Large Coupling Convergence Beyond Definiteness

Christian Koke · 2026

Read Large Coupling Convergence Beyond Definiteness by Christian Koke (2026) 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-2601-18055.txt.

Introduction

Christian Koke studies what happens when one term of an operator family is scaled up without bound. His 2026 preprint follows operators of the form A plus beta times B as beta tends to infinity. The family converges toward an effective operator defined on the kernel of B. The result shows how a strong constraint term forces the whole system onto a smaller space.

Catalog record

CREATORChristian Koke
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE88 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-18055

Browse the full library · About the Tower of Babel

Full text

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

Large Coupling Convergence Beyond Definiteness Christian Koke Abstract . We study convergence of operator families of the form A β = A + β ​ B A_{\beta}=A+\beta B towards an effective operator defined on ker ⁡ ( B ) \ker(B) , as the coupling constant β \beta tends to infinity. Crucially, we focus on the setting where neither A A nor B B can be assumed to be positive- (or negative-) semi-definite. We are hence outside the classical form-theoretic framework, where results based on Kato’s monotone convergence theorem would be applicable. Thus, instead of form methods, our approach builds on classical resolvent identities to study convergence of the family { A β } β \{A_{\beta}\}_{\beta} . Our findings are that: (i) Strong resolvent convergence holds (without further spectral assumptions) if A + β ​ B A+\beta B is self-adjoint and the compression of A A onto ker ⁡ ( B ) \ker(B) is well behaved. (ii) Under the more detailed assumption that 0 ∈ σ ⁡ ( B ) 0\in\sigma(B) is isolated, norm resolvent convergence can be established even if A + β ​ B A+\beta B is merely closed, provided the quasinilpotent part of B B at zero vanishes and certain conditions on the interplay of A A and B B are met. Importantly, if B B is not self-adjoint we find that the limit operator not only depends on ker ⁡ ( B ) \ker(B) as a Hilbert space, but crucially also on the precise form of the Riesz projector at 0 ∈ σ ⁡ ( B ) 0\in\sigma(B) onto ker ⁡ ( B ) \ker(B) . 1. Introduction The asymptotic analysis of operator families of the form A_{\beta}=A+\beta B , where A is a densely defined non-negative self-adjoint operator and B is a non-negative perturbation, is by now well established. Arguably most attention has been paid to the fundamental example of A_{\beta} being an indexed family of Schrödinger operators, with A=-\Delta the free Hamiltonian and B=V a non-negative interaction potential [ 22 , 21 , 20 , 17 ] . In order to account for point interactions, as they appear in certain physical settings [ 1 ] , the condition that B be a true operator was subsequently relaxed: Instead of arguing at the level of operators A,B , the problem was lifted to the level of (densely defined) quadratic forms \mathfrak{a},\mathfrak{b} . When bounded from below, the form \mathfrak{a}_{\beta}=\mathfrak{a}+\beta\mathfrak{b} is then canonically associated a densely defined operator A_{\beta} via the Friedrichs extension. A rich detailed theory of such large coupling perturbations of quadratic forms – as well as resolvent convergence of their associated operators – has since been established [ 24 , 50 , 16 , 14 , 15 , 2 , 3 , 11 ] . Most of these works take as their starting point the abstract framework of strongly perturbed quadratic forms a_{\beta}=\mathfrak{a}+\beta\mathfrak{b} initially developed by Kato [ 31 ] and later extended by Simon [ 46 ] who in their monotone convergence theorems established (generalized) strong resolvent convergence of the operator A_{\beta} associated to \mathfrak{a}_{\beta} to a limit operator \underline{A} defined on \ker(\mathfrak{q}) . Common among most of these works is the need for the considered quadratic forms \mathfrak{a},\mathfrak{b} (and hence also their associated operators A,B ) be positive semi-definite. Surveys into the terrain outside the realm of positive definiteness assumptions have so far been limited: In [ 12 ] e.g. the case where \mathfrak{a} and -\mathfrak{b} (as opposed to \mathfrak{b} ) are assumed to be non-negative, is treated. The series of works [ 28 , 29 , 27 ] considered the setting of Schrödinger operators with strong magnetic fields, and [ 10 , 23 ] considered the example of convergence of two dimensional Dirac operators perturbed by an increasing mass term supported only outside of a domain. In particular, the author is not aware of results on large coupling convergence in the sectorial- or more general closed setting. In this paper, we initiate a systematic study of large coupling convergence beyond positivity assumptions , where neither the perturbing operator B nor the background operator A is assumed to be associated with a semi-definite form. In contrast to most prior approaches, we hence avoid form methods altogether and instead directly work at the abstract operator level A_{\beta}=A+\beta B . Under suitable assumptions, we then establish convergence of the family of resolvents \{(A+\beta B-z)^{-1}\}_{\beta} as \beta\rightarrow\infty . A necessary condition for the resolvents in this family to be well-defined, is that the operators in the family \{A_{\beta}\}_{\beta} are closed, at least for sufficiently large \beta\gg 1 . Outside of lower-boundedness assumptions, there exist two typical settings that guarantee that the sum of two operators A,B is closed [ 31 , 51 ] , which we will hence also consider here: (1) A is bounded relative to B : If A is bounded with respect to a closed operator B , then by definition \mathscr{D}(B)\subseteq\mathscr{D}(A) and there are constants a,b\geq 0 so that (1.1) \displaystyle\quad\|A\psi\|\leq a\|B\psi\|+b\|\psi\|,\quad\forall\psi\in\mathscr{D}(B). Clearly then, A is also relatively bounded with respect to \beta B and by rescaling \beta if necessary, we may assume without loss of generality, that a<1 as soon as ( 1.1 ) is true for any generic a>0 . Standard results (e.g. [ 31 , Theorem IV.1.1] ) then guarantee that A+\beta B is closed on \mathscr{D}(B) . (2) B is bounded and A is closed: Clearly adding a bounded operator to (A,\mathscr{D}(A)) yields an operator that is closed on the same domain. In these settings, we then determine conditions under which the resolvent family \{(A+\beta B-z)^{-1}\}_{\beta} converges to the resolvent of a limit operator \underline{A} densely defined on \ker{B} : In Section 2 we consider the setting where both A and B are still self-adjoint, but no longer positive- (or negative-) semi-definite. For self-adjoint B , the spectral projection P_{\{0\}} then provides a canonical projection onto \ker{B} , and we show that as soon as the compression \underline{A}=PA on \mathscr{D}(A)\cap P\mathscr{A} of A onto \ker{B} is well-behaved, we have strong resolvent convergence. In Section 3 , we then consider the more general non-self-adjoint setting: At a fundamental level, if B is no longer self-adjoint (or more generally normal), we lose access to the Borel functional calculus and only retain the holomorphic one. In order to still define a spectral projection onto \ker(B) , we thus need to assume that 0\in\sigma(B) is isolated, as only then the Riesz-projector onto \{0\} is well defined. This is a stronger assumption on the spectrum of B around 0 than in the preceding section, which explains why it already allows to establish the stronger notion of norm -resolvent convergence in many settings. Crucially, we find that within this general non-self-adjoint setting, we need to assume that the quasi-nilpotent part of B at 0\in\sigma(B) vanishes, if we want to hope for convergence at all. Additionally, we find that the limit operator \underline{A} towards which A_{\beta} convergence in a generalized norm resolvent sense not only depends on \ker{B} as a linear space, but also on the way the Riesz projector P of B at 0\in\sigma(B) projects onto \ker(B) . Throughout our paper, we illustrate our results using examples from partial differential equations and abstract

Read the complete text