On the invariance of the string topology coproduct

Nathalie Wahl

REVIEW 3 major objections 3 minor 23 references

On the invariance of the string topology coproduct

T0 review · 3 major / 3 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read For closed oriented manifolds, the Goresky-Hingston string topology coproduct commutes with the map induced by a homotopy equivalence up to the sign deg(f), making the coproduct a homotopy-type invariant.

desk verdict Important claim and a serious proof attempt, but a load-bearing direction error in the lift Ŵh leaves the main theorem unproven as written. read the letter →

arxiv 1908.03857 v5 pith:RV4Y27K7 submitted 2019-08-11 math.AT math.DGmath.GT

Nathalie Wahl This is my paper · ORCID

classification math.ATmath.DGmath.GT MSC 55P3555P50

keywords stringtopologyGoresky-HingstoncoproducthomotopyinvariancefreeloopspaceChas-SullivanproductThom-Pontryaginconstructiontubularneighborhoodsintersection

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 Goresky-Hingston coproduct in string topology—an operation that cuts a loop at a self-intersection at its basepoint—is homotopy invariant, just as the Chas-Sullivan product (concatenation of loops at a common basepoint) was already known to be. A homotopy equivalence $f$ between closed oriented manifolds intertwines the coproduct on the relative loop-space homology with the coproduct on the target, up to the sign of the degree of $f$; the same holds for the extension-by-zero coproduct and for the associated cohomology products. This matters because the coproduct detects how many times loops in a chain must self-intersect at their basepoint, and the paper shows that any quantity it detects depends only on the homotopy type of the manifold. The proof lifts to the free loop space the Thom-Pontryagin construction of the intersection product, carrying the boundary terms that make the coproduct a relative operation.

What carries the argument

The central device is a lift to loop space of the Thom-Pontryagin definition of the intersection product. The authors embed $M_1$ into $M_2\times D^k$, form two tubular neighborhoods of the composed embedding with isomorphic normal bundles, and use the uniqueness of tubular neighborhoods to obtain a diffeomorphism $h$ of $M_1\times M_2\times D^k$; this $h$ is then lifted to a map $\widehat{h}$ of the loop-space product $\Lambda_{1,2}\times I\times D^k$ that identifies the two pulled-back neighborhoods while preserving the self-intersecting subspace. The retractions $R_i$, which add a geodesic stick to make a loop meet its basepoint at time $s$, convert the geometric comparison into the commutativity of a twelve-piece diagram (2.1); the sign $\deg(f)$ enters through the comparison of Thom classes.

What would settle it

Compute both sides of the coproduct invariance diagram for an explicit degree-one homotopy equivalence between closed oriented manifolds, for instance a self-map of a high-dimensional torus that is a homotopy equivalence but not a diffeomorphism, and check whether the coproduct of a chosen homology class changes by exactly $\deg(f)$; any other factor would refute Theorem A.

Watch

Extended reading notes

Core claim

Theorem A of the paper asserts that for $f\colon M_1\to M_2$ a homotopy equivalence between closed oriented manifolds, the diagram for the relative coproduct $\vee$ on $H_*(\Lambda M_i, M_i)$ commutes with $\Lambda f$ up to the sign $\deg(f)$, and the same is true for the extended coproduct $\widehat{\vee}$ on $H_*(\Lambda M_i)$. Consequently the cohomology products $\circledast$ and $\widehat{\circledast}$ are respected by the induced map in cohomology up to the same sign. In particular, any quantity measured by these operations depends only on the homotopy type of the manifold; the coproduct detects the largest number of simultaneous basepoint self-intersections that must appear in any chain representing a given loop-space homology class, and that number is a homotopy invariant.

Load-bearing premise

The proof assumes that the lifted diffeomorphism $\widehat{h}$ can be chosen to fix pointwise the boundary of the disc and the self-intersecting loops; the commutativity of the central diagram that carries the invariance argument depends on that fixed-point property.

Editorial extensions

If this is right

Any invariant derived from the Goresky-Hingston coproduct, such as basepoint self-intersection multiplicity, becomes a homotopy-type invariant of closed oriented manifolds.
The extension-by-zero coproduct on the full loop space homology is homotopy invariant, so the statement covers classes that are not visible in the relative homology of loops versus constant loops.
The cohomology products $\circledast$ and $\widehat{\circledast}$ are homotopy invariant as well, transferring the result to the cohomological side through the universal coefficient theorem.
The proof demonstrates that compactified string operations—not just the Chas-Sullivan product—can be homotopy invariant, though the authors do not claim their arguments extend to all such operations.

Reading between the lines

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

If Theorem A is correct, any purported formula describing a failure of the coproduct to commute under homotopy equivalences must have an obstruction that vanishes identically for homotopy equivalences; verifying that vanishing is a direct test of the paper's claim.
The sign factor $\deg(f)$ parallels the behaviour of the classical intersection product under degree-$d$ maps, suggesting a potential analogue: the coproduct may be invariant up to a scaling factor for maps of degree $d$, not just $\pm 1$, if the proof's hypotheses are relaxed.
The method may extend to other compactified string operations via the harmonic compactification of moduli space; the paper leaves that as an open question, so testing it on the next simplest operation would be a natural next step.

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.

Desk Editor's Note

read the letter

The claim is the right one and the paper is serious: Hingston and Wahl assert homotopy invariance of the Goresky-Hingston string topology coproduct, a natural next step after invariance of the Chas-Sullivan product. The overall strategy, lifting the Thom-Pontryagin proof of intersection-product invariance to loop space with the Gruher-Salvatore intermediate spaces, is sensible. Section 1's warm-up is clean and genuinely useful, and the construction of the intermediate spaces W^1_{1,2} and Λ_{1,2} is well thought out. Many subdiagrams (1)–(7) and (9)–(12) are checked carefully. The paper is self-contained modulo [15], and the self-citation there is for definitions, not for the target result.

The proof has a load-bearing gap in Proposition 3.5. Proposition 1.4(3) says h∘¯ν₃ = ν₃∘φ, so h maps ¯X₃ onto X₃. Proposition 3.5(1) claims the lift Ŵh satisfies Ŵh(X₃)⊂¯X₃ and Ŵh(X₃^c)⊂¯X₃^c. That is the reverse direction; the intertwining relation âe_I∘Ŵh = h∘âe_I actually gives Ŵh(¯X₃)⊂X₃. The assertion as written contradicts the authors' own Proposition 1.4. Lemma 3.6 needs a map H_*(X₃)→H_*(¯X₃), and Diagram (2.1) uses that arrow; with the direction reversed, the central diagram does not commute as written. This is not a cosmetic typo. The likely fix is to lift h^{-1} instead of h, and to recheck Thom class compatibility, so the gap is probably repairable, but it is real.

The reader's separate objection to Prop 3.5(3) is also correct: the formula s' = max(|x−e(m)|, s) moves the time parameter for points with x∈∂D^k, so Ŵh is not homotopic to the identity relative to that subspace. The weaker property of being homotopic through maps preserving Λ_{1,2}×I×∂D^k ∪ ¯B may suffice for Lemma 3.6, so this is less decisive than the direction error, but the statement as written is false.

Verdict: major revision, not accept. The claim is likely true and much of the machinery is sound, but a serious referee should demand a corrected Proposition 3.5 and a re-verified Diagram (2.1). Given the importance of the claimed result, I would send it to peer review rather than desk reject, but the current version should not be accepted.

Referee Report

3 major / 3 minor

Summary. The paper claims Theorem A: if f:M1→M2 is a homotopy equivalence between closed oriented manifolds, then the induced map Λf commutes with the Goresky–Hingston string topology coproducts ∨ and ∨̂, and with the associated cohomology products, up to the sign deg(f). The proof adapts the Gruher–Salvatore proof of homotopy invariance of the Chas–Sullivan product to the coproduct, working with the Thom–Pontryagin description of the relevant operations. Section 1 gives a warm-up proof of homotopy invariance of the intersection product using composed tubular neighborhoods. Section 2 sets up the key diagram (2.1), whose commutativity is analyzed in twelve subdiagrams in Section 3. The central technical object is a lift ĥ of the diffeomorphism h of Proposition 1.4, which is supposed to interchange the two tubular neighborhoods X3 and ̄X3 of the relevant embeddings. Appendix A proves the composition properties of tubular embeddings used throughout. The paper is self-contained modulo standard differential topology and the authors' previous paper [15] for the definition of the coproduct.

Significance. If the central claim is correct, the paper settles a natural open problem: it establishes that a compactified string topology operation, the Goresky–Hingston coproduct, is a homotopy-type invariant, and it does so with an explicit chain-level diagram. The strategy of lifting the Thom–Pontryagin proof for the intersection product is attractive and is presented carefully, with all subdiagrams itemized and an appendix devoted to the technical composition lemma. The paper also makes a useful contribution by clarifying which parts of the invariance statement are formal and which require invertibility of specific maps. However, the proof as written has a load-bearing gap in the construction of the map ĥ in Section 3.6, and the announced theorem is not established by the submitted text.

major comments (3)

[§3.6, Proposition 3.5(1)] The direction of the lift ĥ is wrong relative to Proposition 1.4(3). Proposition 1.4(3) states that h∘̅ν3 = ν3∘φ, hence h(̅ν3̅N3) = ν3N3 and h^{-1}(ν3N3) = ̅ν3̅N3. Since ̅eI∘ĥ = h∘̅eI, for x∈̅X3 we get ̅eI(ĥ(x)) = h(̅eI(x)) ∈ h(̅ν3̅N3) = ν3N3, so ĥ(̅X3)⊂X3. The asserted inclusions ĥ(X3)⊂̅X3 and ĥ(X3^c)⊂̅X3^c do not follow and are generally false. Lemma 3.6 requires a map H*(X3,̅B1,2)→H*(̅X3,̅B1,2), and Diagram (2.1) uses exactly that arrow. As written, subdiagram (8) is not proved, and the outer commutativity of the central diagram is not established. This error is correctable: defining ĥ via a lift of h^{-1} instead of h gives ĥ(X3)⊂̅X3, and the required fixed-point properties for h^{-1} are available from Proposition 1.4 because h is isotopic to the identity relative to the 0-section and boundary. The gap is genuine but local.
[§3.6, Proposition 3.5(3)] The claimed relative homotopy to the identity is false as stated. The definition of ĥ changes the time parameter through s′ = max(|x−e(m)|,s) for s≤1/2. For a point in Λ1,2×I×∂D^k with |x−e(m)|>s, the third coordinate x is fixed by h (which fixes M1×M2×∂D^k), but s′ differs from s, so ĥ does not fix Λ1,2×I×∂D^k pointwise. Consequently ĥ is not homotopic to the identity relative to Λ1,2×I×∂D^k ∪ ê(F1,2), as asserted. Lemma 3.6 and Proposition 3.11 invoke exactly this property. It is possible that a map-of-pairs homotopy is sufficient for the naturality argument in Lemma 3.6, but the relative statement in the manuscript is incorrect and must be repaired or weakened explicitly.
[§3.9, proof of Theorem A] The proof of Theorem A depends on commutativity of the whole Diagram (2.1). The two issues above concern subdiagram (8), which is the one that connects the two different tubular neighborhoods X3 and ̅X3. Since the direction of ĥ is wrong, the claim 'by Sections 3.1–3.8, we know that each subdiagram in Diagram (2.1) commutes' is not supported. This is a load-bearing point, not a stylistic detail: the outside square of Diagram (2.1) is the heart of the homotopy-invariance proof. The error is repairable within the manuscript's scope, but the present text does not contain a correct proof of the central diagram.

minor comments (3)

[Section 1, after Proposition 1.7] The sentence 'We repeat here that the is not the optimal result' is missing a word; it should read 'this is not the optimal result' or similar.
[Lemma 3.6] The source in the displayed homology group is written as H_{*+k}(Λ1,2×I×D^k, Λ1,2×I×D^k∪̅B1,2); this should be Λ1,2×I×∂D^k ∪ ̅B1,2, as in the surrounding text.
[Proposition 3.5] The phrase 'for h the diffeomorphism of Proposition 1.4(2)' is imprecise: h is the diffeomorphism constructed in the statement and proof of Proposition 1.4, not item (2) of that proposition.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the proof of Theorem A is self-contained and does not reduce to its inputs.

full rationale

The derivation of Theorem A is self-contained modulo standard differential and algebraic topology and the authors' earlier paper [15], which is used only for the definition of the coproduct and for the retraction and concatenation conventions. The target statement—that a homotopy equivalence f induces a map Λf commuting with ∨ and ˆ∨ up to the sign deg(f)—is not assumed anywhere in the proof; it is obtained by proving commutativity of Diagram (2.1) through the twelve subdiagrams in Section 3. Citations to [15], for example 'As in [15], we will denote by ˆ∨ ...' and 'We start by recalling the definition of the string coproduct, as given in [15]', supply definitions and notation rather than the theorem. The uniqueness theorem for tubular neighborhoods used in Proposition 1.4 is an external standard result, and Section 3.7 uses the homotopy-equivalence hypothesis to prove map-level homotopy equivalences, not to assume the desired coproduct commutation. There are no fitted parameters, no quantities relabelled as predictions, and no conclusion-by-citation chain. The skeptical objections about Proposition 3.5 concern whether the written proof establishes the required commutativity of subdiagram (8), including the direction of the lifted map and the relative fixed-point property; if valid, these are technical proof gaps or correctable errors, not circular dependencies. The abstract passage preceding the full text, mentioning a Naef-formula obstruction, is inconsistent with the full text's invariance theorem, but inconsistency is not circularity. Overall, no circular step is exhibited, so the score is 0.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No free parameters or invented entities. The proof relies on standard differential topology (tubular neighborhoods, Thom classes, Vogt's lemma) and on definitions from the authors' prior preprint [15]. The central claim is not a fit to data; it is a proposed theorem.

assumptions (5)

standard math Uniqueness of tubular neighborhoods (Hirsch [16])
Used in Proposition 1.4 to identify the two tubular neighborhoods of the composed embedding and to obtain the diffeomorphism h.
standard math Vogt's lemma [22]
Used in Proposition 3.8 to choose homotopies h1, h2 and K with the compatibility f∘h1 ≃ h2∘(f×1).
standard math Homology suspension theorem
Used in Section 1 and in Section 3.4 to show [1×η∩] is an isomorphism.
domain assumption Definitions of the coproduct, optimal concatenation and retraction maps from [15] (Hingston-Wahl, arXiv:1709.06839)
The paper's central claim concerns these specific operations; the proof imports the construction and properties from the authors' prior preprint.
domain assumption Closed oriented manifold admits a Riemannian metric with injectivity radius ρ
Used throughout to define tubular neighborhoods and geodesic sticks.

how reviews work

0 comments

Cite this review

Pith. "Pith review of On the invariance of the string topology coproduct." pith.science (2026). https://pith.science/paper/RV4Y27K7

@misc{pith2026190803857,
  author       = {Pith},
  title        = {Pith review of: On the invariance of the string topology coproduct},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RV4Y27K7}},
  note         = {Machine review of arXiv:1908.03857}
}

read the original abstract

We give a variant of Naef's formula for the failure of invariance of the string topology coproduct under homotopy equivalences, using an obstruction class built from homotopy data associated to a homotopy equivalence as well as the ``fake diagonal''. The vanishing of our obstruction class can be seen as a way to measure a form of boundedness for homotopy equivalences. We show that the same obstruction rules the failure of invariance for a generalization of the coproduct to higher dimensional loops.

Figures

Figures reproduced from arXiv: 1908.03857 by the authors.

**Figure 2.** For (m, m0 , γ, s) ∈ W11,2, the figure illustrates the 2nd component of L 0 (m, m0 , γ, s, −), a homotopy between γ and f ◦ g ◦ γ through loops originating in W11,2, given at time t by K(m, t, −) ? h2(γ[0, s], t) ? K(m0 , t, −) −1 ∗s 0 K(m0 , t,−)?h2(γ[s,1], t)?K(m, t,−) −1 . and s 0 (t) = s + (1 2 − s) min(t, 1 − t)ε. The time s 0 = s 0 (t) has the property that s 0 (t) = s if t = 0 or t = 1, allowing us to recover… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

23 extracted references · 21 canonical work pages

[15]

Product and coproduct in string topology

Nancy Hingston and Nathalie Wahl. Product and coproduct in string topology. Preprint 2017 (arXiv:1709.06839)

arXiv 2017
[1]

On algebraic structures of the Hochschild complex

Hossein Abbaspour. On algebraic structures of the Hochschild complex. In Free loop spaces in geometry and topology , volume 24 of IRMA Lect. Math. Theor. Phys. , pages 165–222. Eur. Math. Soc., Z¨ urich, 2015

work page 2015
[2]

The goresky-hingston product via higher hochschild homology

Kaj B¨ orjeson. The goresky-hingston product via higher hochschild homology. In preparation

work page
[3]

Glen E. Bredon. Topology and geometry, volume 139 of Graduate Texts in Mathematics . Springer-Verlag, New York, 1993

work page 1993
[4]

Chas and D

M. Chas and D. Sullivan. String topology. Preprint 1999 (arXiv:math.GT/9911159)

arXiv 1999
[5]

Knot contact homology, string topology, and the cord algebra

Kai Cieliebak, Tobias Ekholm, Janko Latschev, and Lenhard Ng. Knot contact homology, string topology, and the cord algebra. J. ´Ec. polytech. Math., 4:661–780, 2017

work page 2017
[6]

Cohen and John D

Ralph L. Cohen and John D. S. Jones. A homotopy theoretic realization of string topology. Math. Ann., 324(4):773–798, 2002

work page 2002
[7]

Cohen, John R

Ralph L. Cohen, John R. Klein, and Dennis Sullivan. The homotopy invariance of the string topology loop product and string bracket. J. Topol., 1(2):391–408, 2008

work page 2008

Show all 23 references

[8]

M. C. Crabb. Loop homology as ﬁbrewise homology. Proc. Edinb. Math. Soc. (2) , 51(1):27–44, 2008

2008
[9]

Drummond-Cole, Kate Poirier, and Nathaniel Rounds

Gabriel C. Drummond-Cole, Kate Poirier, and Nathaniel Rounds. Chain level string topology operations. Preprint 2015 (arXiv:1506.02596)

2015 arXiv
[10]

Rational string topology.J

Yves F´ elix, Jean-Claude Thomas, and Micheline Vigu´ e-Poirrier. Rational string topology.J. Eur. Math. Soc. (JEMS) , 9(1):123–156, 2007

2007
[11]

Loop products and closed geodesics

Mark Goresky and Nancy Hingston. Loop products and closed geodesics. Duke Math. J. , 150(1):117–209, 2009

2009
[12]

Generalized string topology operations

Kate Gruher and Paolo Salvatore. Generalized string topology operations. Proc. Lond. Math. Soc. (3) , 96(1):78–106, 2008

2008
[13]

Loop products on connected sums of projective spaces

Nancy Hingston. Loop products on connected sums of projective spaces. In A celebration of the mathematical legacy of Raoul Bott, volume 50 of CRM Proc. Lecture Notes, pages 161–175. Amer. Math. Soc., Providence, RI, 2010. 24 NANCY HINGSTON AND NATHALIE WAHL

2010
[14]

Compactiﬁed string topology

Nancy Hingston and Nathalie Wahl. Compactiﬁed string topology. In preparation
[16]

Morris W. Hirsch. Diﬀerential topology, volume 33 of Graduate Texts in Mathematics. Springer-Verlag, New York, 1994. Corrected reprint of the 1976 original

1994
[17]

John D. S. Jones. Cyclic homology and equivariant homology. Invent. Math., 87(2):403–423, 1987

1987
[18]

Natural operations on the Hochschild complex of commutative Frobenius algebras via the complex of looped diagrams

Angela Klamt. Natural operations on the Hochschild complex of commutative Frobenius algebras via the complex of looped diagrams. Preprint 2013 (arXiv:1309.4997)

2013 arXiv
[19]

String topology and conﬁguration spaces of two points

Florian Naef and Thomas Willwacher. String topology and conﬁguration spaces of two points. In preparation
[20]

Open and closed string ﬁeld theory interpreted in classical algebraic topology

Dennis Sullivan. Open and closed string ﬁeld theory interpreted in classical algebraic topology. In Topology, geometry and quantum ﬁeld theory , volume 308 of London Math. Soc. Lecture Note Ser. , pages 344–357. Cambridge Univ. Press, Cambridge, 2004

2004
[21]

Loop coproducts in string topology and triviality of higher genus TQFT operations

Hirotaka Tamanoi. Loop coproducts in string topology and triviality of higher genus TQFT operations. J. Pure Appl. Algebra, 214(5):605–615, 2010

2010
[22]

R. M. Vogt. A note on homotopy equivalences. Proc. Amer. Math. Soc. , 32:627–629, 1972

1972
[23]

Hochschild homology of structured algebras

Nathalie Wahl and Craig Westerland. Hochschild homology of structured algebras. Adv. Math., 288:240–307, 2016

2016

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [15]

Product and coproduct in string topology

Nancy Hingston and Nathalie Wahl. Product and coproduct in string topology. Preprint 2017 (arXiv:1709.06839)

arXiv 2017

[2] [1]

On algebraic structures of the Hochschild complex

Hossein Abbaspour. On algebraic structures of the Hochschild complex. In Free loop spaces in geometry and topology , volume 24 of IRMA Lect. Math. Theor. Phys. , pages 165–222. Eur. Math. Soc., Z¨ urich, 2015

work page 2015

[3] [2]

The goresky-hingston product via higher hochschild homology

Kaj B¨ orjeson. The goresky-hingston product via higher hochschild homology. In preparation

work page

[4] [3]

Glen E. Bredon. Topology and geometry, volume 139 of Graduate Texts in Mathematics . Springer-Verlag, New York, 1993

work page 1993

[5] [4]

Chas and D

M. Chas and D. Sullivan. String topology. Preprint 1999 (arXiv:math.GT/9911159)

arXiv 1999

[6] [5]

Knot contact homology, string topology, and the cord algebra

Kai Cieliebak, Tobias Ekholm, Janko Latschev, and Lenhard Ng. Knot contact homology, string topology, and the cord algebra. J. ´Ec. polytech. Math., 4:661–780, 2017

work page 2017

[7] [6]

Cohen and John D

Ralph L. Cohen and John D. S. Jones. A homotopy theoretic realization of string topology. Math. Ann., 324(4):773–798, 2002

work page 2002

[8] [7]

Cohen, John R

Ralph L. Cohen, John R. Klein, and Dennis Sullivan. The homotopy invariance of the string topology loop product and string bracket. J. Topol., 1(2):391–408, 2008

work page 2008