Pith. sign in

REVIEW 2 major objections 3 minor 8 references

Spectral localizers in KK-theory

T0 review · 2 major / 3 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper proves that the even Kasparov-product index homomorphism is computed by a spectral localizer built from smooth cutoffs of the Dirac operator.

desk verdict Solid extension of spectral localizer methods to KK-theory; the proof is largely sound, with a couple of standard but undeferred steps that need referee attention. read the letter →

arxiv 2508.08668 v1 pith:W6JTPYYI submitted 2025-08-12 math.KT math.OA

Jens Kaad This is my paper · ORCID
classification math.KTmath.OA MSC 19K3546L8019K5658B3446L08
keywords KK-theoryspectrallocalizerindexhomomorphismunboundedKasparovmodulesrelativeK-theoryevenHilbertC*-modulesproduct
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper establishes that the index homomorphism induced by a class in even KK-theory via the Kasparov product can be computed from a spectral localizer: a selfadjoint operator built from smooth compactly supported functions of an unbounded Dirac operator, rather than from its spectral projections. The formula only reads the part of the Dirac spectrum lying in a compact interval, with the interval size controlled by explicit operator norms, and it works at the full generality of Hilbert C*-modules where spectral projections may not exist. The proof first treats the commuting involutive case where the localizer collapses to a finite signature computation, then removes the simplifying assumptions by homotopy and perturbation invariance. If the main theorem is correct, index maps from even K-theory to K0(B) admit a direct and computable description that reduces to the known half-signature formula when B=C.

What carries the argument

The spectral localizer is $L_{\kappa,\rho}(H,D,\phi)=\Phi_\rho\gamma H\Phi_\rho+\kappa\Phi_{2\rho}D\Phi_{2\rho}-(1-\Phi_{2\rho}^4)^{1/2}\gamma$, where $\Phi_\rho=\phi(D/\rho)$ for an even localizing function $\phi$ supported in $[-1,1]$ and equal to $1$ on $[-1/2,1/2]$. Because $\phi$ has compact support, each $\Phi_\rho$ is compact, so the pair $(-\gamma,L_{\kappa,\rho})$ defines a class in the relative K-theory of the pair $(\mathcal L(X),\mathcal K(X))$. The argument's backbone is a reduction: first prove the identity under $H^2=1$ and $[D,H]=0$, where a direct computation identifies the class through the phase of $H$ and the half-signature picture, then use homotopy invariance of the class and invariance under selfadjoint bounded perturbations of $D$ to return to arbitrary Lipschitz $H$. Excision in K-theory and the index-isomorphism picture translate the result into $K_0(B)$.

What would settle it

Take $B=\mathbb{C}$, $G_+=G_-=\mathbb{C}^2$, $D_0=I_2$, and $H=2(Q_+\oplus Q_-)-1$ with $Q_+=I_2$ and $Q_-=\mathrm{diag}(1,0)$; then $Q=Q_+\oplus Q_-$ and $H$ is an even selfadjoint involution. Computing Theorem 10.2 for this explicit 4-by-4 example means calculating the Fredholm index of $(QDQ)_0$ and comparing it with $\tfrac12\operatorname{sign}(P_\rho(\kappa D+\gamma H)P_\rho)+\tfrac12\operatorname{sign}(\gamma P_\rho)$ for large $\rho$. If the two integers differ, the main theorem is false.

Watch

Extended reading notes

Core claim

Let $(X,\pi,D)$ be an even compact unbounded Kasparov module from $A$ to $B$ and let $p\in M_n(\operatorname{Lip}_D(A)^\sim)$ be a projection. For every adjointable isometry $V:X^{\oplus n}\to\ell^2(\mathbb{N},B)$, Theorem 9.1 asserts the identity $$\varphi\,$K_0^{{\mathrm{inv}}$}(\operatorname{Ad}(V))\bigl(L(2\pi(p)-1,$D^{{\oplus n}}$)-L(2\pi(s(p))-1,$D^{{\oplus n}}$)\bigr)=\langle [p]-[s(p)], [X,\pi,F_D]\rangle$$ in $K_0(B)$, where $L(H,D)$ is the spectral-localizer class, $s(p)$ is the scalar part of the unitalization, and $F_D=D(1+D^2)^{-1/2}$ is the bounded transform. The right-hand side is the Kasparov-product index homomorphism applied to the class $[p]-[s(p)]$. The left-hand side uses only continuous functions of $D$ with support in $[-\rho,\rho]$, and the admissibility of $(\kappa,\rho)$ is quantified in terms of the spectral gap of $H$, the norm of the commutator $[D,H]$, and the Fourier norm of the localizing function.

Load-bearing premise

The proof of the main formula requires that a certain diagram commutes: pushing a K-theory class forward by an algebra homomorphism and then pairing must equal pairing first and then pushing the result forward. The paper does not prove this directly in the picture it works in, citing an external theorem instead; if that compatibility failed, identity (9.2) would not follow.

Editorial extensions

If this is right

  • The Kasparov-product index homomorphism $\langle\cdot,[X,\pi,F_D]\rangle\colon K_0(A)\to K_0(B)$ can be evaluated from data supported in the compact interval $[-\rho,\rho]$ of the spectrum of $D$, once $\rho$ exceeds the explicit threshold in Proposition 6.8.
  • The formula applies to arbitrary even KK-classes represented by unbounded modules, including all cases where spectral projections of $D$ are not available as adjointable operators.
  • The class is independent of the choice of localizing function and admissible pair, so the invariant is stable under harmless choices of cutoff.
  • When $B=\mathbb{C}$ and spectral projections exist, Theorem 10.2 recovers the earlier half-signature formula $\mathrm{Index}((QDQ)_0)=\tfrac12\operatorname{sign}(P_\rho(\kappa D+\gamma H)P_\rho)+\tfrac12\operatorname{sign}(\gamma P_\rho)$.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The explicit admissibility bounds suggest a truncation algorithm: for $B=\mathbb{C}$, choosing $\rho$ and $\kappa$ from the quoted norms gives a finite-dimensional selfadjoint matrix whose half-signature is guaranteed to equal the index, without prior knowledge of the Dirac spectrum.
  • The two-step proof strategy of proving a commuting case and then transferring by bounded perturbation may carry over to odd KK-theory or to semifinite spectral triples, although the paper only treats the even module case.
  • One could test the robustness of the formula by truncating $\ell^2(\mathbb{N},B)$ at growing finite rank and checking that the resulting class stabilizes once $\rho$ and $\kappa$ meet the paper's bounds.
  • If the naturality of the index isomorphism were proved directly, the main theorem would no longer rest on an imported equivalence between KK-theory pictures, making the whole chain of computations self-contained.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 3 minor

Summary. The paper develops a spectral localizer for even unbounded Kasparov modules over possibly nonunital C*-algebras and uses it to compute the index homomorphism induced by a KK-class via the Kasparov product. The main result, Theorem 9.1, asserts that for an even compact unbounded Kasparov module (X,π,D) from A to B and a projection p in M_n(Lip_D(A)∼), the difference of spectral localizer classes L(2π(p)−1,D^{⊕n}) − L(2π(s(p))−1,D^{⊕n}), after applying an adjointable isometry to the standard module, equals the Kasparov-product index pairing ⟨[p]−[s(p)], [X,π,F_D]⟩ in K0(B). The proof reduces to a computational core in Section 8 where the spectral localizer class is identified with an index class of a compressed unbounded module, and Section 10 recovers the Loring–Schulz-Baldes even index pairing in the special case B=C. The paper also provides quantitative admissibility conditions on the parameters (κ,ρ) in terms of the spectral gap and commutator norms.

Significance. If the main theorem is correct, this is a substantial extension of the spectral localizer method from K-homology to full KK-theory, giving an explicit, finite-spectrum description of the even index homomorphism in a general Hilbert-C*-module setting. The central computation in Section 8 is detailed and appears internally coherent, and the paper carefully avoids any dependence on spectral projections by using continuous functional calculus. The result is concrete and falsifiable, and the recovery of the Loring–Schulz-Baldes theorem is a welcome check. The main conceptual weakness is the reliance on the naturality of the index isomorphism in Proposition 5.5, whose proof as written is incomplete and which is load-bearing for Theorem 9.1.

major comments (2)
  1. [§5.1, Proposition 5.5] The proof of Proposition 5.5 constructs a unital even Kasparov module (E,F) over C([0,1],C) and asserts that it is a homotopy between the two Kasparov modules in (5.6) and (5.7). For this conclusion the operator F must satisfy F^2−1 ∈ K(E), since the representation of the domain algebra C is unital. The fibers of F are the constant operator S = [[0, σ(ps(p))],[σ(s(p)p), 0]] on E1. For a non-degenerate σ and a nonunital C, taking B = C and p=(0,1) in M_1(B∼) gives σ(p)=1_{C∼} acting on C, so E1 ≅ C ⊕ C and F=0; then F^2−1 = −1 is not a compact operator on the Hilbert C-module C⊕C. Thus the assertion that (E,F) is a Kasparov module is not justified in general. Since equation (9.4) in the proof of Theorem 9.1 relies on Proposition 5.5, the proof of the main theorem is incomplete as it stands. The author should either prove the naturality of the index isomorphism in the stated generality, including non-degenerate σ, or cite a reference that gives a complete proof of this naturality rather than merely citing the equivalence of pictures.
  2. [§5.1, equations (5.6)–(5.7)] The identification of (σ_* Index^{-1})([p]−[s(p)]) with the class of (E0, F0) in (5.7) is stated as 'it can be verified', but this is precisely the non-trivial naturality statement. The subsequent homotopy argument cannot serve as that verification because, as noted above, the intermediate module (E,F) is not generally a Kasparov module. The proof therefore needs a different argument; a correct proof or a precise reference to a full naturality theorem for the index isomorphism is required before Theorem 9.1 can be considered established.
minor comments (3)
  1. [§5, Proposition 5.2] The proof of Proposition 5.2 is omitted with the note that it is elementary. Given that the index class depends on the choice of the adjointable isometry V and on representatives, a brief indication of the homotopy-invariance argument would improve readability and verifiability.
  2. [§8, Lemma 8.2] The proof of continuity of t ↦ ν_t(A)G uses the factorization G=(1−Φ^4)^{1/4}G_0 and the bound (8.1). This is correct but quite compressed; expanding the estimate for ∥ν_t(A)−ν_s(A)∥ in terms of ∥μ_t−μ_s∥ would make the argument easier to follow.
  3. [Throughout] There are a number of typographical errors, e.g., 'continous' in Section 8, 'themapt' before the continuity claim in the proof of Lemma 8.2, and inconsistent spacing in displayed formulas. These do not affect the mathematics but should be corrected.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: the spectral-localizer class is computed, not assumed, and the only load-bearing external input is a standard naturality theorem, not a self-citation.

full rationale

I walked the derivation chain from the definition of the spectral localizer (Definition 6.2) through Theorem 8.4 to the main formula (Theorem 9.1). The class L(H,D) is defined directly from D, H, and a localizing function; it is not defined as the index pairing it later computes. Theorem 8.4 establishes, by a concrete computation in relative K-theory, that K^{inv}_0(Ad(V))(L(H,D)) equals φ^{-1}(Index(H_+X,F_{H_+DH_+})), and Theorem 9.1 composes this with the standard description of the Kasparov-product index homomorphism. No fitted parameter is renamed as a prediction, no ansatz is imported from the author's earlier work as the load-bearing content, and no known result is merely relabelled. The paper does contain an explicitly admitted reliance: in §5.1 the author states, 'we have been unable to find a proof which only uses the operator homotopy picture of KK-theory ... and our proof therefore eventually relies on [JeTh91, Theorem 2.2.17]' for the naturality of the index isomorphism under non-degenerate ∗-homomorphisms. This is an external, standard equivalence-of-pictures theorem rather than a self-citation or a definitional reduction; if it failed, the main formula would not follow, so this is a proof-gap/correctness risk, not circularity. The self-citations that appear (e.g. [Kaa20] in Proposition 7.3 for the fact that D+T remains an unbounded Kasparov module, and [KaLe12] for the Kato-Rellich theorem) support auxiliary perturbation facts and are not the target claim. Overall, the central derivation is self-contained relative to standard KK-theory inputs and receives a low circularity score of 1.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

No free parameters are fitted to data; the constants κ and ρ are chosen by explicit inequalities and the final class is independent of them. The paper relies on a set of standard theorems from KK-theory, unbounded Kasparov theory, and Hilbert C*-module functional calculus. No new particles or entities are introduced.

assumptions (7)
  • standard math Baaj-Julg theorem: every even Kasparov module from separable A to σ-unital B can be represented by an unbounded Kasparov module with compact resolvent.
    Invoked in Section 9 (Theorem 4.4) to start from an unbounded representative.
  • standard math Existence of Lipschitz representatives for K-theory classes [Bos90, Théorème A.2.1].
    Used in the introduction and Theorem 9.1 to choose H = 2π(p)-1 with bounded commutator [D,H].
  • domain assumption Naturality of the index isomorphism (Proposition 5.5), proved via [JeTh91, Theorem 2.2.17] rather than the operator homotopy picture.
    Used in Theorem 9.1 identity (9.3)-(9.4). The author flags the reliance explicitly.
  • standard math Kucerovsky's criterion for unbounded Kasparov products [Kuc97, Theorem 13].
    Used in Proposition 4.6 to identify the easy-case unbounded Kasparov product.
  • standard math Continuous functional calculus for regular selfadjoint operators on Hilbert C*-modules via Cayley transform (Theorem 2.4).
    Foundational for defining Φ_ρ and the spectral localizer; cites Lance, Woronowicz, Kustermans.
  • standard math Excision in K-theory [Ros94, Theorem 1.5.9].
    Used to pass between relative K-theory and K-theory of ideals throughout Section 3.
  • standard math Kato-Rellich theorem for Hilbert C*-modules [KaLe12, Theorem 4.5].
    Used in Proposition 2.13 and 7.3 for bounded perturbations of D.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Spectral localizers in KK-theory." pith.science (2026). https://pith.science/paper/W6JTPYYI

@misc{pith2026250808668,
  author       = {Pith},
  title        = {Pith review of: Spectral localizers in KK-theory},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/W6JTPYYI}},
  note         = {Machine review of arXiv:2508.08668}
}
abstract

We study the index homomorphism of even K-groups arising from a class in even KK-theory via the Kasparov product. Due to the seminal work of Baaj and Julg, under mild conditions on the C^*-algebras in question such a class in KK-theory can always be represented by an unbounded Kasparov module. We then describe the corresponding index homomorphism of even K-groups in terms of spectral localizers. This means that our explicit formula for the index homomorphism does not depend on the full spectrum of the abstract Dirac operator D, but rather on the intersection between this spectrum and a compact interval. The size of this compact interval does however reflect the interplay between the K-theoretic input and the abstract Dirac operator. Since the spectral projections for D are not available in the general context of Hilbert $C^*$-modules we instead rely on certain continuous compactly supported functions applied to D to construct the spectral localizer. In the special case where even KK-theory coincides with even K-homology, our work recovers the pioneering work of Loring and Schulz-Baldes on the index pairing.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

8 extracted references · 7 canonical work pages

  1. [1]

    Baaj , Multiplicateurs non bornés, Publications Mathématiques de l’Université Pierre et Marie Curie (1981), no

    [Baa81] S. Baaj , Multiplicateurs non bornés, Publications Mathématiques de l’Université Pierre et Marie Curie (1981), no. 37, 1–44. [BaJu83] S. Baaj and P. Julg, Théorie bivariante de Kasparov et opérateurs non bornés dans les C ∗-modules hilbertiens, C. R. Acad. Sci. Paris Sér. I Math.296 (1983), no. 21, 875–878. MR 715325 [Bla98] B. Blackadar, K-theory...

    work page 1981

  2. [13]

    Mesland and A

    MR 3982949 [MeRe16] B. Mesland and A. Rennie, Nonunital spectral triples and metric completeness in unbounded KK-theory, J. Funct. Anal.271 (2016), no. 9, 2460–2538. MR 3545223 [Mes14] B. Mesland,Unbounded bivariantK-theory and correspondences in noncommutative geometry, J. Reine Angew. Math.691 (2014), 101–172. MR 3213549 SPECTRAL LOCALIZERS IN KK -THEOR...

    work page 2016

  3. [21]

    Kaad and M

    MR 4137615 [KaLe12] J. Kaad and M. Lesch , A local global principle for regular operators in Hilbert C ∗-modules, J. Funct. Anal.262 (2012), no. 10, 4540–4569. MR 2900477 [KaLe13] , Spectral flow and the unbounded Kasparov product, Adv. Math.248 (2013), 495–530. MR 3107519 [Kas80a] G. G. Kasparov , Hilbert C ∗-modules: theorems of Stinespring and Voicules...

    work page 2012

  4. [719]

    The functional calculus for regular operators in Hilbert C*-modules revisited

    MR 582160 (81m:58075) [Kuc97] D. Kucerovsky , The KK-product of unbounded modules, K-Theory 11 (1997), no. 1, 17–34. MR 1435704 (98k:19007) [Kus97] J. Kustermans, The functional calculus of regular operators on HilbertC ∗-modules revisited, arXiv:funct-an/9706007v1. [Lan95] E. C. Lance , Hilbert C ∗-modules, London Mathematical Society Lecture Note Se- ri...

    work page Pith review arXiv 1997

  5. [1957]

    Higson and J

    MR 423094 [HiRo00] N. Higson and J. Roe, Analytic K-homology, Oxford Mathematical Monographs, Oxford University Press, Oxford, 2000, Oxford Science Publications. MR 1817560 [JeTh91] K. K. Jensen and K. Thomsen, Elements of KK-theory, Mathematics: Theory & Applications, Birkhäuser Boston, Inc., Boston, MA,

    work page 2000

  6. [1991]

    Kaad, On the unbounded picture ofKK-theory, SIGMA Symmetry Integrability Geom

    MR 1124848 (94b:19008) [Kaa20] J. Kaad, On the unbounded picture ofKK-theory, SIGMA Symmetry Integrability Geom. Methods Appl.16 (2020), Paper No. 082,

    work page 2020

  7. [1994]

    Schulz-Baldes and T

    MR 1282290 [SBSt21] H. Schulz-Baldes and T. Stoiber , The spectral localizer for semifinite spectral triples, Proc. Amer. Math. Soc.149 (2021), no. 1, 121–134. MR 4172591 [Sui24a] W. D. V. Suijlekom , A generalization of K-theory to operator systems , arXiv:2409.02773 [math.OA] . [Sui24b] , Higher K-groups for operator systems, arXiv:2411.02981 [math.OA] ...

    arXiv 2021

  8. [1998]

    Detecting local topology with the spectral localizer

    MR 1656031 (99g:46104) [BlCu91] B. Blackadar and J. Cuntz, Differential Banach algebra norms and smooth sub- algebras of C ∗-algebras, J. Operator Theory26 (1991), no. 2, 255–282. MR 1225517 (94f:46094) [BoLo16] J. L. Boersema and T. A. Loring , K-theory for real C ∗-algebras via unitary elements with symmetries, New York J. Math.22 (2016), 1139–1220. MR ...

    work page Pith review arXiv 1991

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.