Fen Bilimleri Dergisi, Cilt 32, No. 1, (2011)
On Jost Solution of Diffusion Equation With Discontinuous
Coefficient
Yas¸ar C¸ AKMAK∗and Seval KARACAN Department of Mathematics, Faculty of Science, Cumhuriyet University, SIVAS 58140, TURKEY
Received 08.04.2011; accepted 22.06.2011
Abstract. In this work, integral representation of Jost solution of a diffusion equation is obtained and the properties of representation of the kernel are studied.
AMS subject classifications: Primary 34A55, Secondary 34B24, 34L05 Key words: Diffusion equation, Integral representation, Jost solution
1. Introduction
An important role in the spectral theory of linear operators was played by the trans-formation operator. Marchenko [3] , [5] first applied the transtrans-formation operator to the solution of the inverse problem. Transformation operator was also used in the fundamental paper of Gelfand and Levitan [4] . Yurko [11] studied the inverse problem theory for Sturm Liouville operators on the half-line.
−y00+ q (x) y = λy, x > 0 has a unique solution satisfying the integral equation
e (x, ρ) = exp (iρx) − 1
2iρ +∞Z
x
(exp (iρ (x − t)) − exp (iρ (t − x))) q (t) e (t, ρ) dt. where ρ ∈©ρ : Imρ ≥ 0, ρ 6= 0, λ = ρ2ª. The function e (x, ρ) is called the Jost so-lution [11] .
Huseynov [14] has considered the differential equation
−y00+ q (x) y = λ2y, x ∈ (0, a) ∪ (a, +∞)
with discontinous conditions y (a − 0) = αy (a + 0) , y0(a − 0) = α−1y0(a + 0),
where α 6= 1, α > 0, λ is a complex valued function and satisfies the condition
xq (x) ∈ L1(I) . Then for all λ from the upper half-plane, the Jost solution is unique and it is in the form
e (x, λ) = e0(x, λ) + +∞Z
x
K (x, t) eiλtdt.
∗Corresponding author. Email addresses: ycakmak@cumhuriyet.edu.tr
Let’s consider the diffusion equation with discontinous coefficient
−y00+ [2λp (x) + q (x)] y = λ2ρ (x) y, x ∈ (0, a) ∪ (a, +∞) = I. (1.1) we assume that
(1 + x) q (x) , p (x) ∈ L1(I) , (1.2)
p (x) ∈ BC (I) , (1.3)
where L1(I) is the space of integrable functions and BC (I) denotes bounded, con-tinuous functions on I,
λ is a complex parameter and ρ (x) is a step function: ρ (x) =
½
1, 0 ≤ x ≤ a
α2, a < x ≤ π , 0 < α 6= 1. (1.4) The function e (x, λ) , satisfying the equation (1.1) and condition at infinity lim
x→∞e (x, λ) e
−iλx = 1 is said to be Jost solution of (1.1). It is easy to show that if
q (x) ≡ 0, then the Jost solution is the function e0(x, λ) = ( R1(x) eiλx , x > a α+R 2(x) eiλµ +(x) + α−R 3(x) eiλµ −(x) , 0 < x < a (1.5) where µ±(x) = ±pρ (x)x + a³1 ∓pρ (x)´and α±=1 2 µ α ± 1 α ¶ .
2. Main Result
The main result of the present paper is the following.
Theorem. If a complex-valued function q (x) fulfilled condition (1.2), then there exists the Jost solution e (x, λ) of equation (1.1) for all λ from the upper half-plane, it is unique and representable in the form
e (x, λ) = e0(x, λ) + +∞Z
µ+(x)
K (x, t) eiλtdt, (2.1) where, for each fixed x ∈ (0, a) ∪ (a, +∞), the kernel K (x, .) belongs to the space
L1(µ+(x) , ∞). Moreover, K(x, t) is countinous with partial derivatives
∂K(x, t) ∂x , ∂K(x, t)
∂t for t 6= µ
−(x) and it satisfies the following properties
a)+∞R µ+(x) |K (x, t)| dt ≤ eσ(x)−1, σ (x) =+∞R x [2 |p (t)| + (1 + |t|) |q (t)|] dt (2.2) b)α+R00 2(x) + 2iλαα+R 0 2(x) − 2λα+p (x) R2(x) +2ip (x) K (x, µ+(x))−α+q (x) R 2(x)−2αdK (x, µ +(x)) dx = 0 (2.3)
c)α−R00 3(x) − 2iλαα−R 0 3(x) − 2λα−p (x) R3(x) −2ip (x) K (x, µ−(x) − 0) + 2ip (x) K (x, µ−(x) + 0) −α−q (x) R 3(x)−2αdK (x, µ −(x) − 0) dx +2α dK (x, µ−(x) + 0) dx = 0 (2.4) d)∂ 2K (x, t) ∂x2 −α 2∂2K (x, t) ∂t2 +2ip (x) ∂K (x, t) ∂t = q (x) K (x, t) (2.5) e) lim x+t→∞ ∂K (x, t) ∂x = 0, x+t→∞lim ∂K (x, t) ∂t = 0 (2.6)
Here e0(x, λ) is in the form of (1.5) which is shown above.
e (x, λ) is in the form e (x, λ) = e0(x, λ) + +∞Z x S0(x, t, λ) [2λp (t) + q (t)] e (t, λ) dt, (2.7) where S0(x, t, λ) = sin λ (x − t) λ , a < x < t α+sin λ (µ+(x) − t) λ + α −sin λ (µ−(x) − t) λ , x < a < t sin λα (x − t) λα , x < t < a (2.8) Proof:
Substituting (2.1) into (2.7) , we get
e0(x, λ) + +∞R µ+(x) K (x, t) eiλtdt = e 0(x, λ) ++∞R x S0(x, t, λ) [2λp (t) + q (t)] Ã e0(t, λ) + +∞R µ+(t) K (t, ξ) eiλξdξ ! dt.
Firstly, for x > a, we arrive +∞R x K (x, t) eiλtdt + (R 1(x) − 1) eiλx= =+∞R x S0(x, t, λ) (2λp (t) + q (t)) µ R1(t) eiλt+ +∞R t K (t, ξ) eiλξdξ ¶ dt
Now, if one evaluates the right-hand side of above equation, then it follows that +∞R x S0(x, t, λ) [2λp (t) + q (t)] R1(t) eiλtdt + +∞R x S0(x, t, λ) [2λp (t) + q (t)] µ+∞ R t K (t, ξ) eiλξdξ ¶ dt
=+∞R x sin λ (x − t) 2p (t) R1(t) eiλtdt ++∞R x sin λ (x − t) λ q (t) R1(t) e iλtdt ++∞R x sin λ (x − t) 2p (t) µ+∞ R t K (t, ξ) eiλξdξ ¶ dt ++∞R x sin λ (x − t) λ q (t) µ+∞ R t K (t, ξ) eiλξdξ ¶ dt = −i +∞R x p (t) R1(t) eiλxdt + i +∞R x p (t) R1(t) eiλ(−x+2t)dt +1 2 +∞R x q (t) R1(t) à x R −x+2t eiλudu ! dt − i +∞R x p (t) µ+∞ R t K (t, ξ) eiλ(x−t+ξ)dξ ¶ dt +i+∞R x p (t) µ+∞ R t K (t, ξ) eiλ(−x+t+ξ)dξ ¶ dt +1 2 +∞R x q (t) à +∞R t K (t, ξ) x−t+ξR −x+t+ξ eiλududξ ! dt = −ieiλx+∞R x p (t) R1(t) dt +i 2 +∞R x p µ x + t 2 ¶ R1 µ x + t 2 ¶ eiλtdt −1 2 +∞R x à +∞R (x+t)/2 q (u) R1(u) du ! eiλtdt −i+∞R x µ+∞ R x p (u) K (u, t − x + u) du ¶ eiλtdt +i+∞R x à (x+t)/2R x p (u) K (u, t + x − u) du ! eiλtdt +1 2 +∞R x à +∞R x q (u)t−x+uR t+x−u K (u, ξ) dξdu ! eiλtdt.
Thus, in the case of x > a, one can obtain the following equations
R1(x) = 1 − i +∞R x p (t) R1(t) dt K (x, t) = i 2p µ x + t 2 ¶ R1 µ x + t 2 ¶ −1 2 +∞R (x+t)/2 q (u) R1(u) du −i +∞R x p (u) K (u, t − x + u) du + i (x+t)/2R x p (u) K (u, t + x − u) du +1 2 +∞R x q (u)t−x+uR t+x−u K (u, ξ) dξdu. (2.9)
Similiarly, for 0 < x < a, we obtain R2(x) − 1 = −i α a R x p (t) R2(t) dt − i +∞R a p (t) R1(t) dt R3(x) − 1 = i α a R x p (t) R3(t) dt − i +∞R a p (t) R1(t) dt,
where K (x, t) ≡ 0 for |ξ| < αt − αa + a. Further one can get the following integral equations for the kernel K (x, t) ,
If 0 < x < a, αx − αa + a < t < −αx + αa − a ; K (x, t) = iα + 2α2p µ t + αx + αa − a 2α ¶ R2 µ t + αx + αa − a 2α ¶ −α+ 2α a R (t+αx+αa−a)/2α q (u) R2(u) du +iα − 2α2p µ −t + αx + αa + a 2α ¶ R3 µ −t + αx + αa + a 2α ¶ −α− 2α a R (−t+αx+αa+a)/2α q (u) R3(u) du − α+ 2 +∞R a q (u) R1(u) du +iα − 2 p µ t − αx + αa + a 2 ¶ R1 µ t − αx + αa + a 2 ¶ +α− 2 (t−αx+αa+a)/2R a q (u) R1(u) du − i α a R x
p (u) K (u, t − αx + αu) du
+i
α
(t+αx+αa−a)/2αR
x
p (u) K (u, t + αx − αu) du − 1
2α a R x q (u)t−αx+αuR t+αx−αu K (u, ξ) dξdu −iα++∞R a
p (u) K (u, t − αx + αa − a + u) du −α
+ 2 +∞R a q (u)t−αx+αa−a+uR t+αx−αa+a−u K (u, ξ) dξdu +iα−(t−αx+αa+a)/2R a
p (u) K (u, t − αx + αa + a − u) du
+α − 2 (t−αx+αa+a)/2R a q (u) t−αx+αa+a−uR t+αx−αa−a+u K (u, ξ) dξdu (2.10) if 0 < x < a, -αx + αa − a < t < −αx + αa + a; K (x, t) = iα + 2α2p µ t + αx + αa − a 2α ¶ R2 µ t + αx + αa − a 2α ¶ −α + 2α a R (t+αx+αa−a)/2α q (u) R2(u) du
+iα − 2α2p µ −t + αx + αa + a 2α ¶ R3 µ −t + αx + αa + a 2α ¶ −α − 2α a R (−t+αx+αa+a)/2α q (u) R3(u) du −α + 2 +∞R a q (u) R1(u) du +iα − 2 p µ t − αx + αa + a 2 ¶ R1 µ t − αx + αa + a 2 ¶ +α − 2 (t−αx+αa+a)/2R a q (u) R1(u) du − i α a R x
p (u) K (u, t − αx + αu) du
+i
α
(t+αx+αa−a)/2αR
x
p (u) K (u, t + αx − αu) du − 1 2α a R x q (t)t−αx+αuR t+αx−αu K (u, ξ) dξdu −iα++∞R a
p (u) K (u, t − αx + αa − a + u) du −α + 2 +∞R a q (u)t−αx+αa−a+uR t+αx−αa+a−u K (u, ξ) dξdu +iα−(t−αx+αa+a)/2R a
p (u) K (u, t − αx + αa + a − u) du
+α − 2 (t−αx+αa+a)/2R a q (u)t−αx+αa+a−uR t+αx−αa−a+u K (u, ξ) dξdu (2.11) if 0 < x < a, −αx + αa + a < t < +∞; K (x, t) = iα + 2 p µ t + αx − αa + a 2 ¶ R1 µ t + αx − αa + a 2 ¶ −α + 2 +∞R (t+αx−αa+a)/2 q (u) R1(u) du +iα− 2 p µ t − αx + αa + a 2 ¶ R1 µ t − αx + αa + a 2 ¶ −i α a R x
p (u) K (u, t − αx + αu) du
+i
α
a
R
x
p (u) K (u, t + αx − αu) du − 1 2α a R x q (u)t−αx+αuR t+αx−αu K (u, ξ) dξdu −iα++∞R a
+iα+(t+αx−αa+a)/2R
a
p (u) K (u, t + αx − αa + a − u) du −α + 2 +∞R a q (u)t−αx+αa−a+uR t+αx−αa+a−u K (u, ξ) dξdu −iα−+∞R a
p (u) K (u, t + αx − αa − a + u) du
+iα−(t−αx+αa+a)/2R a
p (u) K (u, t − αx + αa + a − u) du α− 2 (t−αx+αa+a)/2R a q (u)t−αx+αa+a−uR t+αx−αa−a+u K (u, ξ) dξdu. (2.12)
Let’s shown by the method of successive approximations that, equations of (2.10) , (2.11) and (2.12) have solution K (x, .) ∈ L1(µ+(x) , ∞) satisfying property of (2.2).
In the case of x > a K0(x, t) = i 2p µ x + t 2 ¶ R1 µ x + t 2 ¶ −1 2 +∞R (x+t)/2 q (u) R1(u) du K1(x, t) = −i +∞R x p (u) K0(u, t − x + u) du +i (x+t)/2R x p (u) K0(u, t + x − u) du + 1 2 +∞R x q (u) t−x+uR t+x−u K0(u, ξ) dξdu Kn(x, t) = −i +∞R x p (u) Kn−1(u, t − x + u) du +i (x+t)/2R x p (u) Kn−1(u, t + x − u) du +1 2 +∞R x q (u)t−x+uR t+x−u Kn−1(u, ξ) dξdu K (x, t) = ∞ X n=0 Kn(x, t) (2.13)
Let us show by induction that +∞Z αx−αa+a |Kn(x, t)| dt ≤σ n+1(x) (n + 1)!, where, +∞Z x [2 |p (t)| + (1 + |t|) |q (t)|] dt = σ (x) . (2.14) Indeed,
+∞R x |K0(x, t)| dt ≤ 1 2 +∞R x ¯ ¯ ¯ ¯p µ x + t 2 ¶¯¯ ¯ ¯ ¯ ¯ ¯ ¯R1 µ x + t 2 ¶¯¯ ¯ ¯ dt +1 2 +∞R x +∞R (x+t)/2
|q (u)| |R1(u)| dudt. We obtain +∞R x |K0(x, t)| dt ≤ +∞R x [2 |p (t)| + (1 + |t|) |q (t)|] dt. Thus +∞Z x |K0(x, t)| dt ≤ σ (x) . (2.15) For n = 1 +∞R x |K1(x, t)| dt ≤ +∞R x +∞R x
|p (u)| |K0(u, t − x + u)| dudt ++∞R
x
(x+t)/2R
x
|p (u)| |K0(u, t + x − u)| dudt +1 2 +∞R x +∞R x |q (u)|t−x+uR t+x−u |K0(u, ξ)| dξdudt. ≤+∞R x [2 |p (t)| + (1 + |t|) |q (t)|] σ (t) dt. We get +∞Z x |K1(x, t)| dt ≤ σ 2(x) 2! . (2.16) Suppose that +∞Z x |Kn−1(x, t)| dt ≤σ n(x) n! (2.17)
is valid for n − 1. Let’s show that +∞Z x |Kn(x, t)| dt ≤ σn+1(x) (n + 1)!. (2.18) +∞R x |Kn(x, t)| dt ≤ +∞R x +∞R x
|p (u)| |Kn−1(u, t − x + u)| dudt
+ +∞R
x
(x+t)/2R
x
|p (u)| |Kn−1(u, t + x − u)| dudt+1
2 +∞R x +∞R x |q (u)| t−x+uR t+x−u |Kn−1(u, ξ)| dξdudt ≤ +∞R x [2 |p (t)| + (1 + |t|) |q (t)|]σ n(t) n! dt.
Hence +∞Z x |Kn(x, t)| dt ≤ σ n+1(x) (n + 1)!
is valid. It follows by virtue of last equation in the case of x > a, series, (2.13) convergents in L1(x, +∞) for each fixed x, and then
+∞R x |K (x, t)| dt ≤ P∞ n=0 +∞R x |Kn(x, t)| dt ≤ 1 − 1 + σ (x) +σ2(x) 2! + ... + σn+1(x) (n + 1)! + ... = eσ(x)−1. (2.19)
Now, we consider the case of 0 < x < a.
Then, put K (x, t) ≡ 0 for t > −αx + αa − a in the case of αx − αa + a < t <
−αx + αa − a −αx+αa−aR αx−αa+a |K0(x, t)| dt ≤ α+ 2α2 −αx+αa−aR αx−αa+a ¯ ¯ ¯ ¯p µ t + αx + αa − a 2α ¶¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯R2 µ t + αx + αa − a 2α ¶¯ ¯ ¯ ¯ dt +α + 2α −αx+αa−aR αx−αa+a à a R (t+αx+αa−a)/2α |q (u)| |R2(u)| du ! dt +α− 2α2 −αx+αa−aR αx−αa+a ¯ ¯ ¯ ¯p µ −t + αx + αa + a 2α ¶¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯R3 µ −t + αx + αa + a 2α ¶¯ ¯ ¯ ¯ dt +α − 2α −αx+αa−aR αx−αa+a à a R (−t+αx+αa+a)/2α |q (u)| |R3(u)| du ! dt +α+ 2 −αx+αa−aR αx−αa+a µ+∞ R a |q (u)| |R1(u)| du ¶ dt +α − 2 −αx+αa−aR αx−αa+a ¯ ¯ ¯ ¯p µ t − αx + αa + a 2 ¶¯¯ ¯ ¯ ¯ ¯ ¯ ¯R1 µ t − αx + αa + a 2 ¶¯¯ ¯ ¯ dt +α− 2 −αx+αa−aR αx−αa+a à (t−αx+αa+a)/2R a |q (u)| |R1(u)| du ! dt ≤ α + α +∞R x |p (t)| dt + α++∞R x (1 + |t|) |q (t)| dt −α − α +∞R x |p (t)| dt +α−+∞R x (1 + |t|) |q (t)| dt + αα++∞R x (1 + |t|) |q (t)| dt +α−+∞R x |p (t)| dt + 2αα−+∞R x (1 + |t|) |q (t)| dt
< α++∞R x [2 |p (t)| + (1 + |t|) |q (t)|] dt +α−+∞R x [2 |p (t)| + (1 + |t|) |q (t)|] dt +α (α++ 2α−)+∞R x (1 + |t|) |q (t)| dt + α−+∞R x |p (t)| dt < α+σ (x) + α−σ (x) + 2α+∞R x (1 + |t|) |q (t)| dt + 4α+∞R x |p (t)| dt < (α++ α−+ 2α) σ (x) = c1σ (x) . (2.20) Hence −αx+αa−aZ αx−αa+a |K0(x, t)| dt ≤ c1σ (x) .
Put K (x, t) ≡ 0 for t < −αx+αa−a in the case of −αx+αa−a < t < −αx+αa+a to get −αx+αa+aR −αx+αa−a |K0(x, t)| dt ≤ α + 2α2 −αx+αa+aR −αx+αa−a ¯ ¯ ¯ ¯p µ t + αx + αa − a 2α ¶¯¯ ¯ ¯ ¯ ¯ ¯ ¯R2 µ t + αx + αa − a 2α ¶¯¯ ¯ ¯ dt +α+ 2α −αx+αa+aR −αx+αa−a à a R (t+αx+αa−a)/2α |q (u)| |R2(u)| du ! dt +α − 2α2 −αx+αa+aR −αx+αa−a ¯ ¯ ¯ ¯p µ −t + αx + αa + a 2α ¶¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯R3 µ −t + αx + αa + a 2α ¶¯ ¯ ¯ ¯ dt +α − 2α −αx+αa+aR −αx+αa−a à a R (−t+αx+αa+a)/2α |q (u)| |R3(u)| du ! dt +α + 2 −αx+αa+aR −αx+αa−a µ+∞ R a |q (u)| |R1(u)| du ¶ dt +α− 2 −αx+αa+aR −αx+αa−a ¯ ¯ ¯ ¯p µ t − αx + αa + a 2 ¶¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯R1 µ t − αx + αa + a 2 ¶¯ ¯ ¯ ¯ dt +α − 2 −αx+αa+aR −αx+αa−a à (t−αx+αa+a)/2R a |q (u)| |R1(u)| du ! dt < α + α +∞R x |p (t)| dt + α++∞R x (1 + |t|) |q (t)| dt +α − α +∞R x |p (t)| dt + α−+∞R x (1 + |t|) |q (t)| dt +α++∞R x (1 + |t|) |q (t)| dt + α−+∞R x |p (t)| dt
+αα−+∞R x (1 + |t|) |q (t)| dt < α++∞R x [2 |p (t)| + (1 + |t|) |q (t)|] dt +α−+∞R x [2 |p (t)| + (1 + |t|) |q (t)|] dt + (αα−+ α+)+∞R x (1 + |t|) |q (t)| dt + α−+∞R x |p (t)| dt < α+σ (x) + α−σ (x) + αα++∞R x [2 |p (t)| + (1 + |t|) |q (t)|] dt = (α++ α−+ αα+) σ (x) = c2σ (x) . (2.21) Thus −αx+αa+aZ −αx+αa−a |K0(x, t)| dt ≤ c2σ (x) is obtained. For −αx + αa + a < t < +∞, +∞R −αx+αa+a |K0(x, t)| dt ≤α + 2 +∞R −αx+αa+a ¯ ¯ ¯ ¯p µ t + αx − αa + a 2 ¶¯¯ ¯ ¯ ¯ ¯ ¯ ¯R1 µ t + αx − αa + a 2 ¶¯¯ ¯ ¯ dt +α+ 2 +∞R −αx+αa+a à +∞R (t+αx−αa+a)/2 |q (u)| |R1(u)| du ! dt +α− 2 +∞R −αx+αa+a ¯ ¯ ¯ ¯p µ t − αx + αa + a 2 ¶¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯R1 µ t − αx + αa + a 2 ¶¯ ¯ ¯ ¯ dt < α++∞R x |p (t)| dt + α++∞R x (1 + |t|) |q (t)| dt + α++∞R x |p (t)| dt = α++∞R x [2 |p (t)| + (1 + |t|) |q (t)|] dt = α+σ (x) . (2.22) So +∞Z −αx+αa+a |K0(x, t)| dt ≤ α+σ (x) . Then, it follows from (2.20) , (2.21) and (2.22) for 0 < x < a
+∞Z
αx−αa+a
|K0(x, t)| dt ≤ Cσ (x) , (2.23) where C = c1+ c2+ α+, c
Now, for n = 1, if one evaluate integral +∞R αx−αa+a |K1(x, t)| dt then +∞R αx−αa+a |K1(x, t)| dt ≤ 1 α +∞R αx−αa+a µa R x
|p (u)| |K0(u, t − αx + αu)| du ¶ dt +1 α +∞R αx−αa+a µa R x
|p (u)| |K0(u, t + αx − αu)| du ¶ dt + 1 2α +∞R αx−αa+a à a R x |q (u)|t−αx+αuR t+αx−αu |K0(u, ξ)| dξdu ! dt +α+ +∞R αx−αa+a µ+∞ R a
|p (u)| |K0(u, t − αx + αa − a + u)| du ¶ dt +α+ +∞R αx−αa+a à (t+αx−αa+a)/2R a
|p (u)| |K0(u, t + αx − αa + a − u)| du ! dt +α + 2 +∞R αx−αa+a à +∞R a |q (u)| t−αx+αa−a+uR t+αx−αa+a−u |K0(u, ξ)| dξdu ! dt +α− +∞R αx−αa+a µ+∞ R a
|p (u)| |K0(u, t + αx − αa − a + u)| du ¶ dt +α− +∞R αx−αa+a à (t−αx+αa+a)/2R a
|p (u)| |K0(u, t − αx + αa + a − u)| du ! dt +α − 2 +∞R αx−αa+a à (t−αx+αa+a)/2R a |q (u)|t−αx+αa+a−uR t+αx−αa−a+u |K0(u, ξ)| dξdu ! dt < CRa x [2 |p (t)| + (1 + |t|) |q (t)|] σ (t) dt +C+∞R a [2 |p (t)| + (1 + |t|) |q (t)|] σ (t) dt. (2.24) It follows from (2.24) +∞Z αx−αa+a |K1(x, t)| dt ≤ Cσ 2(x) 2! . (2.25) Assume that +∞Z αx−αa+a |Kn−1(x, t)| dt ≤ Cσ n(x) n! (2.26)
is valid for n − 1.Let’s show that +∞Z
αx−αa+a
|Kn(x, t)| dt ≤ Cσ n+1(x)
is valid for n. +∞R αx−αa+a |Kn(x, t)| dt ≤ 1 α +∞R αx−αa+a µa R x
|p (u)| |Kn−1(u, t − αx + αu)| du
¶ dt +1 α +∞R αx−αa+a µa R x
|p (u)| |Kn−1(u, t + αx − αu)| du
¶ dt + 1 2α +∞R αx−αa+a à a R x |q (u)|t−αx+αuR t+αx−αu |Kn−1(u, ξ)| dξdu ! dt +α+ +∞R αx−αa+a µ+∞ R a
|p (u)| |Kn−1(u, t − αx + αa − a + u)| du
¶ dt +α+ +∞R αx−αa+a à (t+αx−αa+a)/2R a
|p (u)| |Kn−1(u, t + αx − αa + a − u)| du
! dt +α+ 2 +∞R αx−αa+a à +∞R a |q (u)|t−αx+αa−a+uR t+αx−αa+a−u |Kn−1(u, ξ)| dξdu ! dt +α− +∞R αx−αa+a µ+∞ R a
|p (u)| |Kn−1(u, t + αx − αa − a + u)| du
¶ dt +α− +∞R αx−αa+a à (t−αx+αa+a)/2R a
|p (u)| |Kn−1(u, t − αx + αa + a − u)| du
! dt +α − 2 +∞R αx−αa+a à (t−αx+αa+a)/2R a |q (u)| t−αx+αa+a−uR t+αx−αa−a+u |Kn−1(u, ξ)| dξdu ! dt < C n! a R x [2 |p (t)| + (1 + |t|) |q (t)|] σn(t) dt +C n! +∞R a [2 |p (t)| + (1 + |t|) |q (t)|] σn(t) dt. Thus +∞Z αx−αa+a |Kn(x, t)| dt ≤ Cσ n+1(x) (n + 1)! (2.28)
Remark: For simplicity, α was taken as α > 1. Other case can be treated in the same way.
Finally, let investigate correlation between the kernel function K (x, t) and co-efficients p (x) , q (x) and ρ (x) in the equation
−y00+ [2λp (x) + q (x)] y = λ2ρ (x) y. For 0 < x < a, e (x, λ) = α+R 2(x) eiλµ +(x) + α−R 3(x) eiλµ −(x) + +∞R µ+(x) K (x, t) eiλtdt.
with µ+(x) < µ−(x) < +∞. If we put e (x, λ) and e00(x, λ) instead of y ve y00 in (1.1) , we get; α+R00 2(x) + 2iλαα+R20 (x) − 2λα+p (x) R2(x) + 2ip (x) K (x, µ+(x)) −α+q (x) R 2(x) − 2αdK (x, µ +(x)) dx = 0 α−R00 3(x) − 2iλαα−R03(x) − 2λα−p (x) R3(x) − 2ip (x) K (x, µ−(x) − 0) +2ip (x) K (x, µ−(x) + 0) − α−q (x) R 3(x) −2αdK (x, µ −(x) − 0) dx + 2α dK (x, µ−(x) + 0) dx = 0 ∂2K (x, t) ∂x2 − α2 ∂2K (x, t) ∂t2 + 2ip (x) ∂K (x, t) ∂t = q (x) K (x, t) lim x+t→∞ ∂K (x, t) ∂x = 0, x+t→∞lim ∂K (x, t) ∂t = 0
In the case x > a, since µ+(x) = x and α = 1, then the same properties of the kernel function.can be easily obtained.
Hence, the proof is completed.
Acknowledgments
This work is supported by the Scientific Research Project Fund of Cumhuriyet Uni-versity under the project number F-287.
References
[1] V.A. Ambartsumyan,1929, ¨Uber eine Frage der Eigenwerttheorie, Z. Physik 53, 690-695.
[2] Povzner, A. V., 1948, On Differential Equations of Sturm-Liouville Type on a Half-Axis, Mat. Sb., 23.
[3] Marchenko V. A. (1950). Some Problems In The Theory of Second-Order Differential Operator, Dock. Akad. Nauk SSSR, 72 , 457-460
[4] Gelfand I. M. and Levitan B. M. (1951). On The Determination of a Differential Equation From Its Spectral Function , Izv. Akad. Nauk SSSR, SEr. Mat. 15 309-360; English transl. In Amer. Math. Soc. Transl. (2) 1 (1955)
[5] Marchenko V. A. (1952). Some Problems In The Theory of Linear Differential Oper-ators, Trudy Moskov. Mat. Obshch. 1 , 327-420
[6] Levitan, B. M., 1964, Generalized Translation Operators and some o its Applications, Jerusalem.
[7] Gasimov, M.G. and Levitan, B.M., 1964, About Sturm-Liouville Differential Opera-tors, Math. Sborn., 63 (105), No. 3.
[8] Naimark, M. A., 1967, Linear Differential Operators, Moscow, Nauka, (in Russian). [9] Gasimov M.G., Guseniov,G.Sh,1981, Determining of The Diffusion Operator From
[10] Amirov, R. Kh. and Yurko, V.A., 2001, On Differential Operators with a Singularity and Discontinuity Conditions Inside the Interval. Ukr. Math. Jour., V.53, No:11, pp. 1443-1458.
[11] Freiling G., Yurko V., (2001) Inverse Sturm Liouville Problems and Their Applica-tions, Nova Science Publishers,
[12] Akhmedova E.N.,2002,On Represantation of Solution of Sturm-Liouville Equation With Discontinious Coefficients, Proceedings of IMM of NAS of Azerbaijan v XVI, (XXIV) pp 5-9.
[13] Amirov, R. Kh., 2006, On Sturm-Liouville Operators with Discotinuity conditions inside an interval, J. Math. Anal. Appl., 317 , pp. 163-176.
[14] Huseynov Hidayet M., Osmanova Jalala A. ,2006, On Iost Solution of Sturm-Liouville Equation With Discontinuty Conditions,IMM of NAS of Azerbaijan, 9, Agayev str., AZ1141, Baku, Azerbaijan
[15] Karacan S. (2009). An Integral Representation of Solution of Diffusion Equation With Discontinuous Coefficient, Msc Thesis, Cumhuriyet Univercity, Sivas.