Approximation of magnetic Schrödinger operators with δ \delta -interactions supported on networks

Markus Holzmann · 2025

Read Approximation of magnetic Schrödinger operators with δ \delta -interactions supported on networks by Markus Holzmann (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-08301.txt.

Introduction

Markus Holzmann studies operators that model quantum particles on networks, where the particle feels a magnetic field and a sharp interaction concentrated on the network itself. Such singular interactions are idealizations, useful but hard to handle directly. Replacing them with smooth, regular interactions that give nearly the same behavior makes both math and computation safer.

Holzmann proves that the smooth versions converge to the singular operator in a strong sense that also controls its spectrum. The network can be a union of smooth surfaces, a graph in the plane, or the border of a region with corners. The result allows complex coefficients and weak assumptions on the fields. Spectral consequences of the approximation are also drawn out.

Catalog record

CREATORMarkus Holzmann
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE63 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-08301

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Approximation of magnetic Schrödinger operators with δ \delta -interactions supported on networks Markus Holzmann Abstract . This paper deals with the approximation of a magnetic Schrödinger operator with a singular δ \delta -potential that is formally given by ( i ∇ + A ) 2 + Q + α δ Σ (i\nabla+A)^{2}+Q+\alpha\delta_{\Sigma} by Schrödinger operators with regular potentials in the norm resolvent sense. This is done for Σ \Sigma being the finite union of C 2 C^{2} -hypersurfaces, for coefficients A A , Q Q , and α \alpha under minimal assumptions such that the associated quadratic forms are closed and sectorial, and Q Q and α \alpha are allowed to be complex-valued functions. In particular, Σ \Sigma can be a graph in ℝ 2 \mathbb{R}^{2} or the boundary of a piecewise C 2 C^{2} -domain. Moreover, spectral implications of the mentioned convergence result are discussed. 2020 Mathematics Subject Classification Primary 81Q10; Secondary 35Q40 1. Introduction Schrödinger operators with singular \delta -potentials supported on sets of measure zero play an important role in mathematical physics. Singular potentials supported on points were already used in the early days of quantum mechanics [ 29 , 39 ] and later systematically analyzed from a mathematical point of view, see the monograph [ 1 ] and the references therein. In the last three decades, there was a substantial interest in singular potentials supported on hypersurfaces \Sigma in \mathbb{R}^{n} , n\geq 2 , like curves in \mathbb{R}^{2} or surfaces in \mathbb{R}^{3} , see the review paper [ 16 ] , the monograph [ 18 ] , or the research papers [ 5 , 6 , 9 , 14 , 15 , 17 , 21 , 33 ] , to mention just a few selected works. Such operators are formally given by H_{A,Q,\alpha}=(i\nabla+A)^{2}+Q+\alpha\delta_{\Sigma}, (1.1) and they describe for attractive interactions (i.e. \alpha<0 ) the propagation of a quantum particle on \Sigma under the influence of a magnetic vector potential A and an electric potential Q allowing the tunneling of the particle between different parts of \Sigma , see [ 5 , 16 ] ; from this point of view, graph like structures \Sigma\subset\mathbb{R}^{2} are of particular interest. Moreover, operators of the form H_{0,0,\alpha} are useful in the design of photonic crystals with prescribed spectral gaps [ 22 , 26 ] . Of course, the operators with the singular potentials are only idealized replacements of more realistic operators with regular potentials, and so the relation of the regular and the singular models was justified via an approximation procedure already in [ 29 ] . For A=Q=0 the approximation of one, two, and three-dimensional Schrödinger operators with \delta -point interactions by Schrödinger operators with scaled regular potentials in the norm resolvent sense was treated systematically in the 1970s and 1980s, see the monograph [ 1 ] and the references therein; the latter convergence implies, in particular, the convergence of the spectra and of functions of the involved operators [ 27 , 40 ] and thus, it justifies the usage of the idealized models as good approximations of their regular counterparts. The approximation of Schrödinger operators with \delta -potentials supported on the sphere in \mathbb{R}^{3} was considered in [ 2 , 37 ] , then the situation that \Sigma is the boundary of a star shaped domain and \alpha is real-valued and smooth was investigated in [ 36 ] , and unbounded and C^{2} -smooth \Sigma in \mathbb{R}^{2} and \mathbb{R}^{3} and constant \alpha\in\mathbb{R} were treated in [ 17 , 21 ] . It is worth mentioning that in [ 38 ] Laplacians with \delta - (and other measure valued) potentials supported on quite general geometries were approximated, but only in the strong resolvent sense, which does not imply the convergence of the spectra of the approximating operators in the Hausdorff metric. Finally, in [ 4 ] the approximation of the Schrödinger operator with a \delta -potential of real-valued strength \alpha\in L^{\infty}(\Sigma) on one bounded or unbounded C^{2} -smooth hypersurface \Sigma\subset\mathbb{R}^{n} , n\geq 2 , was shown, and in [ 5 ] the approximation of a two-dimensional Schrödinger operator with a homogeneous magnetic field and an electric \delta -potential of real-valued strength \alpha\in L^{\infty}(\Sigma) supported on a bounded C^{1,1} -smooth curve was treated. The main motivation in this paper is to show that the operator H_{A,Q,\alpha} formally given by ( 1.1 ) can be approximated in the norm resolvent sense by Schrödinger operators with scaled regular potentials under minimal assumptions on A,Q , and \alpha , and for a wide class of interaction supports \Sigma . Concerning the coefficients, of course from the point of view of quantum mechanics, real-valued functions that are associated with self-adjoint operators are favored, but the analysis in this paper works equally well for complex-valued functions Q and \alpha , yielding a relation to works on Schrödinger operators with non-self-adjoint \delta -potentials [ 10 , 24 , 28 ] . Moreover, the interaction support \Sigma will be allowed to be the finite union of bounded or unbounded C^{2} -smooth hypersurfaces, and so classes of graph-like structures in \mathbb{R}^{2} , that are particularly important for the application to leaky quantum graphs, and piecewise C^{2} -smooth \Sigma\subset\mathbb{R}^{n} , which are an important subclass of Lipschitz domains, are contained. To set the stage in a rigorous way, let N,n\in\mathbb{N} be fixed with n\geq 2 and \Sigma^{(1)},\dots,\Sigma^{(N)}\subset\mathbb{R}^{n} be hypersurfaces as in Definition 2.5 with normal vector fields \nu^{(1)},\dots,\nu^{(N)} such that the Hausdorff measure of \Sigma^{(k)}\cap\Sigma^{(l)} is zero for k\neq l . Then, the network \Sigma is \Sigma:=\bigcup_{k=1}^{N}\Sigma^{(k)}. (1.2) Moreover, assume that A\in L^{2}_{\textup{loc}}(\mathbb{R}^{n};\mathbb{R}^{n}) (1.3) and Q=Q_{1}+Q_{2} with 0\leq Q_{1}\in L^{1}_{\textup{loc}}(\mathbb{R}^{n}),\,Q_{2}\in L^{p^{\prime}}(\mathbb{R}^{n})+L^{\infty}(\mathbb{R}^{n})\text{ with }p^{\prime}\begin{cases}\geq n/2,&n>2,\\ >1,&n=2,\end{cases} (1.4) are fixed. Note that Q_{1} is automatically real-valued, while Q_{2} is allowed to be complex-valued. Next, define the space \mathcal{H}_{A,Q}:=\left\{f\in L^{2}(\mathbb{R}^{n}):|(i\nabla+A)f|,Q_{1}^{1/2}f\in L^{2}(\mathbb{R}^{n})\right\}. (1.5) This space is studied in detail in Section 2.1 , where it is shown that \mathcal{H}_{A,Q} , when endowed with a proper norm, is a Hilbert space and that it is indeed independent of Q_{2} . Moreover, in Section 2.2 it is proved that functions f\in\mathcal{H}_{A,Q} posses Dirichlet traces f|_{\Sigma^{(k)}}\in L^{q}(\Sigma^{(k)}) with q=2\frac{n-1}{n-2+\mu} and \mu\in(0,1] . Eventually, assume that \alpha:=(\alpha^{(k)})_{k=1}^{N}\in L^{p}(\Sigma)+L^{\infty}(\Sigma):=\bigoplus_{k=1}^{N}\bigl(L^{p}(\Sigma^{(k)})+L^{\infty}(\Sigma^{(k)})\bigr) (1.6) with p=\frac{n-1}{1-\gamma} , where \gamma\in(0,1) is fixed. To introduce the Schrödinger operator with a \delta -potential of strength \alpha supported on the network \Sigma , consider the quadratic form \mathfrak{h}_{A,Q,\alpha}[f]:=\int_{\mathbb{R}^{n}}\bigl(\bigl|(i\nabla+A)f\bigr|^{2}+Q|f|^{2}\bigr)\textup{d}x+\int_{\Sigma}\alpha\bigl|f|_{\Sigma}\bigr|^{2}\textup{d}\sigma,\quad\mathrm{dom}\,\mathfrak{h}_{A,Q,\alpha}=\mathcal{H}_{A,Q}, (1.7) where the shortcut notation \int_{\Sigma}\alpha|f|_{\Sigma}|^{2}\textup{d}\sigma:=\sum_{k=1}^{N}\int_{\Sigma^{(k)}}\alpha^{(k)}|f|_{\Sigma^{(k)}}|^{2}\textup{d}\sigma is used. We will see in Proposition 2.8 that \mathfrak{h}_{A,Q,\alpha} is a densely defined, closed, and sectorial form in L^{2}(\mathbb{R}^{n}) . The associated m -sectorial operator H_{A,Q,\alpha} is the rigorous mathematical operator that is associated with the formal expression ( 1.1 ). In order to construct the sequence of operators H_{A,Q,V_{\varepsilon}} that converges to H_{A,Q,\alpha} in the norm resolvent sense, note first that there exists \beta>0 such that for each k\in\{1,\dots,N\} the mapping \iota^{(k)}:\Sigma^{(k)}\times(-\beta,\beta)\rightarrow\mathbb{R}^{n},\quad\iota^{(k)}(x_{\Sigma^{(k)}},t):=x_{\Sigma^{(k)}}+t\nu^{(k)}(x_{\Sigma^{(k)}}), (1.8) is injective, see Proposition 2.6 . Introduce for \varepsilon\in(0,\beta] the set \Omega_{\varepsilon}^{(k)}:=\iota^{(k)}(\Sigma^{(k)}\times(-\varepsilon,\varepsilon)) , and choose a function V^{(k)}\in L^{p}(\mathbb{R}^{n})+L^{\infty}(\mathbb{R}^{n}) with p=\frac{n-1}{1-\gamma} for a \gamma\in(0,1) that is supported in \Omega_{\beta}^{(k)} . Then, one can define for k\in\{1,\dots,N\} V_{\varepsilon}^{(k)}(x):=\begin{cases}\frac{\beta}{\varepsilon}V^{(k)}(x_{\Sigma^{(k)}}+\frac{\beta}{\varepsilon}t\nu^{(k)}(x_{\Sigma^{(k)}})),&\text{if }x=x_{\Sigma^{(k)}}+t\nu^{(k)}(x_{\Sigma^{(k)}})\in\Omega_{\varepsilon}^{(k)},\\ 0,&\text{if }x\notin\Omega_{\varepsilon}^{(k)},\end{cases} and V_{\varepsilon}:=\sum_{k=1}^{N}V_{\varepsilon}^{(k)}; (1.9) it follows from Proposition 2.6 (iv) that V_{\varepsilon}\in L^{p}(\mathbb{R}^{n})+L^{\infty}(\mathbb{R}^{n}) . Consider the quadratic form \mathfrak{h}_{A,Q,V_{\varepsilon}}[f]:=\int_{\mathbb{R}^{n}}\bigl(\bigl|(i\nabla+A)f\bigr|^{2}+(Q+V_{\varepsilon})|f|^{2}\bigr)\textup{d}x,\quad\mathrm{dom}\,\mathfrak{h}_{A,Q,\alpha}=\mathcal{H}_{A,Q}. (1.10) It will turn out as a byproduct in

Read the complete text