The spectrum of non-local discrete Schrödinger operators with a δ-potential
- Fumio Hiroshima^{1}Email author and
- József Lőrinczi^{2}
https://doi.org/10.1186/s40736-014-0007-8
© Hiroshima and Lörinczi; Licensee Springer 2014
Received: 8 February 2014
Accepted: 3 March 2014
Published: 4 October 2014
Abstract
The behaviour of the spectral edges (embedded eigenvalues and resonances) is discussed at the two ends of the continuous spectrum of non-local discrete Schrödinger operators with a δ-potential. These operators arise by replacing the discrete Laplacian by a strictly increasing C^{1}-function of the discrete Laplacian. The dependence of the results on this function and the lattice dimension are explicitly derived. It is found that while in the case of the discrete Schrödinger operator these behaviours are the same no matter which end of the continuous spectrum is considered, an asymmetry occurs for the non-local cases. A classification with respect to the spectral edge behaviour is also offered.
Keywords
Introduction
1.1 Non-local discrete Schrödinger operators on lattice
The spectrum of discrete Schrödinger operators has been widely studied for both combinatorial Laplacians and quantum graphs; for some recent summaries see [3,4,7,9,11,16,19] and the references therein. Specifically, eigenvalue behaviours of discrete Schrödinger operators on \(l^{2}\left (\mathbb {Z}^{d}\right)\) are discussed in e.g. [2,8,10,12,13,15]. However, for discrete non-local Schrödinger operators only few results are known. Typical examples include discrete fractional Schrödinger operators.
In this paper we define generalized discrete Schrödinger operators which include discrete fractional Schrödinger operators and others whose counterparts on \(L^{2}\left (\mathbb {R}^{d}\right)\) are currently much studied [5,6,17,18]. In [14] we have introduced a class of generalized Schrödinger operators whose kinetic term is given by so called Bernstein functions of the Laplacian. These operators are non-local and via a Feynman-Kac representation generate subordinate Brownian motion killed at a rate given by the potential. Their discrete counterparts studied in this paper also have a probabilistic interpretation in that they generate continuous time random walks with jumps on \(\mathbb {Z}^{d}\).
In the present paper we consider a class of Schrödinger operators obtained as a strictly increasing C^{1}-function of the discrete Laplacian and a δ-potential. This includes, in particular, Bernstein functions (see below) of the discrete Laplacian. In the presence of a δ-potential the above probabilistic picture then describes free motion with a “bump” which can be interpreted as an impurity in space. Our aim here is to investigate the spectrum of such operators, specifically, embedded eigenvalues and resonances at the edges of the continuous spectrum.
is the discrete Schrödinger operator with δ-potential. In order to define a non-local version of h, we use Fourier transform on \(l^{2}\left (\mathbb {Z}^{d}\right)\). Let \(\mathbb {T}^{d} =\left [-\pi, \pi \right ]^{d}\) be the d-dimensional torus, and set
The scalar product on is denoted by \((\,f,g)=\int _{\mathbb {T}^{d}}\bar {f}(\theta) g(\theta) d\theta \). The Fourier transform is then defined by for \(\theta = \left (\theta _{1},\ldots,\theta _{d}\right)\in \mathbb {T}^{d} \). Then the discrete Laplacian L transforms as
i.e., the right hand side above is a multiplication operator on . In this paper we use a non-local discrete Laplacian Ψ(L) defined for a suitable function Ψ by applying Fourier transform.
Definition1.
For a given Ψ∈C^{1}((0,∞)) such that dΨ(x)/dx>0, x∈(0,∞), we define the non-local discrete LaplacianΨ(L) by
non-local discrete Schrödinger operator with δ-potential.
An example of such a function is Ψ(u)=u^{α/2}, 0<α<2, which describes a discrete Laplacian of fractional order α/2. Other specific choices will be given in Example 1 below.
Under Fourier transform (1.7) is mapped into
Since σ(L)=[0,2] and Ψ is strictly increasing, it is immediate that σ(Ψ(L))=Ψ([0,2])=[Ψ(0),Ψ(2)]. In what follows we consider the spectrum of H_{ v } instead of h. Note that the map Φ↦v(Ω,Φ)Ω is a rank-one operator, and thus the continuous spectrum of the rank-one perturbation H_{ v } of L is [Ψ(0),Ψ(2)], for every \(v\in \mathbb {R}\). See e.g. [1,20] for rank-one perturbations.
1.2 Ψ(∗)-resonances and Ψ(∗)-modes
As it will be seen below, for a sufficiently large value of −v>0 there exists an eigenvalue E_{−}(v) of H_{ v } strictly smaller than Ψ(0). Suppose that E_{−}(v)↑Ψ(0) as v↑v_{0} with some v_{0}≠0. If Ψ(0) is an eigenvalue of \(H_{v_{0}}\), we call the eigenvector associated with Ψ(0) a Ψ(0)-mode. If Ψ(0) is not an eigenvalue of \(H_{v_{0}}\), we call it a Ψ(0)-resonance. Similarly, for a sufficiently large v>0 it will be seen that there exists an eigenvalue E_{+}(v) strictly larger than Ψ(2). Suppose that E_{+}(v)↓Ψ(2) as v↓v_{2} with some v_{2}≠0. If Ψ(2) is an eigenvalue of \(H_{v_{2}}\), we call the eigenvector associated with Ψ(2) a Ψ(2)-mode, and a Ψ(2)-resonance whenever Ψ(2) is not an eigenvalue of \(H_{v_{2}}\).
Modes and resonances of L + V
2-mode | 2-res. | 0-mode | 0-res. | |
---|---|---|---|---|
d=1 | No | No | No | No |
d=2 | No | No | No | No |
d=3 | No | Yes | No | Yes |
d=4 | No | Yes | No | Yes |
d≥5 | Yes | No | Yes | No |
Modes and resonances of\(\sqrt {L} +V\)
\({\sqrt 2}\)-mode | \({\sqrt 2}\)-res. | 0-mode | 0-res. | |
---|---|---|---|---|
d=1 | No | No | No | No |
d=2 | No | No | No | Yes |
d=3 | No | Yes | Yes | No |
d=4 | No | Yes | Yes | No |
d≥5 | Yes | No | Yes | No |
Eigenvalues
2.1 A criterion for determining the eigenvalues
The following result gives an integral test to spot the eigenvalues of H_{ v }.
Lemma1.
Proof.
To show the necessity part, suppose that E is an eigenvalue and Φ an associated eigenvector. Assuming (Ω,Φ)=0, we have H_{ v }Φ=Ψ(L(θ))Φ=EΦ. Since Ψ(L(θ)) has no point spectrum, this is a contradiction. This gives (Ω,Φ)≠0 and E−Ψ(L(θ))≠0 for almost every \(\theta \in \mathbb {T}^{d}\) with
Thus I(E)<∞ follows, and (Ω,Φ)≠0 implies J(E)≠0.
holds. By J(E)≠0 it follows that there exists v such that (2.5) is satisfied, hence E is an eigenvalue of H_{ v }.
In order to investigate Ψ(∗)-resonances and Ψ(∗)-modes we use Lemma 1 and estimate the two integrals I(E) and J(E) at the two ends E=Ψ(∗) of the interval [Ψ(0),Ψ(2)].
2.2 The location of eigenvalues
Lemma2.
Let \(E\in \mathbb {R}\setminus \left [\Psi (0),\Psi (2)\right ]\). Then there exists v≠0 such that E is an eigenvalue of H_{ v }.
Proof.
In this case it is easily seen that I(E)<∞ and J(E)≠0. Then E is an eigenvalue and v is given by (2.4).
Lemma3.
Let σ_{ p }(H_{ v }) be the point spectrum of H_{ v }. Then σ_{ p }(H_{ v })∩(Ψ(0),Ψ(2))=∅, for every v≠0.
Proof.
Next consider the cases E=Ψ(2) and E=Ψ(0). For a systematic discussion of the eigenvalue behaviour of H_{ v } we introduce the following concept.
Definition2.
Lemma4.
- (1)
J(E)≠0 for both E=Ψ(0) and E=Ψ(2).
- (2)
For E=Ψ(2) we have that I(E)<∞ if and only if d≥1+4a, and J(E)<∞ if and only if d≥1+2a.
- (3)
For E=Ψ(0) we have that I(E)<∞ if and only if d≥1+4b, and J(E)<∞ if and only if d≥1+2b.
Proof.
Since Ψ is strictly increasing, the first statement follows directly.
Thus the lemma follows for E=Ψ(2). For the case of E=Ψ(0) the proof is similar.
From these lemmas we can derive the spectral edge behaviour of H_{ v }. The next theorem is the main result in this paper.
Theorem1.
- (1)Suppose that v>0. Then the following cases occur:
- (i)
Let d<1+2b. Then for all v>0 there exists an eigenvalue E>Ψ(2).
- (ii)
Let 1+2b≤d<1+4b. Then for v>v_{2} there exists an eigenvalue E>Ψ(2), while for v≤v_{2} there is no eigenvalue.
- (iii)
Let 1+4b≤d. Then for v>v_{2} there exists an eigenvalue E>Ψ(2), for v=v_{2} the value E=Ψ(2) is an eigenvalue, while v<v_{2} there is no eigenvalue.
- (i)
- (2)Suppose that v<0. Then the following cases occur:
- (i)
Let d<1+2a. Then for all v<0 there exists an eigenvalue E<Ψ(0).
- (ii)
Let 1+2a≤d<1+4a. Then for v<v_{0} there exists an eigenvalue E<Ψ(0), while for v≥v_{0} there is no eigenvalue.
- (iii)
Let 1+4a≤d. Then for v<v_{0} there exists an eigenvalue E<Ψ(0), for v=v_{0} the value E=Ψ(0) is an eigenvalue, while for v>v_{0} there is no eigenvalue.
- (i)
Proof.
Consider the case v>0 and let d<1+2b. Then for all E>Ψ(2) we have I(E)<∞ and J(E)≠0. Thus there exists v such that E is an eigenvalue of H_{ v }.
Let 1+2b≤d<1+4b. Then for all E>Ψ(2) we have that I(E)<∞ and J(E)≠0. Thus E is an eigenvalue of H_{ v }. Since J(E)<∞, it follows that E↓Ψ(2) as v↓v_{2}>0. However, E=Ψ(2) is not an eigenvalue since I(E)=∞.
Let d≥1+4b. Then for all E>Ψ(2) we have I(E)<∞ and J(E)≠0. Thus E is an eigenvalue of H_{ v }. Since J(E)<∞, we obtain E↓Ψ(2) as v↓v_{2}>0. Since I(E)<∞, we have that E=Ψ(2) is also an eigenvalue. The cases for v<0 can be dealt with similarly.
Remark1.
Note that in general −v_{0}≠v_{2}.
Remark2.
Ψ (0) and Ψ (2)-modes and resonances
v>0 | Ψ(2)-mode | Ψ(2)-res. |
---|---|---|
d<1+2b | No | No |
1+2b≤d<1+4b | No | Yes |
1+4b≤d | Yes | No |
v<0 | Ψ (0)-mode | Ψ(0)-res. |
d<1+2a | No | No |
1+2a≤d<1+4a | No | Yes |
1+4a≤d | Yes | No |
It is worthwhile to see the implications more closely for some specific choices of function Ψ.
Example 1.
- (1)
Discrete Schrödinger operator: Let Ψ(u)=u. Then Ψ is of (1,1)-type and H_{ v }=L+V. See Table 1.
- (2)
Fractional discrete Schrödinger operator: Let Ψ(u)=u^{α/2} for 0<α<2. Then Ψ is of (α/2,1)-type and H_{ v }=L^{α/2}+V.
- (3)
Relativistic fractional discrete Schrödinger operator: Let Ψ(u)=(u+m^{2/α})^{α/2}−m for 0<α<2 and m>0. Then Ψ is of (1,1)-type.
- (4)
Discrete jump-diffusion operator: Let Ψ(u)=u+bu^{α/2} with 0<α<2. Then Ψ is of (α/2,1)-type.
- (5)
Rotationally symmetric geometric discrete α-stable operator: Let Ψ(u)= log (1+u^{α/2}) for 0<α<2. Then Ψ is of (α/2,1)-type.
- (6)
Higher order discrete Laplacian: Let Ψ(u)=u^{ β } for β>1. Then Ψ is of (β,1)-type.
- (7)
Bernstein functions of the discrete Laplacian: Let Ψ be a Bernstein function with vanishing right limits, i.e., \(\Psi : \mathbb {R}^{+} \to \mathbb {R}^{+}\) which can be represented in the form \(\Psi (u)=bu+\int _{0}^{\infty } \left (1-e^{-uy}\right)\nu (dy)\), where b≥0 and ν is a Lévy measure with mass on (0,∞) satisfying \(\int _{0}^{\infty } (1\wedge y) \nu (dy) < \infty \). Then it follows that \(\Psi ^{\prime }(2) =b+\int _{0}^{\infty } y e^{-2y} \nu (dy) \not =0\). Furthermore, since Ψ is concave, we have a=α/2 with some α≥2. Hence Ψ is of (α/2,1)-type with some 0≤α≤2. Note that the first five examples above are specific cases of Bernstein functions.
A classification of spectral edge behaviour
The functions Ψ of the discrete Laplacian can be classified according to the behaviour of the eigenvalues at the two ends of the interval [ Ψ(0),Ψ(2)].
Definition3.
We call Ψnormal type if Ψ is (1,1)-type, and fractional type if Ψ is (α/2,1)-type with 0<α<2.
The two types show qualitatively different behaviour and we discuss them separately.
3.1 Normal type
Let Ψ be of normal type. In this case the spectral edge behaviour is the same as that of the discrete Schrödinger operator L+V. The following result has been obtained in [15].
Proposition1.
- (1)
Let d=1 or 2. For every v>0 there exists an eigenvalue E>Ψ(2), and for every v<0 an eigenvalue E<Ψ(0).
- (2)
Let d=3 or 4. If v>0, then there exists v_{2}>0 such that for all v>v_{2} an eigenvalue E>Ψ(2) exists, and for v≤v_{2} no eigenvalue exists. If v<0, then there exists v_{0}<0 such that for all v<v_{0} an eigenvalue E<Ψ(0) exists, and for v<v_{0} no eigenvalue exists.
- (3)
Let d≥5. If v>0, then there exists v_{2}>0 such that for all v>v_{2} an eigenvalue E>Ψ(2) exists, for v=v_{2} the value E=Ψ(2) is an eigenvalue, and for v<v_{2} no eigenvalue exists. If v<0, then there exists v_{0}<0 such that for all v<v_{0} an eigenvalue E<Ψ(0) exists, for v=v_{0} the value E=Ψ(0) is an eigenvalue, and for v>v_{0} no eigenvalue exists.
Thus the spectral edge behaviour for positive and negative v is qualitatively the same, and the details only depend on the dimension d.
3.2 Fractional type
In the fractional type case we have the following spectral edge behaviour.
Theorem2.
- (1)
If v>0, then the spectral edge behaviour is the same as for normal type Ψ with v>0.
- (2)If v<0, then we have the following cases:
- (i)
Let d<1+α. Then for every v<0 there exists an eigenvalue E<0.
- (ii)
Let 1+α≤d<1+2α. There exists v_{0}<0 such that for all v<v_{0} an eigenvalue E<0 exists, while for v≤v_{0} no eigenvalue exists.
- (iii)
Let d≥1+2α. There exists v_{0}<0 such that for all v<v_{0} an eigenvalue E<0 exists, for v=v_{0} the value E=0 is an eigenvalue, and for v>v_{0} no eigenvalue exists.
- (i)
In the fractional case it is seen that the edge behaviours for positive and negative v are in general different from each other, in contrast with the normal type case.
3.2.1 The case of α=1
3.2.2 Massless and massive cases
Consider the Bernstein function \(\Psi (u)=\sqrt {u+m^{2}}-m\) with m≥0. This allows to define the relativistic discrete Schrödinger operator \(\sqrt {L+m^{2}}-m+V\). Then it follows that Ψ(u) is (1,1)-type for m>0, and (1/2,1)-type for m=0. In particular, the edge behaviours of \(\sqrt {L+m^{2}}-m+V\) for m>0 are symmetric, and the edge behaviours of \(\sqrt {L}+V\) and \(\sqrt {L+m^{2}}-m+V\) are different. More generally, consider the Bernstein function Ψ(u)=(u+m^{2/α})^{α/2}−m, with 0<α<2 and m≥0. This defines the rotationally symmetric relativistic α-stable operator (L+m^{2/α})^{α/2}−m. We conclude that Ψ(u) is of (1,1)-type for m>0 but of (α/2,1)-type for m=0. Thus the edge behaviours of (L+m^{2/α})^{α/2}−m+V and L^{α/2}+V are different.
Declarations
Acknowledgements
FH is financially supported by Grant-in-Aid for Science Research (B) 23340032 from JSPS. JL thanks Institut Mittag-Leffler, Stockholm, for the opportunity to organise the research-in-peace workshop “Lieb-Thirring-type bounds for a class of Feller processes perturbed by a potential" during the period 25 July – 9 August 2013. JL also thanks London Mathematical Society for a travel grant to this workshop. FH thanks the invitation to this workshop, and we both express our gratitude to IML for the kind hospitality and inspiring work environment, where most of this paper has been prepared.
Authors’ Affiliations
References
- Aronszajn, N.: On a problem of Weyl in the theory of singular Strum-Liouville equations. Am. J. Math. 79, 597–610 (1957).MathSciNetView ArticleGoogle Scholar
- Bellissard, J., Schulz-Baldes, H.: Scattering theory for lattice operators in dimension d ≥3. arXiv:1109.5459v2 (2012).Google Scholar
- Berkolaiko, G., Carlson, R., Fulling, SA., Kuchment, PA., (Eds): Quantum graphs and their applications, Vol. 415 (2006).Google Scholar
- Berkolaiko, G., Kuchment, PA.: Introduction to Quantum Graphs. AMS Math. Surv. Monographs186 (2012).Google Scholar
- Bogdan, K., Byczkowski, T.: Potential theory of Schrödinger operator based on fractional Laplacian. Probab. Math. Stat. 20, 293–335 (2000).MathSciNetGoogle Scholar
- Potential analysis of stable processes and its extensions. Lecture Notes in Mathematics(Bogdan, K., et al., eds.), Vol. 1980. Springer (2011).Google Scholar
- Chung, F.: Spectral Graph Theory. CBMS Regional Conf. Series Math, Washington DC (1997).Google Scholar
- Damanik, D., Killip, R., Simon, B.: Schrödinger operators with few bound states. Commun. Math. Phys. 258, 741–750 (2005).MathSciNetView ArticleGoogle Scholar
- Exner, P., Keating, JP., Kuchment, PA., Sunada, T., Teplyaev, A., (Eds): Analysis on graphs and its applications, Vol. 77 (2008).Google Scholar
- Exner, P., Kuchment, PA., Winn, B.: On the location of spectral edges in -peridoc media. J. Phys. A43, 474022 (2010).MathSciNetView ArticleGoogle Scholar
- Grigor’yan, A.: Heat kernels on manifolds, graphs and fractals. In: European Congress of Mathematics, Barcelona, July 10-14, 2000, Progress in Mathematics 201, pp. 393–406. Birkhäuser (2001).Google Scholar
- Higuchi, Y., Matsumoto, T., Ogurisu, O.: On the spectrum of a discrete Laplacian on with finitely supported potential. Linear Multilinear Algebra59, 917–927 (2011).MathSciNetView ArticleGoogle Scholar
- Higuchi, Y., Shirai, T.: The spectrum of magnetic Schrödinger operators on a graph with periodic structure. J. Funct. Anal. 169, 456–480 (1999).MathSciNetView ArticleGoogle Scholar
- Hiroshima, F., Ichinose, T., Lőrinczi, J.: Path integral representation for Schrödinger operators with Bernstein functions of the Laplacian. Rev. Math. Phys. 24, 1250013 (2012).MathSciNetView ArticleGoogle Scholar
- Hiroshima, F., Sasaki, I., Shirai, T., Suzuki, A.: Note on the spectrum of discrete Schrödinger operators. J. Math-for-Industry4, 105–108 (2012).MathSciNetGoogle Scholar
- Korotyaev, E., Saburova, N.: Schrödinger operators on periodic discrete graphs. arXiv:1307.1841 (2013).Google Scholar
- Kulczycki, T.: Gradient estimates of q-harmonic functions of fractional Schrödinger operator. Potential Anal. 39, 69–98 (2013).MathSciNetView ArticleGoogle Scholar
- Lőrinczi, J., Małecki, J.: Spectral properties of the massless relativistic harmonic oscillator. J. Diff. Eq. 253, 2846–2871 (2012).View ArticleGoogle Scholar
- Post, O.: Spectral Analysis on Graph-Like Spaces, Lecture Notes in Mathematics, Vol. 2039. Springer (2012).Google Scholar
- Simon, B., Wolff, T.: Singular continuous spectrum under rank one perturbations and localization for random Hamiltonians. Commun. Pure App. Math. 39, 75–90 (1986).MathSciNetView ArticleGoogle Scholar
Copyright
This article is published under license to BioMed Central Ltd. \This is an Open Access article distributed under the terms of the Creative Commons Attribution License(http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.