MATHEMATISCHE

www.mn-journal.org

NACHRICHTEN
+(-. ,(,=1(7&1 A4(%1.-0 *(4 B'. A&4&%(1:7='3A.4%(1:7 B3A. .6>&B:(, 8:B'
7(-A1.5 0A.7B4&1 A&4&-.B.4
#9 )9 "&4:-(;/
/
N2@*I*&*5 03 B$*K5E$*I9@ $26 N230AE$*I02 :59K20F0LI5@7 +PQ5G )9$65ER 03 ;9I5295@7 "- CK06H$5% @*- '17 :$@KG52*J/'M7
+PQ5GI@*$2

<595I%56 /M D&25 'OOM7 A5%I@56 = D&25 'OO=7 $995.*56 '( D&25 'OO=
>&QFI@K56 02FI25 M D&25 'OO4

".3 8(420 BI!567 .$A$Q0FI9JKR.5AQ0FI9 *R.5 5,&$*I02@7 202JF09$F .A0QF5E@7 @.59*A$F .$A$E5*5A7 525ALR I2*5J
LA$F7 I2*5LA$F 5,&$*I02
!+$ ?@///< >AIE$AR8 (MB/O? ;59026$AR8 (M>OM
N2 *K5 .A5@52* .$.5A *K5 &2I,&5 @0F%$QIFI*R 03 *#0 202JF09$F .A0QF5E@ 30A *K5 EI!56 .$A$Q0FI9JKR.5AQ0FI9 *R.5
5,&$*I02 #I*K 90E.F5! @.59*A$F .$A$E5*5A I@ .A0%56- ;59*0A@ 30A %$F&5@ 03 *K5 @.59*A$F .$A$E5*5A #K5A5 *K5@5
.A0QF5E@ K$%5 &2I,&5 @0F&*I02@ $A5 @K0#2- +2I,&525@@ 03 *K5 @0F&*I02 I@ .A0%56 QR *K5 E5*K06 03 525ALR
I2*5LA$F $26 5!I@*5295 I@ .A0%56 QR *K5 E5*K06 03 I2*5LA$F 5,&$*I02@- N2 .$A*I9&F$A 9$@5@7 5IL52%$F&5@ $26
[email protected] 5IL523&29*I02@ 03 *K5 @*&6I56 .A0QF5E@ $A5 30&26-

Math. Nachr. 281, No. 7, 959 – 970 (2008) / DOI 10.1002/mana.200510652

NT
RI
EP
R
Math. Nachr. 281, No. 7, 959 – 970 (2008) / DOI 10.1002/mana.200510652

Some non-local problems for the parabolic-hyperbolic type equation with

complex spectral parameter
E. T. Karimov∗1
1
Institute of Mathematics and Information Technologies, Uzbek Academy of Sciences, F. Khodjaev st. 29, Tashkent-125,
Uzbekistan

Received 15 June 2005, revised 7 June 2007, accepted 23 June 2007

Published online 5 June 2008

Key words Mixed, parabolic-hyperbolic type equations, non-local problems, spectral parameter, energy inte-
gral, integral equation
MSC (2000) Primary: 35M10; Secondary: 35P05
In the present paper the unique solvability of two non-local problems for the mixed parabolic-hyperbolic type
equation with complex spectral parameter is proved. Sectors for values of the spectral parameter where these
problems have unique solutions are shown. Uniqueness of the solution is proved by the method of energy
integral and existence is proved by the method of integral equations. In particular cases, eigenvalues and
corresponding eigenfunctions of the studied problems are found.

c 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction
It is known that the theory of mixed type equations is one of the modern part of the theory of partial differential
equations. Recently a circle of problems for mixed type equations was considerably extended.
The theory of mixed parabolic-hyperbolic type equations is also one of the actual directions of the theory of
mixed type equations. For the first time the necessity of consideration of parabolic-hyperbolic type equation was
specified in 1959 by I. M. Gel’fand [5]. He gave an example, connected to the movement of gas in a channel,
surrounded by a porous environment: inside the channel the movement of the gas is described by the wave
equation, outside by the equation of diffusion. The basic bibliography about the history of the occurrence and
development of this subject can be found in the book of T. D. Djuraev [4].
After the seventieth of the last century, the study of the spectral properties of boundary-value problems for
mixed type equations was started. Interests to study these kinds of equations are explained with that, some
many-dimensional analogues of main boundary-value problems can be studied by reducing them (with the help
of Fourier transform) to the problems with spectral parameter.
We note basic works closely related to our work. They are the work of K. B. Sabytov [7] in which the
uniqueness of the Tricomi problem solution with real spectral parameter has been proved and the work of
A. S. Berdyshev [2, 3], where some spectral properties of local, non-local boundary-value problems for the
parabolic-hyperbolic type equations have been studied.
Non-local conditions, which were used in the present work, can be found in the work of A. K. Urinov [10] and
in the book of M. S. Salakhitdinov and A. K. Urinov [8].
In this work along with proving unique solvability of the stated problems, some sectors for values of spectral
parameters, where problems have unique solutions are shown. And in particular cases, eigenvalues and corre-
sponding eigenfunctions of the studied problems are found.

∗ e-mail: [email protected], Phone: +998 71 162 75 44, Fax: +998 71 162 73 57

c 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
960 Karimov: Non-local problems for the parabolic-hyperbolic type equation

2 The statement of problems

We consider the equation

Lu = su, (2.1)

where


uxx − uy , y > 0,
Lu =
uxx − uyy , y < 0,

s = λ at y > 0 and s = −µ at y < 0. Here λ, µ are given complex numbers.

Let Ω be a simple connected domain, located in the plane of independent variables x, y, at y > 0 sectionally
bounded AA0 , BB0 , A0 B0 (A(0, 0), B(1, 0), A0 (0, 1), B0 (1, 1)) and at y < 0 bounded with characteristics
AC : x + y = 0, BC : x − y = 1 of Equation (2.1).
We introduce the following notations:
• Ω1 = Ω ∩ {y > 0},
• Ω2 = Ω ∩ {y < 0},
• AB = {(x, y) : y = 0, 0 < x < 1},
• θ0 and θ1 are points of intersection of characteristics, which are outgoing from the point (x, 0) ∈ AB with
characteristics AC, BC, respectively, i.e.,
x x  
x+1 x−1
θ0 = ,− , θ1 = , .
2 2 2 2

Under a regular solution of Equation (2.1) in Ω, we understand a function u(x, y) ∈ C(Ω) ∩ C 1 (Ω) ∩
2,1
Cx,y (Ω1 )∩ C 2 (Ω2 ) satisfying Equation (2.1) in Ω1 ∪ Ω2 .
Problem S1 Find a regular solution of Equation (2.1) in Ω satisfying the conditions

u(x, y)|AA0 = ϕ1 (y), u(x, y)|BB0 = ϕ2 (y), o ≤ y ≤ 1, (2.2)

√ √
0, µ 0, µ
a1 (x)A0x [u(θ0 )] + b1 (x)A1x [u(θ1 )] + c1 (x)u(x, 0) = d1 (x), x ∈ AB. (2.3)

Problem S2 Find a regular solution of Equation (2.1) in Ω satisfying the condition (2.2) and
√
  √
 
1, µ d 1, µ d
a2 (x)A0x u(θ0 ) + b2 (x)A1x u(θ1 ) + c2 (x)uy (x, 0) = d2 (x), x ∈ AB. (2.4)
dx dx

Here ai (x), bi (x), ci (x) (i = 1, 2) are given real-valued functions, moreover a2i (x) + b2i (x) = 0, for all
√
x ∈ [0, 1] and ϕ1 (y), ϕ2 (y), di (x) are, generally speaking, complex-valued functions. Under µ we understand
the square root of the complex number µ, with | arg(µ)| ≤ π.
We note that the special case of Problem S1 (b1 (x) = 0, c1 (x) = 0) was studied in [6].
Below we give definitions and properties of some integral and integral-differential operators, which were
introduced and studied in the book of M. S. Salakhitdinov and A. K. Urinov [8].
 x  n

t−m ∂
Amx [f (x)] = f (x) −
n,µ
f (t) J0 µ (x − m)(x − t) dt,
m x−m ∂t
 x  1−n

x−m ∂
mx [f (x)] = f (x) +
Bn,µ f (t) J0 µ (t − m)(t − x) dt,
m t−m ∂x
  x
1 2
C0,µ
mx [f (x)] = sign(x − m) f 
(x) + µ f (t)J 1 [µ(x − t)] dt .
2 m

Here m, n = 0, 1, J k (z) = Γ(k + 1)(z/2)−k Jk (z), Jk (z) is the “k”-th order Bessel’s function of the first kind,
Γ(z) is Euler’s Gamma-function, see [1].


c 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com
Math. Nachr. 281, No. 7 (2008) 961

Theorem 2.1 If f (x) ∈ C[a, b], then for for all m ∈ [a, b], x ∈ [a, b] the following equalities hold:

An,µ
mx {Bmx [f (x)]} = f (x),
n,µ
Bn,µ
mx {Amx [f (x)]} = f (x),
n,µ
(2.5)
 x  x
0,µ 1,µ
Amx Bmt [f (t)] dt = f (t)J0 [µ(x − t)] dt, (2.6)
m m
 
d 0,µ
A1,µ
mx Bmx [f (t)] = sign(x − m)C0,µ
mx [f (x)]. (2.7)
dx
Here the function f (x) should belong to the class C 1 (a, b) for the equality (2.7).
The proof of this theorem can be found in [8].

3 The main functional relations

Lemma 3.1 If u(x, y) is a solution of Equation (2.1) in Ω1 satisfying the homogeneous condition (2.2) and
u(x, 0) = τ (x), 0 ≤ x ≤ 1, then
 1   1   
δ2
 eδx τ (x)ν(x) dx + eδx |τ  (x)|2 + (λ) − |τ (x)|2 dx = 0 (3.1)
0 0 2

holds in Ω1 .
Here δ is some real number, uy (x, 0) = ν(x), 0 < x < 1.
The proof of this lemma was given in [6].
Lemma 3.2 If u(x, y) is a solution of Equation (2.1) in Ω2 satisfying the condition (2.3) or (2.4), then
 x  1
√ √
α(x)τ (x) = 2d1 (x) + a1 (x) ν(t)J0 [ µ(x − t)] dt + b1 (x) ν(t)J0 [ µ(t − x)] dt
(3.2)
√0 √ x
− a1 (x)τ (0)J0 [ µx] − b1 (x)τ (1)J0 [ µ(1 − x)], 0 ≤ x ≤ 1,
√ √
0, µ 0, µ
β(x)ν(x) = −2d2 (x) + a2 (x)C0x [τ (x)] − b2 (x)C1x [τ (x)], 0 < x < 1, (3.3)
hold in Ω2 .
Here α(x) = a1 (x) + b1 (x) + 2c1 (x), β(x) = a2 (x) − b2 (x) − 2c2 (x).

P r o o f. The solution of the Cauchy problem for Equation (2.1) in Ω2 has the form (see [9]):


1 1 x+y
u(x, y) = [τ (x + y) + τ (x − y)] + ν(t)J0 µ((x − t)2 − y 2 ) dt
2 2 x−y

 (3.4)
√  x+y J1 µ((x − t)2 − y 2 )
µy
+ τ (t)
dt.
2 x−y (x − t)2 − y 2

We can find u(θ0 ) and u(θ1 ), using (3.4)

  x
 1 

1 x ∂ 1 x
u(θ0 ) = τ (x) + τ (t) J0 µt(t − x) dt + τ (0) − ν(t)J0 µt(t − x) dt.
2 0 t ∂x 2 2 0

Taking
√  x

0, µ x ∂
B0x [τ (x)] = τ (x) + τ (t) J0 µt(t − x) dt,
0 t ∂x
 x √  x

1, µ
B0t [f (t)] dt = f (t)J0 µt(t − x) dt
0 0

www.mn-journal.com

c 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
962 Karimov: Non-local problems for the parabolic-hyperbolic type equation

into account, we have

 √  √

1 0, µ
x
1, µ
u(θ0 ) = τ (0) + B0x [τ (x)] − B0t [ν(t)] dt . (3.5)
2 0

Similarly we can find

 √  1 √

1 0, µ 1, µ
u(θ1 ) = τ (1) + B1x [τ (x)] − B1t [ν(t)] dt . (3.6)
2 x

We find, by differentiating (3.5), (3.6) with respect to x,


d 1 d 0,√µ √
1, µ
u(θ0 ) = B [τ (x)] − B0x [ν(x)] , (3.7)
dx 2 dx 0x

d 1 d 0,√µ √
1, µ
u(θ1 ) = B [τ (x)] + B1x [ν(x)] . (3.8)
dx 2 dx 1x
√ √
0, µ 0, µ
By applying the integral operators A0x , A1x to both sides of (3.5), (3.6) and by taking account of (2.5), (2.6)
we get
√
  x
0, µ 1 √ √
A0x [u(θ0 )] = τ (0)J0 [ µx] + τ (x) − ν(t)J0 [ µ(x − t)] dt ,
2 0
√
  1
0, µ 1 √ √
A1x [u(θ1 )] = τ (1)J0 [ µ(1 − x)] + τ (x) − ν(t)J0 [ µ(t − x)] dt .
2 x
√ √
1, µ 1, µ
Similarly applying the operators A0x , A1x to both sides of (3.7), (3.8) and taking (2.5), (2.7) into account
we obtain
 
√
1, µ d 1  0,√µ 
A0x u(θ0 ) = C0x [τ (x)] − ν(x) ,
dx 2
 
√
1, µ d 1 √
0, µ

A1x u(θ1 ) = −C1x [τ (x)] + ν(x) .
dx 2
After substituting the obtained equalities into conditions (2.3), (2.4), one can easily get (3.2), (3.3), respectively.

At studying Problem S1 (resp. S2 ) the following cases are possible:

1) α(x) = 0, a1 (x) = 0, b1 (x) ≡ 0 (β(x) = 0, a2 (x) = 0, b2 (x) ≡ 0);
2) α(x) = 0, a1 (x) ≡ 0, b1 (x) = 0 (β(x) = 0, a2 (x) ≡ 0, b2 (x) = 0);
3) α(x) = 0, a1 (x) = 0, b1 (x) = 0 (β(x) = 0, a2 (x) = 0, b2 (x) = 0);
4) α(x) ≡ 0 (β(x) ≡ 0).

4 Uniqueness of the solution

4.1 Case No. 1
Lemma 4.1 Let α(x) = 0, a1 (x) = 0, b1 (x) ≡ 0 and the following conditions be valid:
 
√ a1 (x) a1 (x)
δ ≤ −2| ( µ)|, > 0, ≤ 0, 0 ≤ x ≤ 1. (4.1)
α(x) α(x)
Then
 1 
I= eδx τ (x)ν(x) dx ≥ 0 (4.2)
0

holds at d1 (x) ≡ 0.


c 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com
Math. Nachr. 281, No. 7 (2008) 963

P r o o f. Substituting (3.2) into (4.2) and representing a function J0 [◦] as (see [1])
 1
√ 1  − 1 √
J0 [ µ(x − t)] = 1 − ξ 2 2 cos µξ(x − t) dξ,
π −1

after some transformations we have

⎧
 ⎨ 1
1 2
1  
I= 1 − ξ 2 dξ [−a3 (x) − a3 (x)(δ − 2µ2 ξ)] eδ−2µ2 ξ Φ2j1 (x, ξ) dx
4π −1 ⎩ 0
j=1
2
 4

+ a3 (1)eδ−2µ2 ξ Φ2j1 (1, ξ) + +a3 (1)eδ+2µ2 ξ Φ2j1 (1, ξ) (4.3)
j=1 j=3
⎫
 1 4
 ⎬
+ [−a3 (x) − a3 (x)(δ + 2µ2 ξ)] eδ+2µ2 ξ Φ2j1 (x, ξ) dx .
0 ⎭
j=3

Here

a1 (x)
a3 (x) = ,
α(x)
 x
Φj1 (x, ξ) = Φj (t, ξ) dt,
0


±µ2 ξx Φ1 (x, ξ),
[ν11 (x, ξ) ± ν22 (x, ξ)] e =
Φ3 (x, ξ),


Φ2 (x, ξ),
[ν12 (x, ξ) ± ν21 (x, ξ)] e±µ2 ξx =
Φ4 (x, ξ),
νj1 (ξ, t) = νj (t) cos µ1 ξt, νj2 (ξ, t) = νj (t) sin µ1 ξt, j = 1, 2,
ν(x) = ν1 (x) + iν2 (x),
√
µ = µ1 + iµ2 .

From (4.3) one can readily see that when conditions (4.1) are valid the inequality (4.2) holds.

Theorem 4.2 Let the conditions of Lemma 4.1 hold. If

√
(λ) ≥ 2( ( µ))2 , (4.4)

then Problem S1 cannot have more than one solution.

P r o o f. Let u(x, y) be a solution of the homogeneous Problem S1 . Then according to Lemma 4.1 the in-
√
equality (4.2) holds at δ = −2| ( µ)|.
√
Setting δ = −2| ( µ)| in (3.1) we get
 1  1
 2
 √ 
I+ e |τ (x)| dx +
δx
(λ) − 2( ( µ))2 eδx |τ (x)|2 dx = 0.
0 0

Here one can see if (4.2) and (4.4) hold, then every item will be nonnegative, hence they should be equal to zero.
From here we have τ (x) ≡ 0. It is known that the problem for Equation (2.1) in Ω1 with homogeneous condition
(2.2) and u(x, 0) = τ (x) = 0, called the first boundary-value problem, has only the trivial solution. Hence
u(x, y) ≡ 0 in Ω1 . Since u(x, y) ∈ C(Ω), we can conclude u(x, y) ≡ 0 in Ω.

www.mn-journal.com

c 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
964 Karimov: Non-local problems for the parabolic-hyperbolic type equation

4.2 Case No. 2

Lemma 4.3 Let α(x) = 0, a1 (x) ≡ 0, b1 (x) = 0. If
 
 √  b1 (x) b1 (x)

δ≥2 µ ,  > 0, ≥ 0, 0 ≤ x ≤ 1, (4.5)
α(x) α(x)
then the inequality (4.2) holds at d1 (x) ≡ 0.
The proof of this lemma will conduct similarly to the proof of Lemma 4.1.
Theorem 4.4 Let conditions (4.4) and (4.5) be fulfilled. If there exists a solution of Problem S1 , then it is
unique.
The theorem will be proved similarly as Theorem 4.2, using Lemmas 3.1 and 4.3.

4.3 Case No. 3

√ 
Lemma 4.5 Let α(x) = 0, a1 (x) = 0, b1 (x) = 0. If µ = µ2 = 0 and
   
a1 (x) b1 (x) a1 (1) b1 (0)
≤ 0, ≥ 0, + ≥ 0, 0 ≤ x ≤ 1, (4.6)
α(x) α(x) α(1) α(0)
then the inequality (4.2) holds at d1 (x) ≡ 0.
This lemma can be proved using the proofs of Lemmas 4.1 and 4.3 simultaneously.
Theorem 4.6 If conditions of Lemma 4.5 are fulfilled and (λ) ≥ 0, then the Problem S1 cannot have more
than one solution.
2

 setting (λ) − 2 ≥ 0, we get τ (x) ≡ 0. Here we

δ
P r o o f. From (3.1) taking Lemma 4.5 into account √and
should put δ = 0, because, if δ > 0, then in (4.1) at µ = 0 the following conditions should be fulfilled
a3 (1) ≥ 0, δa3 (x) ≤ 0, a3 (x) ≤ 0, x ∈ [0, 1],
in order to fulfill inequality (4.2). It is not difficult to see, that the function, satisfying these conditions, is equal
to zero, which is not possible.
If δ < 0, then we can conclude from (4.5) that when the following conditions
b3 (0) ≥ 0, δb3 (x) ≥ 0, b3 (x) ≥ 0, x ∈ [0, 1],
b1 (x)
are valid, where b3 (x) = the inequality (4.2) holds. But for δ < 0, the function, which satisfies the above
α(x) ,
given conditions, identically equals zero.
Hence, δ = 0. It means that we get τ (x) ≡ 0 at (λ) ≥ 0. Taking the homogeneous conditions (2.2) and
u(x, y) ∈ C(Ω) into account, we obtain u(x, y) ≡ 0 in Ω (see the proof of Theorem 4.2).

4.4 Case No. 4

Theorem 4.7 Let α(x) ≡ 0, let (λ) ≥ 0 and let one of these conditions be valid:
1) a1 (x) = 0, b1 (x) ≡ 0;
2) a1 (x) ≡ 0, b1 (x) = 0;
3) a1 (x) = 0, b1 (x) = 0, a1 (x) = b1 (x) and
 1 1
|K(x, t, µ)|2 dx dt < 1, (4.7)
0 0
where
⎧
⎪
⎪ ∂ √
⎨ {a1 (x)J0 [ µ(x − t)]} /[a1 (x) − b1 (x)], x ≥ t,
∂x
K(x, t, µ) =
⎪
⎪ ∂ √
⎩ {b1 (x)J0 [ µ(t − x)]} /[a1 (x) − b1 (x)], x ≤ t.
∂x
Then, if a solution of Problem S1 exists, it is unique.


c 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com
Math. Nachr. 281, No. 7 (2008) 965

The proof of this theorem can be conducted as in the work [8].

Below we give uniqueness theorems for Problem S2 without proofs, because the proofs can be conducted
similarly to Problem S1 , only with some changes.
Case No. 1: β(x) = 0, a2 (x) = 0, b2 (x) ≡ 0.
Theorem 4.8 Let the condition (4.4) hold and let
 
 √  a2 (x) a2 (x)

δ ≤ −2 µ ,  > 0, ≤ 0, 0 ≤ x ≤ 1.
β(x) β(x)

If a solution of Problem S2 exists, then it is unique.

Case No. 2: β(x) = 0, a2 (x) ≡ 0, b2 (x) = 0.
Theorem 4.9 Let condition (4.4) hold and let
 
 √  b2 (x) b2 (x)

δ≥2 µ ,  < 0, ≤ 0, 0 ≤ x ≤ 1.
β(x) β(x)

If the Problem S2 has a solution, then it is unique.

Case No. 3: β(x) = 0, a2 (x) = 0, b2 (x) = 0.
Theorem 4.10 If (λ) ≥ 0 and
   
√ a2 (x) b2 (x) a2 (1) b2 (0)
( µ) = 0, ≤ 0, ≤ 0, − ≥ 0, 0 ≤ x ≤ 1,
β(x) β(x) β(1) β(0)

then Problem S2 cannot have more than one solution.

Case No. 4: β(x) ≡ 0.
Theorem 4.11 Let (λ) ≥ 0 and let one of these conditions be valid:
1) a2 (x) = 0, b2 (x) ≡ 0;
2) a2 (x) ≡ 0, b2 (x) = 0;
3) a2 (x) = 0, b2 (x) = 0, a2 (x) = −b2 (x) and
 1 1
|N (x, z, µ)|2 dx dz < 1,
0 0

where
⎧ √  1
⎪ a2 (x)J1 [ µ(x − t)]
⎪
⎪ √ µ dt, x ≥ t,
⎨ a2 (x) + b2 (x)µ(x − t)
z
N (x, z, µ) =  1 √
⎪
⎪ b2 (x) J1 [ µ(t − x)]
⎪
⎩− µ √ dt, x ≤ t,
a2 (x) + b2 (x) z µ(t − x)

then, if a solution of Problem S2 exists, it is unique.

Corollary 4.12 Let λ = µ. If
1) one of the conditions (4.1), (4.5) is valid, then Problem S1 can have eigenvalues only outside of the domain
 √ 
D1 = λ : |(λ)| ≥ | (λ)|/ 3 ;

2) the condition (4.6) or one of the conditions of Theorem 4.7 is valid, then Problem S1 can have eigenvalues
only outside of the domain

D2 = {λ : (λ) ≥ 0} .

www.mn-journal.com

c 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
966 Karimov: Non-local problems for the parabolic-hyperbolic type equation

Corollary 4.13 Let λ = µ. If

1) the conditions of Theorem 4.8 or 4.9 are fulfilled, then Problem S2 can have eigenvalues only outside of
the domain D1 ;
2) the conditions of Theorem 4.10 or 4.11 are realized, then Problem S2 can have eigenvalues only outside
of the domain D2 .
Below we give some examples about the existence of eigenvalues and corresponding eigenfunctions of the
stated problems in particular cases. For the representation of eigenfunctions in the domain Ω1 we use Green’s
function of the first boundary-value problem for Equation (2.1), represented as

∞  
1 (x−ξ−2n)2
−λ(y−η)− 4(y−η)
(x+ξ−2n)2
−λ(y−η)− 4(y−η)
G(x, y, ξ, η) =
e −e . (4.8)
2 π(y − η) n=−∞

Note that the eigenvalues are located outside of the domain D2 .

Example 4.14 Let b1 (x) = 0, and let c1 (x) = 0. Then µ = 0, λn = − 14 − (nπ)2 , n = 1, 2, . . . , are
eigenvalues of Problem S1 and the corresponding eigenfunctions are representable as
⎧ 1
⎪
⎨ exp(t/2) sin nπtG(x, y, t, 0) dt, y > 0,
un (x, y) = 0
⎪
⎩
exp((x + y)/2) sin nπ(x + y), y < 0.

Example 4.15 Let a1 (x) = 0 and let c1 (x) = 0. Then µ = 0, λn = − 14 − (nπ)2 , n = 1, 2, . . . , are
eigenvalues of Problem S1 and the corresponding eigenfunctions can be written in the form
⎧ 1
⎪
⎨ exp(−t/2) sin nπtG(x, y, t, 0) dt, y > 0,
un (x, y) = 0
⎪
⎩
exp((x − y)/2) sin nπ(x − y), y < 0.

Example 4.16 Let

a1 (x) = m1 c1 (x), b1 (x) = m2 c1 (x), a2 (x) = p1 c2 (x), b2 (x) = p2 c2 (x).

2
Then µ = 0, λn = − q4 − (nπ)2 , n = 1, 2, . . . , are eigenvalues of Problems S1 and S2 , and the corresponding
eigenfunctions have the form
⎧ 1
⎪
⎪
⎨ exp(qt/2) sin nπtG(x, y, t, 0) dt, y > 0,
un (x, y) = 0
⎪
⎩ 1 {(1 + q) exp(q(x + y)/2) sin nπ(x + y) + (1 − q) exp(q(x − y)/2) sin nπ(x − y)},
⎪
y < 0,
2
where
m 1 + m2 + 2 p1 + p2
q= = , m1 = m2 , p1 − p2 = 2, mi , pi (i = 1, 2) ∈ R\{0}.
m1 − m2 p1 − p2 − 2

5 Existence of the solution

5.1 Case No. 1
Theorem 5.1 Let the conditions of Lemma 4.1 and the condition (4.4) hold, let a1 (x), c1 (x), d1 (x) ∈
C 1 [0, 1] ∩ C 2 (0, 1), and let ϕ1 (y), ϕ2 (y) ∈ C 1 [0, 1]. Then the regular solution of Problem S1 exists.

c 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com
Math. Nachr. 281, No. 7 (2008) 967

P r o o f. Passing to the limit in Equation (2.1) at y → +0, from the parabolic part of domain Ω we have

τ  (x) − ν(x) − λτ (x) = 0, (5.1)

and the condition (2.2) will pass to

τ (0) = ϕ1 (0), τ (1) = ϕ2 (0). (5.2)

The solution of Equation (5.1) satisfying condition (5.2) has form (see [4])
 1
τ (x) = r(x) + G∗ (x, t, λ)ν(t) dt, (5.3)
0

where
 1
r(x) = ϕ1 (0) + x[ϕ2 (0) − ϕ1 (0)] + λ {ϕ1 (0) + t[ϕ2 (0) − ϕ1 (0)]} G∗ (x, t, λ) dt,
0
⎧ √ √
⎪
⎪ sh λ(1 − t)sh λx
⎪
⎨ √ √ , 0 ≤ t ≤ x,
∗ λsh λ
G (x, t, λ) = √ √
⎪
⎪ sh λtsh λ(1 − x)
⎪
⎩ √ √ , x ≤ t ≤ 1.
λsh λ
We get from (3.2)
 x
2d1 (x) a1 (x) √ √
τ (x) = + ν(t)J0 [ µ(x − t)] dt − τ (0)J0 [ µx] . (5.4)
α(x) α(x) 0

By differentiating (5.3) and (5.4) with respect to x once and considering τ  (x), taking account of τ (0) = ϕ1 (0),
we have
 1
ν(x) − ν(t)K1 (x, t, λ, µ) dt = k1 (x), (5.5)
0

where

     
α(x)  2d1 (x) √ a1 (x) √ a1 (x) √
k1 (x) = r (x) − − ϕ1 (0) µ J1 [ µx] − J0 [ µx] ,
a1 (x) α(x) α(x) α(x)

√ 1 √ sh√λ(1 − t) sh√λx −  a (x)  J √µ(x − t) − √µ a (x) J √µ(x − t),

1

1

α(x)
 λ sh λ α(x) α(x)
0 1

a (x)

K1 (x, t, λ, µ) = 0 ≤ t ≤ x,
1

 √ 1 √ sh√λt sh√λ(1 − x), x ≤ t ≤ 1.
λ sh λ
The unique solvability of the integral equation (5.5) follows from the uniqueness of the solution of Problem S1 ,
since Problem S1 is equivalently reduced to the integral equation (5.5).
And so Problem S1 in the domain Ω1 is equivalent to the first boundary-value problem for Equation (2.1) and
this solution is representable as
 1  y   y 
∂G  ∂G 
u(x, y) = G(x, y, ξ, 0)τ (ξ) dξ + ϕ1 (η) dη − ϕ2 (η) dη, (5.6)
0 0 ∂ξ ξ=0 0 ∂ξ ξ=1

where G(x, y, ξ, η) is defined by formula (4.8). And the solution of Problem S1 in Ω2 will be determined by
formula (3.4).
Under the conditions imposed on the given functions a1 (x), c1 (x), d1 (x), from the continuity and limitation
of Bessel’s function of the first kind and also from the properties of the function G∗ (x, t, λ), we can conclude
that there exists a regular solution of Problem S1 .

www.mn-journal.com

c 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
968 Karimov: Non-local problems for the parabolic-hyperbolic type equation

Theorem 5.2 Let the conditions of Theorem 4.8 hold. If a2 (x), c2 (x), d2 (x) ∈ C[0, 1] ∩ C 1 (0, 1), ϕ1 (y),
ϕ2 (y) ∈ C 1 [0, 1], then a regular solution of Problem S2 exists.
P r o o f. We obtain from (3.3) that
  √

x
J1 [ µ(x − t)]
β(x)ν(x) = −2d2 (x) + a2 (x) τ (x) + µ τ (t) √ dt .
0 µ(x − t)

We substitute the expression of function τ (x) by formula (5.3) into the obtained equality and after some trans-
formations we have
 1
ν(x) + ν(t)N1 (x, t, λ, µ) dt = n(x), (5.7)
0

where
  √
a2 (x)  J1 [ µ(x − t)]
x
2d2 (x)
n(x) = r (x) + r(t) √ dt − ,
β(x) 0 µ(x − t) a2 (x)
  x √
a2 (x) ∂ µJ1 [ µ(x − z)]
N1 (x, t, λ, µ) = − G∗ (x, t, λ) + G∗ (x, t, λ) √ dz .
β(x) ∂x 0 µ(x − z)
It follows from the uniqueness of the solution of the problem that Equation (5.7) is uniquely solvable.
One can easily check that n(x) ∈ C[0, 1] ∩ C 1 (0, 1), N1 (x, t, λ, µ) is continuous and continuously differen-
tiable at x = t. We can conclude from these that τ (x) ∈ C 1 [0, 1] ∩ C 2 (0, 1), ν(x) ∈ C 1 (0, 1), i.e., there exists a
regular solution of Problem S2 .

5.2 Case No. 2

Theorem 5.3 Let the conditions (4.4), (4.5) be fulfilled and let b1 (x), c1 (x), d1 (x) ∈ C 1 [0, 1] ∩ C 2 (0, 1).
Then there exists a regular solution of Problem S1 .
Theorem 5.4 Let the conditions of Theorem 4.9 hold. If b2 (x), c2 (x), d2 (x) ∈ C 1 [0, 1] ∩ C 2 (0, 1), then there
exists a regular solution of Problem S2 .
The proofs of these theorems are similar to the proofs of Theorems 5.1 and 5.2.

5.3 Case No. 3

Theorem 5.5 Let the conditions of Theorem 4.6 hold. If a1 (x), b1 (x), c1 (x), d1 (x) ∈ C 2 [0, 1] ∩ C 3 (0, 1),
then there exists a regular solution of Problem S1 .
P r o o f. In this case, (3.2) looks like to
 1
d3 (x)
τ (x) = + ν(t)K2 (x, t, µ) dt, (5.8)
α(x) 0

where
√ √
d3 (x) = 2d1 (x) − a1 (x)τ (0)J0 [ µx] − b1 (x)τ (1)J0 [ µ(1 − x)],
⎧
⎪ a1 (x) √
⎨ α(x) J0 [ µ(x − t)], x ≥ t,
⎪
K2 (x, t, µ) =
⎪
⎩ b1 (x) J0 [√µ(t − x)], x ≤ t.
⎪
α(x)
By differentiating (5.8) twice with respect to x and substituting into (5.1), after some calculations we have
 1
ν(x) − ν(t)K2∗ (x, t, µ) dt = k2 (x), (5.9)
0

c 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com
Math. Nachr. 281, No. 7 (2008) 969

where
 
d3 (x) d3 (x) ∂2
k2 (x) = −λ , K2∗ (x, t, µ) = K2 (x, t, µ) − λK2 (x, t, µ).
α(x) α(x) ∂x2

As the problem has unique solution, Equation (5.9) is unique solvable and it is not difficult to see that u(x, y) ∈
C(Ω) ∩ C 1 (Ω) ∩ Cx,y
2,1
(Ω1 ∪ AB) ∩ C 2 (Ω2 ).

Theorem 5.6 Let the conditions of Theorem 4.10 be valid. If a2 (x), b2 (x), c2 (x), d2 (x) ∈ C[0, 1] ∩ C 1 (0, 1),
then Problem S2 has a regular solution.

P r o o f. Acting like in Theorem 5.2 we have

 1
ν(x) = l(x) + l(t)RL (x, t, λ) dt,
0

where
 1
2d2 (x) a2 (x) + b2 (x)  1
l(x) = − + r (x) + r(t)L0 (x, t, µ) dt,
β(x) β(x) β(x) 0

RL (x, t, λ, µ) is the resolvent of the kernel

 1
a2 (x) + b2 (x) ∂ ∗ 1
L(x, t, λµ) = G (x, t, λ) − G∗ (x, z, λ)L0 (x, z, µ) dz,
β(x) ∂x β(x) 0
⎧ √
⎪ J1 [ µ(x − t)]
⎪
⎨a2 (x)µ √µ(x − t) ,
⎪ x ≥ t,
L0 (x, t, µ) = √
⎪
⎪ [ µ(t − x)]
⎪−b2 (x)µ J1√
⎩ , x ≤ t.
µ(t − x)
We can conclude that, when the conditions of the theorem hold, Problem S2 has a regular solution.

5.4 Case No. 4

Theorem 5.7 Let (λ) ≥ 0 and let one of the following conditions be valid:
1) a1 (x) = 0, b1 (x) ≡ 0, a1 (x), d1 (x) ∈ C[0, 1] ∩ C 1 (0, 1), a1 (0)ϕ1 (0) = 2d1 (0);
2) a1 (x) ≡ 0, b1 (x) = 0, b1 (x), d1 (x) ∈ C[0, 1] ∩ C 1 (0, 1), b1 (1)ϕ2 (0) = 2d1 (1);
3) a1 (x) = 0, b1 (x) = 0, a1 (x) = b1 (x), a1 (x), b1 (x), d1 (x) ∈ C 1 [0, 1] ∩ C 2 (0, 1) and condition (4.7).
Then a regular solution of Problem S1 exists.
Theorem 5.8 Let (λ) ≥ 0 and let one of the following conditions hold:
1 √ √
1) a2 (x) = 0, b2 (x) ≡ 0, a2 (x), d2 (x) ∈ C 1 [0, 1], 0 ad22 (t)
(t) J0 [ µ(1 − t)] dt = ϕ2 (0) − ϕ1 (0)J0 [ µ];
1 d2 (t) √
2) a2 (x) ≡ 0, b2 (x) = 0, b2 (x), d2 (x) ∈ C 1 [0, 1], J [ µt] dt
0 a2 (t) 0
= − ϕ12(0) ;

3) a2 (x) = 0, b2 (x) = 0, a2 (x) = −b2 (x), a2 (x), b2 (x), d2 (x) ∈ C 1 [0, 1] and
 1  1
|N (x, z, µ)|2 dx dz < 1.
0 0

Then Problem S2 has a regular solution.

In this case (3.2) and (3.3) will decompose, the function ν(x) can be found easily, and then the function τ (x)
will be defined by formula (5.3).
The solutions of Problems S1 and S2 will be defined in Ω1 by formula (5.6) and in Ω2 by formula (3.4).
www.mn-journal.com

c 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
970 Karimov: Non-local problems for the parabolic-hyperbolic type equation

Acknowledgements The author would like to express his gratitude towards Professor Heinrich Begehr and A. S. Berdyshev
for their useful remarks in preparing of this paper.

References
[1] H. Bateman and A. Erdélyi, Higher Transcendental Functions, Volume 2 (Nauka, Moscow, 1966).
[2] A. S. Berdyshev, On non-local boundary-value problems with deviation from a characteristic for the parabolic-
hyperbolic equation, Izv. Akad. Nauk UzSSR, Ser. Phys.-math. Nauk 3, 14–18 (1989).
[3] A. S. Berdyshev, Basis property of the system of radical functions of the non-local boundary-value problem for the
parabolic-hyperbolic equation, Dokl. Akad. Nauk 366, No. 1, 7–9 (1999).
[4] T. D. Djuraev, A. Sopuev, and M. Mamajonov, Boundary-value Problems for the Parabolic-hyperbolic Type Equations
(Fan, Tashkent, 1986).
[5] I. M. Gel’fand, Some questions of analysis and differential equations, Uspekhi Mat. Nauk Ser. 3 XIV, No. 87, 3–19
(1959); Amer. Math. Soc. Transl. (2) 26, 173–200 (1963).
[6] E. T. Karimov, About the Tricomi problem for the mixed parabolic-hyperbolic type equation with complex spectral
parameter, Complex Var. Theory Appl. 56, No. 6, 433–440 (2005).
[7] K. B. Sabytov, To the theory of mixed parabolic-hyperbolic type equations with a spectral parameter, Differentsial’nye
Uravneniya 25, No. 1, 117–126 (1989).
[8] M. S. Salakhitdinov and A. K. Urinov, Boundary-value Problems for Mixed Type Equations with a Spectral Parameter
(Fan, Tashkent, 1997).
[9] A. N. Tikhonov and A. A. Samarskij, Equations of Mathematical Physics (Nauka, Moscow, 1966).
[10] A. K. Urinov, About the uniqueness of the solution of some non-local problems for the general equation of Tricomi with
complex spectral parameter, Uzbek. Mat. Zh. 3, 83–91 (1993).

c 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mn-journal.com