Abstract
It is known that the uniqueness of potential function of the Sturm–Liouville problem can be shown from the nodal points. In this article, we solve the inverse nodal problem of the reconstruction of the potential function q from the nodal data by a pointwise limit. We show that this convergence is in the L1. It is mentioned that this method is based on the works of Law and Yang, but we have applied the method for the Sturm–Liouville problem depending on eigenparameter boundary conditions.
1. Introduction
Inverse problems of spectral theory have been an important part of the research in mathematical physics. The development of the inverse problem, its close relative, is quite different. In general, a set of eigenvalues is not sufficient to determine the operator. It needs more data that are spectral function, norming constants, etc., to determine the operator. For example, two sets of eigenvalues Citation1, one set of eigenvalues plus a symmetric potential function Citation2 and a spectral function Citation3. On the other hand, some questions relating to the Sturm–Liouville problem were solved by Gesztesy and Simon Citation4,Citation5 and Malamud Citation6.
Recently, a new class of inverse problems has attracted the attention of researchers. This is the so-called inverse nodal problem. The inverse nodal problem was initiated by McLaughlin Citation7, which is different from the classical inverse spectral theory of Gelfand and Levitan Citation3. She found that the nodal set of the Dirichlet problem for the Sturm–Liouville equation can determine q up to a constant. Later, Hald and McLaughlin Citation8 and Browne and Sleeman Citation9 showed that only the knowledge of nodal points can determine the potential function of the regular Sturm–Liouville problem. Yang Citation10 showed that this uniqueness result is valid for any q. In the past few years, the inverse nodal problem of the regular and singular Sturm–Liouville problem has been investigated by several authors Citation7–9,Citation11,Citation12.
Consider the following equation:
with the boundary conditions
where q(x) is real and integrable on [0, π], and μ and a ≠ 0 are real parameters. It is well known that the above problem has only real and simple eigenvalues Citation13.
In Hochstadt's paper Citation13, a solution of (1.1) shall be sought in the form
By this induction, we see that
so that
By using (1.3) we get Citation13
where tan λoπ = −a and
We observe that μ dependence enters into the terms, vanishing like
.
Let λo(q, a) < λ1(q, a) < ··· →∞ be the eigenvalues of the (1.1)–(1.3) and i = 1, 2, …, n − 1, be nodal points of the nth eigenfunction. Let λn be the nth eigenvalue and
be the ith nodal point of the nth eigenfunction yn. Also, let
be the ith nodal domain of the nth eigenfunction and
be the associated nodal length. It is shown that the set of all nodal points
is dense in [0, π]; in fact, a judicious choice of one nodal point
for each yn, n > 1 also forms a dense set in [0, π]. The simplest method for choosing a dense subset of the nodes is to choose all the eigenfunctions. In addition to this, it will be bound that the nodes for yn are roughly equally spaced. Hence, we can choose the first node in
and the second in
The third, fourth, fifth and sixth nodes should lie in
and
and so on. This method gives a dense set of nodes and it is not even necessary to choose the next node from the next eigenfunction and any finite amount of nodes can be deleted.
We define the function jn(x) on (0, π) by Hence fix x and n, j = jn(x) implies
The inverse nodal problem was first studied by McLaughlin Citation7 and Hald and McLaughlin Citation8, then this method was extended in Citation14,Citation15. In this article, these results are expanded. In other words, for the Sturm–Liouville problem which includes μ parameter and a general boundary condition, a reconstruction formula for q is given. Our main theorems are as follows.
THEOREM 1.1
Suppose that q ∈ C1 on [0, π], then
where μ is as expressed in (1.1).
We note that with the asymptotic expression for λn in Theorem 1.1 implies that , where Fn is determined only by the nodal data and the constant
in the form
and μ is as expressed in (1.1).
THEOREM 1.2
Fn converges to q in L1.
2. Main results
In this section, some lemmas which are necessary for proof of the Theorem 1.1 and Theorem 1.2 are given.
LEMMA 2.1
Assume that q ∈ L1 (0, π). Then as n → ∞
Proof
From (1.4), we know that
If y(x, λ) = 0, then as long as cos λx is not close to zero, then
Now, we take λ = λn and
. Since the Taylor's expansion for the arctangent function is given by
then
Therefore
The nodal length is
LEMMA 2.2
Suppose that f ∈ L1(0, π). Then for almost every x ∈ (0, π), with j = jn(x)
Proof
For f ∈ L1(0, π), almost everywhere. Then, given ξ > 0, when n is sufficiently large and for almost every x ∈ (0, π)
This proves the lemma.▪
Proof of Theorem 1.1
When we consider (2.2) in the form
so that
By Lemma 2.2
then
or
It remains to be shown that for almost every x ∈ (0, π),
tends to zero as n → ∞.
We get a sequence of continuous functions qk which converges to q in L1(0, π). Then qk has a subsequence converging to q almost everywhere in (0, π). For simplicity, we call this subsequence qk. We take any x such that qk(x) converges to q(x). Then for a given ϵ > 0, we can fix a large k such that |qk(x) − q(x)| < ϵ. Hence,
Now, by Lemma 2.1
and this term tends to zero for n → ∞. By (1.6) the first term An satisfies, when n is sufficiently large,
On the other hand,
Because qk is continuous, this term is arbitrarily small for every x ∈ (0, π). For n → ∞,
Then, Dn → 0. Hence we conclude that
This proves the theorem.
LEMMA 2.3
We take that the sequence fk ∈ [0, π] converges to f ∈ L1, then, for any large enough n, with j = jn(x), as k → ∞
Proof
By Lemma 2.1 and the observation that the integral is constant on any nodal interval
and for k → ∞ this term converges to zero.▪
LEMMA 2.4
Suppose that q ∈ L1 (0, π), then, as n → ∞ with j = jn(x)
Proof
First, we show that if q is continuous on [0, π] the result satisfies. Let By using the intermediate value theorem, there exists ζ ∈ (a, x) such that
If x is close enough to a, the difference can be arbitrarily small. Then, for all ϵ > 0, when n is large enough, with j = jn(x)
In the above process, we assume that
The estimate also holds if
Hence if q ∈ C[0, π], then
converges to q(x) uniformly on [0, π]. Thus
can be arbitrarily small. Since C[0, π] is dense in L1(0, π), for any q ∈ L1(0, π) there exists a sequence qk ∈ C [0, π] convergent to q in L1(0, π). Hence fix n sufficiently large,
By the above process and Lemma 2.3, when k is large enough, the first two terms are arbitrarily small. Hence, as k → ∞
Proof of Theorem 1.2
When we consider the value of Fn, we obtained that
It sufficies to show that as n → ∞
By using Lemma 2.1 we have
Hence, we only need to prove that for n → ∞
and
From Lemma 2.3, the first limit holds and the second limit also holds. On the other hand, the sequence of functions
converges to 0 for almost every x ∈ (0, π). Furthermore,
and
Then, we may apply the Lebesque dominated convergence theorem to show that (2.3) is valid. The proof is completed.
Acknowledgements
This work was supported in part by the Firat University, Scientific Research Project Unit under grant number: 1540.
References
- Borg, G, 1945. Eine umkehrung der Sturm-Liouvillesehen eigenwertaufgabe, Acta Math. 78 (1945), pp. 1–96.
- Hochstadt, H, 1973. The inverse Sturm–Liouville problem, Commun. Pure Appl. Math. 26 (1973), pp. 715–729.
- Gelfand, IM, and Levitan, BM, 1951. On the determination of a differential equation from its spectral function, Amer. Math Soc. Trans. 1 (1951), pp. 253–304.
- Gesztesy, F, and Simon, B, 2000. Inverse spectral analysis with partial information on the potential. II: The case of discrete spectrum, Trans. Am. Math. Soc. 352 (2000), pp. 2765–2787.
- Gesztesy, F, and Simon, B, 2000. On local Borg–Marchenko uniqueness results, Commun. Math. Phys. 211 (2) (2000), pp. 273–287.
- Malamud, MM, 1999. Uniqueness questions in inverse problems for systems of differential equations on a finite interval, Trans. Moscow Math Soc. 60 (1999), pp. 204–262.
- McLaughlin, JR, 1988. Inverse spectral theory using nodal points as data–a uniqueness result, J. Diff. Eq. 73 (1988), pp. 354–362.
- Hald, O, and McLaughlin, JR, 1989. Solutions of inverse nodal problems, Inv. Prob. 5 (1989), pp. 307–347.
- Browne, PJ, and Sleeman, BD, 1996. Inverse nodal problem for Sturm–Liouville equation with eigenparameter depend boundary conditions, Inverse Problems 12 (1996), pp. 377–381.
- Yang, X-F, 1997. A solution of the inverse nodal problem, Inv. Probl. 13 (1997), pp. 203–213.
- Cheng, YH, Law, C-K, and Tsay, J, 2000. Remarks on a new inverse nodal problem, J. Math. Anal. Appl. 248 (2000), pp. 145–155.
- Koyunbakan, H, 2005. Inverse nodal problem for singular differential operator, J. Inverse Ill-posed Probl. 13 (5) (2005), pp. 435–440.
- Hochstadt, H, 1967. On inverse problems associated with second-order differential operators, Acta Math. 119 (1967), pp. 173–192.
- Shen, CL, and Tsai, TM, 1995. On a uniform approximation of the density function of a string equation using eigenvalues and nodal points and some related inverse nodal problems, Inv. Prob. 11 (1995), pp. 1113–1123.
- Chen, Y-T, Cheng, YH, Law, CK, and Tsa, J, 2002. Convergence of the reconstruction formula for the potential function, Proc. Amer. Math. Soc. 130 (2002), pp. 2319–2324.