ABSTRACT
This paper is concerned with the inverse problem for non-selfadjoint Sturm–Liouville operator with discontinuity conditions inside a finite interval. Firstly, we give the definitions of generalized weight numbers for this operator which may have the multiple spectrum and then investigate the connections between the generalized weight numbers and other spectral characteristics. Secondly, we obtain the generalized spectral data, which consists of the generalized weight numbers and the spectrum. Then the operator is determined uniquely by the method of spectral mappings. Finally, we give an algorithm for reconstructing the potential function and the coefficients of the boundary conditions and the coefficients of the discontinuity conditions.
1. Introduction
In this paper, we consider the following non-selfadjoint boundary value problem for the equation:(1) (1)
on the interval with the boundary conditions(2) (2)
and the discontinuity conditions(3) (3)
at , where is a complex-valued function, h, H, are complex numbers and , .
There has been extensive study of inverse problems for Sturm–Liouville operator with discontinuity conditions inside a finite interval since the discontinuities are connected with non-smooth material properties. The inverse problem for selfadjoint Sturm–Liouville operator with different type discontinuity has been considered and solved by different methods in [Citation1–Citation10]. [Citation11,Citation12] studied the inverse spectral problem for discontinuous Sturm–Liouville operators with boundary conditions linearly dependent on the spectral parameter. The inverse problem for non-selfadjoint Sturm–Liouville operator with discontinuity inside an interval has been investigated in [Citation13,Citation14] when the spectrum is simple.
Recently, many authors paid more attention on the inverse problem for the non-selfadjoint operator with multiple spectrum (see [Citation15–Citation20] and the references therein). Especially, Buterin [Citation18] considered the inverse problem for the boundary value problem (Equation1(1) (1) ), (Equation2(2) (2) ) with an arbitrary behaviour of the spectrum and gave generalized weight numbers more naturally and proved that a multiple spectrum and the generalized weight numbers determine the potential function and boundary conditions uniquely. In this paper, we add the discontinuity conditions () at to the boundary value problem (Equation1(1) (1) ), (Equation2(2) (2) ) and give the generalized weight numbers for discontinuous non-selfadjoint Sturm–Liouville operator with multiple spectrum, and recovering this operator from its spectral characteristics by spectral mappings (see [Citation21]).
This paper is organized as follows. In Section 2, some basic definitions and useful properties are given. We devote Section 3 to give the useful definition of generalized spectral data. The connections between the generalized spectral data and other spectral characteristics are investigated in Section 4. In Section 5, by the method of spectral mappings, we prove that the given generalized spectral data uniquely determine the potential q and the coefficients h, H, , , respectively, and then give an algorithm for reconstructing the operator .
2. Preliminaries
Let , be continuously differentiable functions on [ 0, d] and . Denote . If and satisfy the discontinuity conditions (Equation3(3) (3) ), then(4) (4)
Let , be solutions of Equation (Equation1(1) (1) ) satisfying the discontinuity conditions (Equation3(3) (3) ) and the initial conditions(5) (5)
respectively. Then . Denote , then is independent of x. From (Equation5(5) (5) ), we obtain(6) (6)
In the following, we give three powerful and important lemmas, the rigourous proof of these lemmas which can be referred to [Citation2,Citation22–Citation26] and no proof will be given here.
Lemma 2.1:
The zeros of coincide with the eigenvalues , of L. and are corresponding eigenfunctions of L.
Proof See [Citation18, p.740] and [Citation14, p.3].
Lemma 2.2:
Let , Im. For ,(7) (7) (8) (8) (9) (9) (10) (10)
where , .
In particular, for , we obtain(11) (11) (12) (12)
Proof The proof is similar to the selfadjoint case, see [Citation2, p.145–146].
Lemma 2.3:
The roots , of
are separated. For fixed and sufficiently large ,(13) (13)
where . By Rouché theorem, we have
so
where , , and is a bounded sequence
Proof The proof can be refer to [Citation2, p.146] and [Citation26, Lemma 3].
3. The generalized spectral data
The algebraic multiplicity of the eigenvalue () is the order of it as a root of , i.e. . In throughout the paper, we use multiplicity instead of algebraic multiplicity for short. By the virtue of Lemma 2.3, for sufficient large n, .
Let , , . For , , we have(14) (14) (15) (15)
From (Equation6(6) (6) ), we infer that
i.e. and , , , are generalized eigenfunctions of L. Let(16) (16)
It is easy to see that , are complete systems of eigenfunctions and generalized eigenfunctions of L (refer to [Citation24, Theorem 1.3.2]). Naturally, we can define the generalized weight numbers , for L by the following equations:(17) (17)
When the multiplicity , the generalized weight numbers coincide with the weight numbers for the selfadjoint Sturm–Liouville operator with discontinuity conditions inside a finite interval (see [Citation2, p.143 (10)]).
Definition 1:
The numbers are called the generalized spectral data of L.
4. The Weyl function
Denote by , the solutions of equation (Equation1(1) (1) ) under the conditions
and the discontinuity conditions (Equation3(3) (3) ). The functions and are called the Weyl solution and the Weyl function for L, respectively. Evidently,(18) (18) (19) (19) (20) (20)
The symbol denotes the characteristic function of the boundary value problem consisting of the equation (Equation1(1) (1) ), the discontinuity conditions (Equation3(3) (3) ) and the boundary conditions . The zeros of are expressed in terms of ; it is easy to show that . Then is a meromorphic function with zeros in and poles in .
Next, we prove that the generalized spectral data determine the Weyl function uniquely by the following theorem. This is a generalization of corresponding result of non-selfadjoint Sturm–Liouville operator without discontinuities (see [Citation18, p.741 (9)]).
Theorem 4.1:
The Weyl function and the generalized spectral data of L satisfy the following equalities:(21) (21) (22) (22)
where is Kronecker delta.
Proof Firstly, considering the contour integral
where , M, , is assumed to be counterclockwise. By the virtue of Lemma 2.3, it yields for sufficiently small fixed and sufficiently large N. The formulae (Equation12(12) (12) ), (Equation13(13) (13) ) and (Equation20(20) (20) ) yield
for sufficiently large . Hence . Using the residue theorem (see [Citation25, V. §2.]), we calculate
Thus(23) (23)
Set , and in light of (Equation23(23) (23) ), we get (Equation21(21) (21) ).
Secondly, let us prove that coefficients and the generalized weight numbers determine each other uniquely by the formula (Equation22(22) (22) ). On account of (Equation20(20) (20) ) we have , together with the identity (Equation21(21) (21) ), we find(24) (24)
Since , , are the zeros of with the multiplicity , the Taylor series of at , , is . If we plug it back to (Equation24(24) (24) ) and let approaches , then . The proof of(25) (25)
follows in a similar manner. From (Equation18(18) (18) ), we get , . Owing to (Equation14(14) (14) )–(Equation16(16) (16) ), an easy induction gives(26) (26)
Moreover, since , are solutions of equation (Equation1(1) (1) ) and satisfy the discontinuity conditions (Equation3(3) (3) ), from (Equation4(4) (4) ), we know the function is continuous on , hence
By the initial conditions (Equation5(5) (5) ) and equality (Equation6(6) (6) ), we obtain
Hence . A simple manipulation leads to the solution that
Using (Equation14(14) (14) ), (Equation15(15) (15) ) and integrating by parts, we get(27) (27)
By substituting (Equation26(26) (26) ) in (Equation27(27) (27) ) and taking the definition of generalized weight numbers (Equation17(17) (17) ) into account, we obtain(28) (28)
Combining (Equation28(28) (28) ) and (Equation25(25) (25) ), we conclude that
Since , , continuing by induction we finally obtain the relation (Equation22(22) (22) ).
5. The inverse problem
Inverse Problem 5.1:
Recovering the operator L from one of the following conditions: (i) the generalized spectral data ; (ii) the two spectra , ; (iii) the Weyl function .
Remark 1:
According to Lemma 2.3, we know the spectrum uniquely determines the characteristic function . Similarly, the characteristic function is uniquely determined by its zeros . Combining (Equation20(20) (20) ), (Equation21(21) (21) ) and (Equation22(22) (22) ), we see that the statements (i)–(iii) of Inverse Problems 5.1 are equivalent. The numbers can also be used as spectral data.
5.1. The uniqueness theorem
Before giving the uniqueness theorem, we introduce some symbols initially. We agree that L, denote the operators of the same form but with different coefficients , , , , , . That is to say if a certain symbol represents an object related to L, then will denote the analogous object related to , and .
Theorem 5.2:
[The uniqueness theorem]If , , , then , i.e. a.e. on , , , , and .
Proof Because of Theorem 4.1, we know the generalized spectral data uniquely determines the Weyl function . It suffices to prove that if , then . It follows from (Equation12(12) (12) ), (Equation13(13) (13) ) and (Equation18(18) (18) ) that(29) (29)
Define the matrix by the following formula
i.e.(30) (30)
Formula (Equation19(19) (19) ) yields(31) (31)
Combining (Equation18(18) (18) ) and (Equation31(31) (31) ), we see that
Owing to (Equation18(18) (18) ) and (Equation31(31) (31) ), for each fixed x, the functions are meromorphic functions in . Put . According to (Equation11(11) (11) ), (Equation29(29) (29) ) and (Equation31(31) (31) ), we obtain(32) (32)
By (Equation18(18) (18) ) and (Equation31(31) (31) ), we see that if , then for each fixed x, the functions are entire in . Combining with (Equation32(32) (32) ), we derive , . Taking (Equation30(30) (30) ) into consideration, we get(33) (33)
for all x and . Together with (Equation7(7) (7) ), we see that for , , ,
where for , and for . Combining (Equation19(19) (19) ) and (Equation33(33) (33) ) this yields , , i.e. for all x and and consequently .
5.2. Solution of the inverse problem
Without loss of generality, we consider the inverse problem of recovering L from the generalized spectral data . Like [Citation2, p.153 (60)], choose an arbitrary model boundary value problem such that(34) (34)
where . Set , , , , , , , , , ,
For , , denote
where , . Similarly, by replacing with in the above definitions, we define , , , , , . Using the fact that is continuous on , , , , , , , , , , are continuous functions of .
By the same methods as in [Citation2, p.153–156], using Lemmas 2.2, 2.3, (Equation4(4) (4) ), (Equation17(17) (17) ), (Equation22(22) (22) ) and Schwarz’s lemma [Citation25, VI. §2.], we get the following estimates as , :(35) (35)
The similar estimates are also valid for , .
Lemma 5.3:
The following representations hold:(36) (36) (37) (37)
where the series converge absolutely and uniformly with respect to .
Proof From (Equation34(34) (34) ), we obtain and , then by virtue of (Equation7(7) (7) ), it yields(38) (38)
In the same way, we derive that(39) (39)
Let , using (Equation9(9) (9) )–(Equation10(10) (10) ), (Equation13(13) (13) ), (Equation18(18) (18) ) and (Equation39(39) (39) ), we arrive at(40) (40)
Further, combining (Equation19(19) (19) ) and (Equation31(31) (31) ), we see that(41) (41)
It follows from (Equation11(11) (11) ), (Equation12(12) (12) ), (Equation29(29) (29) ), (Equation31(31) (31) ), (Equation38(38) (38) ), (Equation40(40) (40) ) and (Equation41(41) (41) ) that(42) (42)
Analogously, we have(43) (43)
Let real numbers a, b be , , , . Consider closed contour (with counterclockwise circuit) in the -plane, where . By the standard method (see [Citation27, p.46–70]), using (Equation18(18) (18) ), (Equation30(30) (30) )–(Equation32(32) (32) ), and Cauchy’s integral formula (see [Citation25, IV. §5.]), we obtain the identity(44) (44)
where
Using (Equation42(42) (42) ), we acquire
uniformly respect to and on bounded sets. Similarly, we have the relation(45) (45)
where
uniformly with respect to and , on bounded sets. Calculating the integral in (Equation44(44) (44) ) by the residue theorem (see [Citation25, V. §2.]), we have, in light of (Equation21(21) (21) ),
for sufficiently large N. Passing to the limit in (Equation44(44) (44) ) as , we obtain
Taking derivative to the both sides of this equation with respect to the corresponding number of times and substituting into , we arrive at (Equation36(36) (36) ). Analogously, using the same method on (Equation45(45) (45) ), it yields
and taking the definitions of , into account we get (Equation37(37) (37) ).
Note that there exists , such that for , . Moreover, an argument similar to the one used in [Citation21, Lemma 1.3.4] shows that the infinite series
and
converge absolutely and uniformly on [ 0, d] and , respectively. Therefore, is square integrable on , where
Lemma 5.4:
The following relations hold
Proof The rigourous proof of this lemma is similar to [Citation21, Lemma 1.3.5], [Citation2, Lemma 5].
Remark 2:
For each fixed , the relation (Equation36(36) (36) ) can be considered as a system of linear equations with respect to , , . But the series in (Equation36(36) (36) ) converge only ‘with brackets’, i.e. the terms in them cannot be dissociated. Therefore, it is inconvenient to use (Equation36(36) (36) ) as a main equation of the inverse problem. Below we will transfer (Equation36(36) (36) ) to a linear equation in the Banach space of bounded sequences.
Denote . For each fixed , we define the vector
by the formula(46) (46)
We also define a block-matrix
by the following formula
Similarly we introduce , and , by replacing by , and by . Using (Equation35(35) (35) ), we get the estimates(47) (47)
Consider the Banach space B of bounded sequences with the norm . It follows from (Equation47(47) (47) ) that for each fixed the operators and (here I is the identity operator), acting from B to B, are bounded, and
Theorem 5.5:
For each fixed , the main equation(48) (48)
for the vector is uniquely solvable in the Banach space B. Moreover, the operator is bounded in B.
Proof Rewriting (Equation36(36) (36) ) in the form
substituting here (Equation46(46) (46) ) and taking into account our notations of and , we arrive at(49) (49)
which is equivalent to (Equation48(48) (48) ) and the series in (Equation49(49) (49) ) converges absolutely and uniformly for . Similarly, by the definitions of , , (Equation37(37) (37) ) becomes
which is equivalent to
Replacing L for , one gets analogously
Hence the operator exists, and it is bounded in B.
Equation (Equation48(48) (48) ) is called the main equation of the inverse problem. Using the solution of the main equation one can construct the function q, the coefficients , of the discontinuity conditions, and the coefficients h, H of the boundary conditions. Thus, we obtain the following algorithm for solving the inverse problem.
Additional information
Funding
Notes
No potential conflict of interest was reported by the authors.
References
- Hald OH . Discontinuous inverse eigenvalue problems. Commun Pure Appl Math. 1984;37:539–577.
- Yurko VA . Integral transforms connected with discontinuous boundary value problems. Integral Transform Spec Funct. 2000;10:141–164.
- Shieh CT , Yurko VA . Inverse nodal and inverse spectral problems for discontinuous boundary value problems. J Math Anal Appl. 2008;347:266–272.
- Shahriari M , Akbarfam AJ , Teschl G . Uniqueness for inverse Sturm-Liouville problems with a finite number of transmission conditions. J Math Anal Appl. 2012;395:19–29.
- Fu SZ , Xu ZB , Wei GS . The interlacing of spectra between continuous and discontinuous Sturm-Liouville problems and its application to inverse problems. Taiwanese J Math. 2012;16:651–663.
- Yang CF . Inverse problems for the Sturm-Liouville operator with discontinuity. Inverse Probl Sci Eng. 2014;22:232–244.
- Ozkan AS . Half-inverse Sturm-Liouville problem with boundary and discontinuity conditions dependent on the spectral parameter. Inverse Probl Sci Eng. 2014;22:848–859.
- Wang YP , Yurko VA . On the inverse nodal problems for discontinuous Sturm-Liouville operators. J Differ Equ. 2016;260:4086–4109.
- Xu XC , Yang CF . Inverse spectral problems for the Sturm-Liouville operator with discontinuity. J Differ Equ. 2017;262:3093–3106.
- Güldü Y , Arslantaş M . Direct and inverse problems for Sturm-Liouville operator which has discontinuity conditions and Coulomb potential. Chin J Math. 2014;2014:1–10.
- Ozkan AS , Keskin B . Spectral problems for Sturm-Liouville operator with boundary and jump conditions linearly dependent on the eigenparameter. Inverse Probl Sci Eng. 2012;20:799–808.
- Wang YP . Inverse problems for discontinuous Sturm-Liouville operators with mixed spectral data. Inverse Probl Sci Eng. 2015;23:1180–1198.
- Neamaty A , Khalili Y . The uniqueness theorem for differential pencils with the jump condition in the finite interval. Iran J Sci Technol Trans A Sci. 2014;38:305–309.
- Manafov MD . Inverse spectral problems for energy-dependent Sturm-Liouville equations with finitely many point δ-interactions. Electron J Differ Equ. 2016;2016:1–12.
- Brown BM , Peacock RA , Weikard R . A local borg-marchenko theorem for complex potentials. J Comput Appl Math. 2002;148:115–131.
- Tkachenko V . Non-selfadjoint Sturm-Liouville operators with multiple spectra. Oper Theory Adv Appl. 2002;134:403–414.
- Marletta M , Weikard R . Weak stability for an inverse Sturm-Liouville problem with finite spectral data and complex potential. Inverse Prob. 2005;21:1275–1290.
- Buterin SA . On inverse spectral problem for non-selfadjoint Sturm-Liouville operator on a finite interval. J Math Anal Appl. 2007;335:739–749.
- Albeverio S , Hryniv R , Mykytyuk Y . On spectra of non-self-adjoint Sturm-Liouville operators. Selecta Math. 2008;13:571–599.
- Horváth M , Kiss M . Stability of direct and inverse eigenvalue problems: the case of complex potentials. Inverse Probl. 2011;27:095007.
- Yurko VA . Method of spectral mappings in the inverse problem theory. Inverse and ill-posed problems series. Utrecht: VSP; 2002.
- Levin BY . Lectures on Entire Functions[English translation], V. Tkachenko (Transl.). Providence: American Mathematical Society; 1996.
- Bellman R , Cooke KL . Differential-difference equations. New York (NY): Academic Press; 1963.
- Marchenko VA . Sturm-Liouville operators and applications [English translation], A. Iacob (Transl.). Basel: Birkhäuser Verlag; 1986. Original work published 1977
- Conway JB . Functions of one complex variable. vol. 2. New York (NY): Springer-Verlag; 1995.
- Amirov RK . On Sturm-Liouville operators with discontinuity conditions inside an interval. J Math Anal Appl. 2006;317:163–176.
- Freiling G , Yurko VA . Inverse Sturm-Liouville Problems and Their Applications. Huntington: Nova Science Publishers; 2001.