Higher order spectral shift of Euclidean Callias operators

Oliver Fürst · 2025

Read Higher order spectral shift of Euclidean Callias operators by Oliver Fürst (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-2506-01647.txt.

Introduction

Fürst presents a 2025 mathematics paper on trace formulas for Callias-type operators, a Dirac-style family over odd-dimensional space. Classic results need the potential to settle outside a core region, making the operator Fredholm. Fürst drops that demand, letting the potential take values in unbounded operators and misbehave at infinity. The operator then escapes the Fredholm world entirely.

A trace formula still survives, now written with higher order spectral shift functions. A tidy functional equation ties the pieces together, widening one found earlier by Pushnitski. Under a mild pointwise condition, a regularized index exists even here. The work claims the first multi-dimensional non-Fredholm extension of the Callias index story.

Catalog record

CREATOROliver Fürst
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE135 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-01647

Browse the full library · About the Tower of Babel

Full text

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

Higher order spectral shift of Euclidean Callias operators Oliver Fürst Abstract We consider Dirac-Schrödinger operators over odd-dimensional Euclidean space. The conditions for the potential are based on those of C. Callias in his famous paper on the corresponding index problem. However, we treat the case where the potential can take values in unbounded operators of a separable Hilbert space, and crucially, we also do not assume that the potential needs to be invertible outside a compact region. Hence, the Dirac-Schrödinger operator is not necessarily Fredholm. In the setup we discuss, it however still admits a related trace formula in terms of the underlying potential. In this paper we express the trace formula for these Callias-type operators in terms of higher order spectral shift functions, leading to a functional equation which generalizes a known functional equation found first by A. Pushnitski. To the knowledge of the author, this paper presents the first multi-dimensional non-Fredholm extension of the Callias index theorem involving higher order spectral shift functions. More precisely, we also show that under a Lebesgue point condition on the higher order spectral shift function associated to the potential, the Callias-type operator admits a regularized index, even in non-Fredholm settings. This corresponds to a known Witten index result in the one-dimensional case shown by A. Carey et al. The regularized index that we introduce is a minor extension of the classical Witten index, and we present an index formula, which generalizes the classical Callias index theorem. As an example, we treat the case of ( d + 1 ) (d+1) -massless Dirac-Schrödinger operators, for which we calculate the associated higher order spectral shift functions. 1 Introduction The main goal of this paper is to generalize the classical Callias index theorem to the non-Fredholm case by regularizing both sides of the index formula through higher order spectral shift functions. Moreover we present a functional equation for the involved spectral shift functions. Before we proceed in more detail, we begin with some historical background. C. Callias considered in his paper on axial anomalies [ 11 ] the index problem for a Dirac-Schrödinger operator D with matrix-valued potential A in odd dimension d , i.e. \displaystyle D=\mathbbm{i}c\nabla\otimes\mathbbm{1}_{\mathbb{C}^{m}}+\mathbbm{1}_{\mathbb{C}^{r}}\otimes M_{A}, (1.1) where M_{A} is the multiplication operator in L^{2}(\mathbb{R}^{d},\mathbb{C}^{m}) induced by the matrix function \mathbb{R}^{d}\ni x\mapsto A(x)\in\mathbb{C}^{m\times m} , and \displaystyle c\nabla=\sum_{i=1}^{d}c^{i}\partial_{x^{i}}, (1.2) where (c^{i})_{i=1}^{d} are Clifford matrices of rank r=2^{\frac{d-1}{2}} , i.e. for i,j=1,\ldots,d , \displaystyle c^{i}c^{j}+c^{j}c^{i}=-2\delta_{ij}\mathbbm{1}_{\mathbb{C}^{r}}. (1.3) The central result of Callias’ paper is the following index theorem ( [ 11 ] [Theorem 2]): Let A be a smooth family of m\times m -matrices with \left|A\left(x\right)\right|\geq C>0 for all \left|x\right|\geq c for some constants c , C . Assume that A is asymptotically homogeneous of order 0 as x\to\infty . Then D is Fredholm, and, with U\left(x\right):=A\left(x\right)\left|A\left(x\right)\right|^{-1} , k=\frac{d-1}{2} , \displaystyle\operatorname{ind}D=\frac{1}{2k!}\left(\frac{\mathbbm{i}}{8\pi}\right)^{k}\lim_{R\to\infty}\int_{S_{R}}\operatorname{Tr}_{\mathbb{C}^{m}}U\left(\operatorname{d}U\right)^{\wedge\left(d-1\right)}, (1.4) here, S_{R} denotes the (d-1) -sphere of radius R around 0 , and \left(\operatorname{d}U\right)^{\wedge\left(d-1\right)} the (d-1) -fold exterior product of (matrix-)one-forms \operatorname{d}U . There are some issues with the original proof (for example, trace-class membership of involved operators), which, however, have been rectified by F. Gesztesy and M. Waurick in [ 23 ] , where the authors also generalize some of the conditions. Since the publication of Callias’ paper, a plethora of generalizations have emerged in the Fredholm case, for which we suggest the following incomplete (and subjective) list of papers: [ 7 ] , [ 1 ] , [ 9 ] , [ 2 ] , [ 33 ] , [ 10 ] , [ 27 ] , [ 14 ] , [ 26 ] , [ 8 ] , [ 35 ] , [ 16 ] , [ 15 ] , [ 34 ] . In any case, some condition of invertibility of the potential outside a compact region is necessary in the Fredholm case. In this paper, we crucially suspend this assumption. In this case, one still retains a trace formula for the semi-group difference, \displaystyle\operatorname{Tr}_{L^{2}(\mathbb{R}^{d},H)}\operatorname{tr}_{\mathbb{C}^{r}}\left(e^{-tD^{\ast}D}-e^{-tDD^{\ast}}\right), (1.5) where H is a complex separable Hilbert space, which takes the role of \mathbb{C}^{m} . This trace formula, formulated in terms of the underlying potential A (allowed to be unbounded operator-valued), is the central result of [ 17 ] , which we shall discuss in more detail below. In the one-dimensional case d=1 , where Callias type index theorems correspond to the famous ”Index=Spectral Flow”-theorem, there is already a comprehensive body of work ( [ 32 ] , [ 20 ] , [ 13 ] , [ 12 ] to give a (incomplete) selection of the main results) dealing with the non-Fredholm case. The key to an invariant description beyond comparative trace formulas in terms of resolvents or semigroups is the translation into a functional equation of two spectral shift functions. The first result of this kind is due to A. Pushnitski in [ 32 ] [Theorem 1.1], which we quote and paraphrase here: Let (A(t))_{t\in\mathbb{R}} be a family of self-adjoint operators in H . If A_{\pm}=\lim_{t\to\pm\infty}A(t) exist in an appropriate sense, and \displaystyle\int_{\mathbb{R}}\left\|A^{\prime}(t)\right\|_{S^{1}(H)}\operatorname{d}t<\infty, (1.6) then there exist unique 1 1 1 We omit the normalization procedure here, since spectral shift functions are generally just unique up to a constant. Krein spectral shift functions of the pairs (A_{+},A_{-}) and (DD^{\ast},D^{\ast}D) . The Krein spectral shift function \xi_{T,S}:\mathbb{R}\rightarrow\mathbb{R} of a suitable pair of self-adjoint operators (T,S) has the property that for admissible functions f:\mathbb{R}\rightarrow\mathbb{C} , \displaystyle\operatorname{Tr}(f(T)-f(S))=\int_{\mathbb{R}}f^{\prime}(\lambda)\xi_{T,S}(\lambda)\operatorname{d}\lambda. (1.7) The spectral shift functions then satisfy the functional equation \displaystyle\xi_{DD^{\ast},D^{\ast}D}(\lambda)=\frac{1}{\pi}\int_{-\sqrt{\lambda}}^{\sqrt{\lambda}}\xi_{A_{+},A_{-}}(s)\frac{\operatorname{d}s}{\sqrt{\lambda-s^{2}}}, (1.8) for almost every \lambda>0 . In [ 13 ] [Theorem 4.3] the authors show that if \xi_{A_{+},A_{-}} admits a left- and right Lebesgue point at 0 , then also \xi_{DD^{\ast},D^{\ast}D} admits a right Lebesgue point at 0 , and \displaystyle\xi_{DD^{\ast},D^{\ast}D}(0+)=\frac{\xi_{A_{+},A_{-}}(0-)+\xi_{A_{+},A_{-}}(0+)}{2}. (1.9) Moreover, it is shown that the value at the right Lebesgue point 0 , the value \xi_{DD^{\ast},D^{\ast}D}(0+) gives an invariant description of the Witten index \operatorname{ind}_{W}D , which was introduced in [ 22 ] , independent of the chosen regularization (resolvent, semi-group). Since the Witten index agrees with the Fredholm index if both indices exist, and the spectral flow of a Fredholm family coincides with the spectral shift function of its endpoints (again, if both objects exist), formula ( 1.9 ) can be regarded as a non-Fredholm extension of the ”Index=Spectral Flow”-theorem. Since we also have a principal trace formula in the other cases d\geq 3 odd 2 2 2 For d even the trace is always trivial. , presented in [ 17 ] [Theorem 6.5], it is suggestive that it should be the starting point to develop a framework centred around spectral shift functions, allowing us to rewrite the trace formula invariantly through a functional equation like ( 1.8 ). For the sake of this introduction we present the principal trace formula without the precise conditions on the potential A , which takes values in the self-adjoint operators of a separable Hilbert space H with common domain \operatorname{dom}A_{0} of a given self-adjoint ”model operator” A_{0} (In the one-dimensional case discussed above one simply puts A_{0}=A_{-} ). Under the assumptions made in [ 17 ] [Hypothesis 2.11] on A=A_{0}+B we have for t>0 , \displaystyle\operatorname{Tr}_{L^{2}(\mathbb{R}^{d},H)}\operatorname{tr}_{\mathbb{C}^{r}}\left(e^{-tD^{\ast}D}-e^{-tDD^{\ast}}\right) \displaystyle= \displaystyle\frac{2}{d}(4\pi)^{-\frac{d}{2}}t^{\frac{d}{2}}\int_{\mathbb{R}^{d}}\int_{s\in\Delta_{d-1}}\operatorname{Tr}_{\mathbb{C}^{r}\otimes H}\left(\prod_{j=0}^{d-1}\left(\mathbbm{i}c\nabla A_{\phi}\right)(x)e^{-ts_{j}A_{\phi}^{2}(x)}\right)\operatorname{d}s~\operatorname{d}x, (1.10) where \phi\in C_{c}^{\infty}(\mathbb{R}^{d}) , such that \phi\equiv 1 near 0 is arbitrary and A_{\phi}:=A_{0}+(1-\phi)(A-A_{0}) .

Read the complete text