R O C Z N IK I P O L SK IE G O TO W A RZY STW A M A TEM A TYCZNEGO Séria I : P E A C E M A TEM A TY CZN E X I X (1977)
F. B
abanskiand J . M
usialek(Krakôw)
On a certain bicaloric problem
1. In this paper we shall give the constructions of solutions of the mixed problem for the equations
(1) ' P 2TJ(x, t) = - f 7(x, «), P U = ( B l - B t)U , and
(la) P 2u (x,t) = —f { x ,t ,u { x ,t ) ) ,
in the domain S — {(x ,t): — a < x < + a , 0 < t < T } , where a, T are positive constants, with the limit conditions:
(2) U(x, 0) = М х ) for xe( ~ a , + a ),
(3) B t U (x, 0) =M<D) for xc( —a , -j-a),
(4) U {a,t) = / 3(t) for te(0, T),
(5) P (U (a 7t)) = / 4(<) for £e(0,JF), (6) U { - a ,t ) = f 5(t) for te(0, T),
(7)
1 «s' II <for t*(0, T ).
The problem of finding a solution of (1) or (la) satisfying (2)-(7) will be called briefly (B — C — B )x and (B — C — B )2 respectively. We shall prove that the solution of the problems (B — C —B )1 and (B — G — B )2 is the sum of any solution of the equation
(lb) P 2U {x,t) = 0
and any solution of equation (1) and (la) respectively.
2. In order to construct the solution of the problems (В — С — В ){
(i = 1 , 2 ) we shall use the convenient Green function.
Let
ai = {(&> t): x = + a
70 < t < T }, a
2= {(a?, t): x = —a, 0 < t < T },
and give any points X {x ,t)e S , Y (y ,s )e S \ja 1Kja2. Let X \ = X \ = X
and -Xjn+i^L+iJ <0 f °r n — 0 ,1 ,2 ,..., be the symmetric image of the
point X \n{œ\ni t) with respect to the axis ax and Х \п+2(х\п+2, t) the sym
metric image of the point X ln+1(xln+1, t) with respect to the axis a2. X \n+l {x22n+1, *) be the symmetric image of the point X 22n{x22n, t) with respect to the axis a2 and X ln+2{x22n+2, t) the symmetric image of the point Х \п+х
( 4 +i,<) with respect to the axis ax.
It is easy to verify by induction that
xln = ( — l ) n{x — 2na), x2n = ( — l ) n(x + 2na), w = 0 , 1, 2, ...
Let us consider two sequences of the functions:
Vln{X , Y) = и гп(
0) , г , у ,
8) = K (t, s)ex-p[\YX1n\2(4:(s-t))-1)
K (t, s)exp((y — ( — l ) ” (æ — 2na))2(ê(s — <))_1) for s < t,
0 for s ^ t y
and •
U2n(X , Y) = U l(ce,t,y ,s)
E( t , s)esp((ÿ — ( — l ) ” {aîH-2îta))2(4(s — f))” 1} for s < t y
0 for t.
and
V \(X, X) = V l(X , Y) = X ( « ,s ) e x p ||r X |J (4(S - 0 ) - 1) K (t, s)exp((ÿ — æ)! (4(s — <))_1) for s < t,
0 f or 8 > t,
where K (t,s) — [2 (тг(£ — s))*|-1.
3. Now we shall prove the following
T
heorem1 [2]. The Green function for the problems (B — C — D)i (i = 1, 2) is of form G (X , Y) — (t- s)Q (X , Y), where
Q(X, Y) = U (X , Y )+ Ü [(X , Y) —TJ\(X, Y ) - U 22(X , Y) +
00 00
+ 2 ( V l i +i{Z , T )-U l< (X , Y)) + 2 ( ü î i ( Z , Y )~ U l,_ y(X , Y)).
i = l г'=2
P roof. The function Q(X, Y) is the Green function for the first Fourier problem of the equation
(lc) Pu{X ) = 0
in the domain 8 X = {(x ,t): —a < x < + a , t > 0 } ([1], p. 475). Using
the estimations analogous to those in [1], pp. 476-477, we can easy verify
that the series
Q i(X, Y) = % ( V l2i(X , Y ) - V \ i+1(X , Y)),
г = 0
00
Q ,(X, Y) = 2 ( V l i ( X , Y ) - V l ^ X , Y))
г = 1
are quasi-uniformly convergent with the derivatives to the order 4 for X e S x and
2/e(
— oo,+
oo),0 < s < t. From the equality P XG = - Q follows that the function Q satisfies equation (lb) and P x G(X, Y) — 0.
For Y ea2 we get |TX L-il = № Xl{\ and \ТХ\{\ = bence Qx
= Q2 = 0. For Y eax we have |Y X l2i\ = № Х\{_г\ and \YX22i\ = |Г Х *<+1|
hence Qx = Q2 = 0, and we obtain P x G (X , F) = —Q(X, Y) = 0 for
~Y € dj U ^2 •
4. Using some lemmas dealing with convergence of certain integrals and series and Green function G we shall prove that the function
7
(8) Щ Х ) = У Ч ( Х ) ,
t = l
where
-f-a 4-ct
иг(Х) = f M y)Q {X , y, 0)dy, u2{X) = f h(y)tQ {X,y, 0)dy,
—a —à
where Л(у) = f i (y) — I>tu1(y10)t
t t
щ (Х ) = J f a(s)J)vQ (X f a, s)ds, щ = - J / 4(s)(i- s )i> 3/Ç (X , a, e)<fe,
о 0
t
u5{X) = J / 5(e)-D,,Q(X, - a ,8 )d s ,
оt
u6(X) = J f e(s)(t — s)DyQ{X, —a, s)ds
0
and
1 +a
M ,m = / / f ,(y ,s )G (X , Y)dyds
0 - a
solves in $ the (В — C — D)x problem.
Let
where t
0and T are
a r b it r a r ypositive numbers. Let
OQF J X , Y) = £ ( Щ (Х , Y ) - U î i±l(X , Y)) (n =
1 , 2), 1OÛ
(9)
H J X , Y)
= Y)| + |î7?j±1( X , Г)|) ( » = 1 ,2 ) , i= 1oo
(10) H ^ ( X , Y) = JT ’ I.D Y>+ иш ( Х , Г))| (n = 1 ,2 ),
1 = 1where j , p = 0,1, 2, 3,4, A: = 0 , 1 , 2 , j + fc < 6.
We shall prove that the function 77(X) defined by formula (8) is of class Cjkp in 8. Moreover, the derivatives Dxjtk U l) exist and are continuous in
8.
Lemma 1.
The series defined by formulas
(9), (10)are uniformly con
vergent in every set V.
P roof. The common majorant of this series is the series C ] ? i " 2, where О is a positive constant ([1], p. 477). <=1
By Lemma 1 we get
Lemma 2.
There exist the derivatives D j k pF n(X,
T )continuous in V and
00
D ^ F J X , Y) = ^ Y )~ 0?,±1(X , Г)) (n = 1, 2).
S — 1
By Lemma 2 follows
Lemma
3. I f the functions f { (i —
1,..., 7) are continuous and bounded, then the functions щ (i = 1, ..., 7) are of class Cjkp in 8 and satisfy equa
tion
( l b ) .P roof. By Lemma 2 we can change the integration, differentiation and summation, and we get the first part of the thesis of Lemma 3. Now we shall prove that the functions щ (i = 1, 3) satisfy equation (lb).
For i = 2, 4, 5, 6 the proof is analogous. Since Q(X, Y) and (t — s)Q(X, Y) are the solutions of equation (lb) thus by Lemma 2 we obtain
-\-d
Р *«Л Х ) = / fi(y )F
2x Q {X , у ,
0)dy = 0
—a
and
tP u,(X ) = f f s(s)P
2x (Pv(t- s)Q (X , a, i))ds = 0.
0
5. In order to prove that the function 77 (defined by formula (8))
satisfies the limit conditions (2)-(7) we shall prove some lemmas.
Lem m a
4. I f the function / x is continuous and bounded in the interval ( —a, + a), then
lim%(X) = /i(a?0) as t)-*(œ0, 0+), a?0e (— a, -fa).
P roof. Let
-fa
1г(Х) = At~* j fi(y )e x p (- (y - œ )2(ét)-1)dy
—a
and
+ a oo
U X ) = A r * f h ( y ) { - V \ ( X , Y) + 2 ( V h ( X , X ) - V \ i_1(X , Г)) +
г=1 oo
+ ^ { U t i( X , Y ) - U l i_1(X ,Y )))d y,
i —1
for у
e < ( —a, -fa), for y c B \ ( - a , -fa>.
+ 0O
I ^ X ) = 4 Г { J A M e x p f - ^ - a O ^ ^ d y ^ / i K )
— 00
as (a?, t)-^(x0, 0+), x0e( — «» + « )•
The function Сг$ is a majorant for I 2(X) and I 2(X)->0 when (oo, t)->(a?0, 0+), Æ0e( — a, -fa), Oj being positive constant.
Lemma 5.
I f the functions h and
/ г- (г = 1, . . . , 7)are continuous and bounded, thenlimu^X) = 0 as (oo, t)-+(x0, 0+), a?0e (—a, -fa) (i = 2 , . . . , 7).
P roof. From Lemma 4 for i = 2 we get
-\~alimw2(X) = limtf J h(y)Q(X, у , 0)dy = limtt(ai0) = 0
—a
as (oo, t)->(x
q, 0+), and x0€(— a, -fa). The common majorant for u{ (i = 3, ..., 7) is the function C2(x0±_a)~mt*, where m and C2 are positive constants.
Let
T/ N I МУ) for У е ( - а , + a > , My) = { ( 0 for y e lt ,\ ( —a, -fa>.
Now we shall prove where A = (2(nÿ) h
Let
M y )
M y ) =
0By [1], p. 450, we obtain
Lem m a 6.
Let the function f x be of class C2 and bounded with two first derivatives for cue ( —a, -\-a) and
(11) D 2xux(X )->f" {œ0) as (sc, t)->(x0, 0+), x0e ( - a , + « );
then
Щ иг + и^-^^оОо) as (x, t)-+(x0, 0+), x0e {—a, -fa).
P roof. Since the function ux satisfies equation (lb) in the interior of Vx and by (11) we get
DfUx(X) = Dxux(X)~->fx (й/0) as (x^ t)— >(л?0, 0^), я?о€(
and by Lemma 4 and Weierstrass theorem ([1], p. 446) we obtain
+ o o +o o
Dtuz{X) = ( f h (y)Q (X ,y,0)dy + t j h(y)DtQ (X , у , 0)dy}
— OO — OO
4 (® o )~ /i(® o ) when (x, t-)^(x0, 0+), х0е ( ~ а , f a ) . Hence И т Н ^ ^ + ^г) = /
2(^
0) as (a?, t)->(®0, 0+), a?0e (— a, f a ) .
Lem m a7. I f the functions f { (i = 1, ..., 7) are continuous and bounded, then
limDfU^X) = 0 when (x, t)-+(x0, 0+), x0e {— a, f a ) ,
i — 1, 7; i Ф 2.
P roof. The common majorant of the integrals I>iui is the function Сг{®0± а ) ~ т $, where m and Cz are positive constants.
Now we shall prove that the function U(X) defined by formulas (8) satisfies the boundary conditions (4)-(7).
Let
K x(X , a, s) — {a — x){t — s)~4'3exp( — (a — æ)2(4(i — s))-1), and
Щ(-Х) = J х{ Х ) J 2{X ), where
t
Jy(X ) = A j f 3(s )K (X ,a ,s )d s
0and
t OO
J 2{X) = Л f f 3( s ) ( U 2i( X , Y) ^2i+l (X , Y)) +
о г—1
oo
+ y < U U X , Y ) - l 72
v,(X , Г)))| <fe.
» = 1
Let
J / 4 l/s(*) fo r 's€(0,t), f
3(s) —
l 0 for *€ .B \(0 ,t).
L
emma8. Let the function f
3be continuous and bounded; then
\im J
1(X) = f
3(t0) as (x ,t)-> {a ,t0), t
0> 0 and
lim J2(X) = 0 as (æ, t)-+(a, t0), t0> 0.
P roof. Since
-\-t
(12) J 3(X) — A J K x{X , a , s)ds = 1
—oo
thus
t tJ x(X) = A f d (s,t
0)X
1(X , a, s)ds + A J f M K ^ X , a, s)ds,
— OO —00
where •
d(s, t0) = f
3(s) f
3(t0).
By formula (12) we get
t
Ji( X ) fs(to) = A J*d (sj t
0)K 1(X j u, s)ds.
— OO
Let e be an arbitrary positive number. We shall prove that there exists a positive number Ô such that if is —
10\ < ô and \a — x\ < ô, then W i(X)-fAto)\ < 0 4e, where is a positive constant.
Indeed as f
3is continuous at the point t0, we have (13) \d{s,t0) \ < e for S€(t0— ô, t0+ ô).
Let
1
1= {(s)î — oo <i s <Z t
0— J 1
2— {(s) î t
3Ô <C
S<C to} , Is = {(*)•• t0 < S < t ] ,
and
3M (X) = y Mt(X ),
i= l
where
Mx(X) = A J d ( s , t
0)K x(X , a , s)ds,
M
2{X) = A j d { s , t
0)K x(X , a, s)ds,
and M
3(X) = A f d(s, t
0)K x(X , a, s)ds. Let G
6= sup|/3|.
Is
By virtue of (12) and (13) we have the estimations
\Мг(Х)\ < 2AC
5J X x(X , s)ds <
2AC
5j (a — x)(t — s)~ zds
h h
<5\ 23
< 2AC,
j (a — x) ds < 2AG&(a — x)t
0( B
< s
for x > \ — \ (
2AG
5at
0— e), and
|J f 2(X)| < 2As J K x(X , a, s)ds < As J К г(Х , a, s)ds = As,
J 2 —oo
and |ilf3(X)| < Xe.
Now we shall prove the second part of the thesis of Lemma 8, we have limD„{Ul((X , X) + V l ^ X , Г)) = limD ^ U ^ X , Y )+ U \M (X , T)) = 0,
•
as (x, t)->(a, t0), to>
0, and by uniform convergence of the integral uz at the point (a, t0) we get 1ш1«72(Х) = 0 as (x , t)->(a, t0).
By Lemma 8 we obtain
Lemma 9.
Let the function f z be continuous and bounded
;then 1шш3(Х) = / 3(t0) when (x, t)-+(a~, tQ), t
0> 0.
Now we shall prove
Т/
еутупуга10. I f the functions f { (i = 1 ,. . . , 6 ) are continuous and bounded, then
1шшг-(Х) = 0 as (x ,t)^ -(a ~ ,t
0) ,t 0> 0 for i = 1, ..., 6, i ф 3.
P roof. Since the integrals ui are uniformly convergent at the point (a, t0) and Q(a, t0, y, 0) = 0, then
lim%(X) = 0 as (x, t)-+(a~, t0), t
0> 0 for i — 1, 2.
The proof of conditions 1шшг-(Х) = 0 as (x, t)->(a~, t0), t0 >
for i — 4, 5, 6 is similar to the proof of the convenient condition in Lemma 8.
By uniform convergence of the integral u
7at the point (a ,t 0) and from the condition Q (a,t, y , s) = 0 for у Ф + a follows that
+ a t
lim «7( I ) = j J f
7{ y ,
8) ( t -
8)Q(œ,t
0, y ,
8)dyd
8 = 0—a 0
as (x , t)->(a~, t0).
Similarly to condition (4) we can verify condition (6).
Now we shall prove condition (5). The proof of condition (7) is the same.
Т
литупуга11. I f the functions f { (i = 1, ..., 7) are continuous and bounded, then HmPTJ(X) = / 4(<0) when (sc, t)->(a~, t0), t0> 0.
P roof. By (8) and Lemma 3 we get
7
P V (X ) = У Ч ( Х ) ,
i = 1
where
-f-a H-ct
zx(X) = f fx(y)Px Q (X , у, о)dy, z
2(X) = f h(y)Px tQ(X, y,0)dy,
—a —a
t
S*(X) = P x [ f f s ( s)Dy Q (X , a, s) dsj,
0 t
*A X) = P x (ffi(s)I> y(t - s )Q (X , a, s)ds}, '
0
t
z
5(X) = P x ^ f b(s)DvQ(X,
0 t
z
6(X) = P x ( f f e(s)B v( t -
8)Q(X, - a , «)<&),
0+ a t
z
7(X) = P X ( J j f
7( y ,
8) ( t -
8)Q (X , Y)dydsy
—a 0
By Theorem 1 we get P x Q(X, у , 0) = 0 and zx(X) = 0. From uni
form convergence of the integral z
2at the point (a, t0), t
0> 0 we obtain
+ a + a
lim«2(X) = lim J h(y)Px (tQ(X, y ,
0))dy - lim J h (y )(-Q (X , y, Q))dy
—a —a
= - / h(y)Q(a, t0, y, 0)dy = 0.
—a
Since
t
z
3(X) = j f
3(s)B
2xB yQ(X, Y )d s- lim f
3(s)B yQ(X, a, s ) -
0 s-><
t t
- Jfz (s)B iyQ (X ,a ,s)d s = j f
3(s)Px Q (X , a, s)ds
о 0
and lim fs(s)B y Q(X, a, s) = 0
and
DyQ{a, t
0, a, s) —
0, and P xQ(X, a, s) —
0, thus lim £3(X ) = 0 when(x, t)->(a~, t0), t0>
0 . F o r the integral z4 we havet
z
4(X) = - f f
4(s)(t-s)L>yP x Q (X ,a ,s)d s + limf
4(s)(t-s)I> yQ(X, a, s) +
о t
+
jf M D yQ(X, a,
8)d ss
оSince lim
Q(X, a, s)
=Q(a, t0, a, s)
= 0 when(со, t)'-+(a~, t0), t
0> 0, andP x Q ( a ,t ,a ,s )
= 0 and limf
4(s)(t — s)D yQ (a~,t, a, s)
= 0 ass->t.
B y Lem m a 8 we obtain
t
1н п г4(Х ) = lim j
f
4(s)DyQ(x, a, t, s)ds
=f
4(t0)
as(со, t)-+(a~, t0).
о
W e can prove sim ilarly as for
zi (i
= 3 , 4 ) thatИ т ^ - ( Х ) = 0 when
(со, t)-+(a~, t0), t
0> 0 fori
= 6, 6 . W e shall prove that И т ^7( Х ) = 0 as(x, t)->(a~, t0), t
0> 0.O f course since
Q(a, t, у , s)
= 0 thust + a
lim z7(.X) = — l i m j J
f
7(y, s)Q (X, Y)dyds
0 —a t + a
= f f М У, *)Q (a ,t
0,y , s)dyds,
0 —a
when
(x, t)->(a, t0), t
0> 0 . N ow we shall proveL
emma12. Let the function f
7be continuous and bounded', then P 2
u7(X) = - M X ) .
P r o o f .
t + a * t + a
Р гщ (Х ) = / f U y ,s ) D lQ ( X , Y )d y d s - 'j j f,(y ,s )D t Q(X, Y )dyds-
0 —a 0 —a
+a
- lim
f f
7(y, s)Q(X, Y)dy.
++* - a
Hence
+ a + a
Р ги,(Х) = f f,(y , s)P x Q(X, T )d y d s - lim / f,(y ,e )Q {X , Y)dyds
—a —a
= - л д а .
By Lemmas 3-12 we get
Theorem 2.
I f the functions f {
(i — 1, . . . , 7)satisfy the assumptions of Lemmas
3, 6,then the function U(X) defined by formula
(8)is the solu
tion of the problem (B — G — D )x.
6
. Let ns now consider equation (la) and
t +a
(Id) v(X) = f f f(y , s,v (y , s))G {X , Y)dyds.
0 —a
We shall nse the following
Lemma 13 ([3], p. 10).
I f the function f(Y , u(Y)) is continuous and bounded in the set jS and satisfies the Lipschitz condition with constant a < 1 uniformly with respect to Y eS, then there exists the solution v(X) of equa
tion (Id).
Lemma
14. Let the function f [ X , u{X)) satisfies the assumptions of Lemma 13) then the function v(X) satisfies the homogeneous boundary conditions (2)-(7) and equation (la).
The proof of Lemma 14 is similar to the proof of Lemmas 3, 5, 7, 10 and 12.
By Lemmas 13, 14 we obtain
Th eo rem 3.
I f the function f(Y ,u (Y j) satisfies the assumptions of Lemma
13,then the function u(X ) = u
1{X )-\-...Jr u%(X)-\-v{X
)1where u{
(i = 1, ..., 6) are defined by formula (8) is the solution of the problem (B — C — D)2.
References
[1] M. K rz y z a n sk i, Partial differential equations of second order, vol. I, Warszawa 1971.
[2] J . M ilew ski, Wybrane zagadnienia graniczne dla rôwnan parabolicznych rzçdôw wyzszych (unpublished).
[3] W. P o g o rz e lsk i, Bownania calkowe i ich zastosowania, t. II, Warszawa 1958.