuous at x 0 and satisfy the assumptions of Lemma 2, and let for every ô > 0
f 001 1/2
(2/2)- / 2
(a>2
) I O ? + (2/2 - ^ a)2]-1
# 2 - >0
! Î/2 —•*'21 > ^
as x2 -» a?2, #i -> 0, a?i ■> 0 ; йетг
(c) S 3(X )-+f 2(x°2) as (я?и a?a) (0, a$);
(d) $4(.X) -> 0 as (a?!, a?2) -> (0, a?2);
(e) [^ i(X )+ ^2(X) ] - > 0 as (®j, a?2) -> (0, x2).
P ro o f of (c) is similar to this of Lemma 3, and this of (d) and (e) to this of Lemma 4.
Lemma 5 implies the boundary condition (6). Lemmas 1-5 and Theorem 2 imply .
Theorem 3. The assumptions of Lemmas 1-4 being supposed, the function defined by (4) or (4a) forms the solution of the Laplace equation
in the quadrant x x > 0, x 2 > 0 with boundary conditions (5) and (6).
References
[1] M. K rz y z an ski, B o w n a n ia rozniezkowe czqstkowe rzçdu drugiego. Czçsc I, W ar
szawa 1957.
[2] J. M usialek, F u n k e ja Greena oraz rozw iqzanie z ag a d n ien ia N eu m an n a i D iricM eta, Zeszyty Naukowe AGrH, Krakow 1970.