On the spectral flow theorem of Robbin-Salamon for finite intervals

Joa Weber; Urs Frauenfelder

REVIEW 3 major objections 4 minor 17 references

On the spectral flow theorem of Robbin-Salamon for finite intervals

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · 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 on finite intervals, adding spectral boundary conditions turns the operator $\partial_s + A(s)$ into a Fredholm operator whose index is the spectral flow $\varsigma(A)$ of the path.

desk verdict The finite-interval spectral flow definition is internally inconsistent, so the main index formula in Theorem A is not well-defined for the very paths it needs to cover; the Fredholm machinery around it is substantial but the central result as written collapses. read the letter →

arxiv 2412.16332 v1 pith:77I5EAUC submitted 2024-12-20 math.SG math-phmath.APmath.DGmath.DSmath.MP

Urs Frauenfelder , Joa Weber This is my paper · ORCID

classification math.SGmath-phmath.APmath.DGmath.DSmath.MP MSC 58J3047A5346B70

keywords spectralflowFredholmindexHessianpathsinterpolationspacesboundaryconditionsfiniteintervaloperatorsRobbin-Salamontheoremsymplecticgeometry

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

On a finite interval $[-T,T]$, the operator $D_A\xi=\partial_s\xi+A(s)\xi$ has infinite kernel and is not Fredholm unless boundary conditions are added; here $A(s)$ is a continuous path of symmetrizable Fredholm operators of index zero, called Hessians. The paper shows that imposing $\pi_+^{A(-T)}\xi(-T)=0$ and $\pi_-^{A(T)}\xi(T)=0$, where $\pi_\pm$ are spectral projections in the interpolation space $H_{1/2}$, makes the augmented operator Fredholm. The Fredholm index equals the spectral flow $\varsigma(A)$, the net number of eigenvalues crossing zero along the path. The same formula is proved for half-infinite intervals and the real line, completing the classical spectral flow theorem for all four interval types under continuous paths.

What carries the argument

The paper's central object is the augmented operator $\mathcal{D}_A$ together with the spectral flow $\varsigma(A)$. The proof rests on five ingredients: (1) a Rabier-type semi-Fredholm estimate for $D_A$ that needs no weak derivative; (2) the constant-invertible case, where $\mathcal{D}_A$ is an isomorphism; (3) an index-difference formula in terms of spectral content $\rho_A(\lambda,\mu)$, the number of eigenvalues of $A$ between two resolvent values; (4) a path-concatenation theorem, asserting that index and spectral flow are additive when the path is cut at an invertible time; and (5) the imported continuity of spectral projections $\pi_\pm(A)$ as $A$ varies, used to deform a general path to a constant invertible path without changing the index.

What would settle it

Restrict to the model Hilbert-space pair $(\ell^2,\ell^2_f)$ and take $A(s)$ diagonal with two prescribed eigenvalue paths $a_1(s), a_2(s)$, one of which crosses zero with multiplicity two. In the finite-dimensional truncation, compute the augmented operator's index by solving the linear ODE with the stated boundary projections and compare the answer to the eigenvalue-crossing count $\varsigma(A)$; a mismatch would refute Theorem A, and agreement in such examples would test the resolvent-shift mechanism at non-simple crossings.

Watch

Extended reading notes

Core claim

The central claim, Theorem A, is that for every Hessian path $A\in\mathcal{A}^*_I$ on a finite, half-infinite, or real interval, the augmented operator $\mathcal{D}_A\colon P_1(I)\to P_0(I)\times H_{1/2}^+(A(-T))\times H_{1/2}^-(A(T))$ given by $\xi\mapsto(\partial_s\xi+A(s)\xi,\ \pi_+^{A(-T)}\xi(-T),\ \pi_-^{A(T)}\xi(T))$ is Fredholm and satisfies $\operatorname{index}\mathcal{D}_A=\varsigma(A)$. For finite intervals the new content is the boundary conditions; without them the kernel is infinite-dimensional. For half-lines one boundary projection suffices, and on the real line no boundary condition is needed, recovering the classical theorem. The proof is constructive: it decomposes the path into sub-paths made invertible by subtracting resolvent values $\lambda_j$, proves the index is zero on each invertible sub-path, and uses concatenation and the spectral content $\rho_A(\lambda,\mu)$ to add up the eigenvalue crossings.

Load-bearing premise

The proof depends on a result imported from a companion preprint rather than proved here: the spectral projection of an invertible operator onto its positive eigenspace varies continuously as the operator varies. If that continuity fails, the deformation argument that equates the index with the spectral flow collapses.

Editorial extensions

If this is right

Any two finite-interval Hessian paths with the same spectral flow are connected through Fredholm operators of equal index, so spectral flow is a complete homotopy invariant for this boundary-value problem.
The finite-interval result extends to half-infinite intervals and to the real line within the same framework, giving a uniform statement for continuous paths without differentiability assumptions.
Because the proof replaces transversality by resolvent shifts, non-simple crossings of eigenvalues at zero are allowed; no infinite-dimensional perturbation theory is needed.
The identity $\operatorname{index}\mathcal{D}_A = -\operatorname{index}\mathcal{D}_{-A^*}$ follows immediately from $\varsigma(A)=-\varsigma(-A^*)$.
Cutting a path at an invertible time gives additive indices, so the index can be computed interval by interval, which is exactly what the Floer-theoretic gluing constructions require.

Reading between the lines

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

If the imported continuity of spectral projections holds, the same finite-interval argument could be adapted to Banach-space settings or to higher-order operators where transversality is harder, since the proof's structure is purely analytic.
A practical by-product is a numerical recipe for the index: approximate a path by invertible sub-paths, count eigenvalues between successive resolvent shifts at each breakpoint, and sum; this bypasses solving the boundary-value problem directly.
The theorem does not address paths whose endpoints are non-invertible; testing whether the boundary projections can be defined by a limiting procedure would clarify the stability of the index formula at the boundary of $\mathcal{A}^*_{I_T}$.

Share X Bluesky LinkedIn Reddit HN

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

You should know two things before reading this paper. First, the Fredholm theory for the augmented operator DA with interpolation boundary conditions is done carefully and in detail — the estimates in §4.2.1, the cokernel identification in Proposition 4.13, and the concatenation theorem 4.17 are real work, and the resolvent-shift trick to avoid Robbin–Salamon's transversality is clever. Second, the definition of spectral flow on finite intervals is broken, and it breaks the main theorem.

The problem is Definition 3.1. The paper fixes a zero branch a_0(s) ≡ 0 and requires continuous ordered branches with a_{−1}(s) ≤ 0 ≤ a_1(s). If an eigenvalue crosses zero from negative to positive, the continuous branch that started as a_{−1}(−T) < 0 must become positive after the crossing, which contradicts a_{−1}(s) ≤ a_0(s) = 0. The only way to keep the ordering is to set that branch equal to 0 for all s after the crossing, which duplicates the artificial zero and violates condition (ii) unless 0 is actually in the spectrum there. The paper's own Normalization example in Lemma 3.2 shows exactly this: it claims “since arctan(T) > 0, we have 0 = a_{−1}(T)”, while simultaneously declaring a_0(s) ≡ 0. That is not a minor typo — the literal definition forces ς(A) = 0 for every path, since only the fixed branch a_0 hits zero. So the right-hand side of Theorem A is not defined for paths with crossings, which are precisely the cases the theorem is supposed to handle.

This is more fundamental than the reliance on [FW24, Thm. D] from a viXra preprint, which the reader flagged. That reliance is real — the continuity of spectral projections is load-bearing in the homotopy step — but it is secondary. The eigenvalue bookkeeping in §4.2.7 is built on the flawed ς, so the index formula as stated does not survive.

What survives is the Fredholmness of DA, the bijectivity for constant invertible paths, and the concatenation index additivity. Those are useful pieces, and the approach to the real-line theorem via finite intervals is promising once the spectral flow is defined correctly (e.g., by letting the zero label move, or using a crossing form).

This paper is for specialists in Floer theory and spectral flow. The methods deserve attention, but the central theorem as written is not correct. I would send it to a serious referee, because the Fredholm analysis is substantive and the spectral flow defect may be repairable, but the referee must be told to focus on Definition 3.1 and the Normalization example. In current form, I would not cite the main theorem.

Referee Report

3 major / 4 minor

Summary. The paper studies operators of the form D_A = ∂_s + A(s) on finite intervals, half-lines, and the real line, where s ↦ A(s) is a continuous path of symmetrizable Fredholm operators of index zero (called Hessians) between a Hilbert space pair (H_0,H_1). On a finite interval, the unconstrained operator is not Fredholm, so the authors introduce an augmented operator D_A whose target includes boundary components given by positive/negative spectral projections of the endpoint Hessians. The main result, Theorem A, asserts that D_A is Fredholm and that its index equals the spectral flow ς(A) of the path A. For the real line this recovers the Robbin-Salamon theorem, but the finite-interval case is presented as the new content. The proof proceeds through a Rabier-type estimate, a finite-interval estimate, a cokernel identification with the kernel of the adjoint operator, a path-concatenation theorem, and a telescoping eigenvalue count. The paper is carefully structured and the finite-interval Fredholm theory is developed in detail, but the index formula relies on a continuity theorem for spectral projections imported from another paper by the same authors.

Significance. If the central theorem is correct, it gives a spectral-flow formula for finite-interval operators of the form ∂_s + A(s) under only continuity of A, without the differentiability assumption of Robbin-Salamon, and with a proof that avoids infinite-dimensional transversality theory. The approach via interpolation spaces and spectral projections is potentially useful for Floer-theoretic applications, and the finite-interval concatenation argument is an attractive way to reduce the real-line theorem to a local index bookkeeping. The paper is also explicit about the mathematical structures it needs: the cokernel identification in Proposition 4.13, the concatenation theorem 4.17, and the spectral-content computation in Lemma 4.16 are all substantial and mostly self-contained. However, a load-bearing ingredient, the continuity of the spectral projections π_±^A in A, is imported as Theorem 4.20 from [FW24], a viXra preprint by the same authors, and no proof is supplied here. This prevents the paper from being fully verifiable as it stands.

major comments (3)

[Definition 3.1 and Lemma 3.2] The definition of the finite-interval spectral flow is ambiguous. Conditions (i)-(ii) and the phrase 'inclusion of the zero function' can be read as requiring a_0(s) ≡ 0. Under that reading, no path in which an eigenvalue crosses from negative to positive can admit the required continuous ordered branches: the branch starting at a_{-1}(-T) < 0 would have to become positive while remaining ≤ a_0(s) = 0. The proof of the Normalization property in Lemma 3.2 is inconsistent with that reading, since it concludes a_{-1}(T) = 0 while the endpoint spectrum {arctan(T), 0} forces a_0(T) = arctan(T) > 0. The intended object appears to be the ordered branches of the augmented multiset spec(A(s)) ∪ {0}, where a_0 is the branch whose value at -T is the inserted 0 but which is free to move at later times. If that is the intended meaning, Definition 3.1 must state it explicitly, specify the multiplicity convention for the added 0, and justify existence and uniqueness of the ordered branches. This issue is load-bearing because ς(A) is the right-hand side of Theorem A.
[§4.2.7, Step 1 and Theorem 4.20] The proof of the index formula for paths consisting of invertible operators deforms a general invertible path to the constant path A(0) and uses continuity of the projections r ↦ π_±^{A(±rT)} in L(H_{1/2}) to apply Theorem D.1. This continuity is imported from [FW24, Thm. D], stated here as Theorem 4.20, but it is not proved in the paper and [FW24] is a viXra preprint by the same authors. This is a load-bearing step both for the finite-interval formula and for the later half-line and real-line arguments that cite the same theorem. Please either include a full proof of Theorem 4.20 in an appendix or replace it with a peer-reviewed reference. In addition, Theorem 4.20 is stated for the class L^*_{sym0}(H_1,H_0), while Step 1 applies it to symmetrizable operators in A^*_{I_T}; the paper should explain why the stated hypotheses cover the actual application.
[§4.2.7, Step 2 and Step 3] There is an indexing inconsistency in the final index computation. Step 2 defines shifts λ_0, ..., λ_N and asserts that A(s) - λ_j ι is invertible on [t_j, t_{j+1}] for j = 0, ..., N, with t_{N+1} = T. Step 3 and equation (4.67), however, use λ_{N+1}, and the telescoping argument in (4.70) ends with ν↑(t_{N+1}; λ_{N+1}). As written, λ_{N+1} is undefined. The final interval [t_N, T] is covered by A(s) - λ_N ι being invertible, so either the concatenated operator should be D^{λ_0,λ_N}_A or the list of shifts must be extended and the condition for j = N restated. This is a notational error in a central computation and should be repaired.

minor comments (4)

[Definition 1.4] The third displayed line defines A^*_R using A_{I_-}; this should be A_R, since I_- is the half-line used for A^*_{I_-}.
[Definition 1.5 and Section 2.2] The notation H_{1/2}(A) is potentially confusing: the interpolation space H_{1/2} depends only on the fixed Hilbert space pair (H_0,H_1), while H^±_{1/2}(A) depend on A through the spectral projections. Please clarify this distinction explicitly.
[§4.2.3, proof of Step 3] The target space is written as W(I_T; A_t, A_T), which appears to be a typo for W(I_T; A_σ, A_T); please correct.
[§4.3.6 and §4.5] The concatenation formulas write ς(D_A|_{[0,T]}) and ς(D_A|_{[-T,T]}), but the argument of ς should be the path A restricted to the interval, not the operator D_A. Please adjust the notation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the finite-interval index formula is derived from direct eigenvalue bookkeeping, and the only self-citation used in the proof, [FW24, Thm. D], is an independent continuity result rather than the Robbin–Salamon index theorem.

full rationale

The central claim, index(DA) = ς(A), is not assumed as an input. The proof of Theorem A proceeds by first establishing the Fredholm property of the augmented operator, then proving the index formula by homotoping invertible paths to constants (Step 1), decomposing a general path into subintervals where shifted operators are invertible (Step 2), and summing eigenvalue-crossing contributions via the spectral content ρ and the counting function ν↑ (Step 3). The identification with the spectral flow is made at the end through equation (4.68) and the surrounding bookkeeping, not by importing the desired index formula. The paper does rely on [FW24, Thm. D] to know that the spectral projections π± vary continuously with the operator; this is a self-citation, but it is a separate statement about continuity of spectral projections with stated assumptions that do not include the index theorem. It is parameter-free and externally checkable, so under the rules here it counts as independent support rather than circularity. A separate reviewer concern is that Definition 3.1 is delicate: fixing a zero branch a0(s) ≡ 0 and requiring continuous branches satisfying (i) and (ii) is problematic when an eigenvalue crosses zero, and the Normalization example in Lemma 3.2 appears to assign the value 0 to a branch that should be positive at T. However, that is a definitional or correctness issue with the right-hand side ς(A), not a case of the proof deriving its conclusion from that conclusion; the index computation itself is performed with eigenvalue counts. Accordingly, no circular step can be exhibited, and the circularity score is 0.

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

No empirical free parameters are fitted and no new physical or mathematical entities are postulated. The paper introduces boundary condition spaces and augmented operators, but these are constructions within an existing Hilbert space pair framework, not independent entities with separate falsifiable handles.

assumptions (5)

standard math The Hilbert space pair has compact dense inclusion H1 -> H0 and both spaces are separable.
Used throughout; cited from [FW24, Cor. A.5] and Lemma A.7 rather than proved here.
standard math H-self-adjoint pair operators have real discrete spectrum and an orthonormal eigenvector basis.
Proved in Appendix E using the Hilbert-Schmidt theorem for compact self-adjoint inverse operators; this underpins the spectral projections and spectral flow definitions.
standard math Interpolation spaces H1/2 transform consistently under equivalent inner products, via the Stein-Weiss interpolation theorem.
Used in Section 2.2 to make the boundary condition spaces well defined up to equivalent norms.
domain assumption The spectral projections πA± depend continuously on A in L(H1/2) for invertible H0-symmetric pair operators.
Cited as [FW24, Thm. D] and invoked in Step 1 of the Theorem A index formula and in the half-line homotopy argument; not proved in this paper.
standard math Openness of invertibility and the quantitative invertibility lemma for bounded operators.
Proved in Appendix B and used repeatedly in the Rabier-style estimates and perturbation arguments.

how reviews work

0 comments

Cite this review

Pith. "Pith review of On the spectral flow theorem of Robbin-Salamon for finite intervals." pith.science (2026). https://pith.science/paper/77I5EAUC

@misc{pith2026241216332,
  author       = {Pith},
  title        = {Pith review of: On the spectral flow theorem of Robbin-Salamon for finite intervals},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/77I5EAUC}},
  note         = {Machine review of arXiv:2412.16332}
}

abstract

In this article we consider operators of the form $\partial_s\xi+A(s)\xi$ where $s$ lies in an interval $[-T,T]$ and $s\mapsto A(s)$ is continuous. Without boundary conditions these operators are not Fredholm. However, using interpolation theory one can define suitable boundary conditions for these operators so that they become Fredholm. We show that in this case the Fredholm index is given by the spectral flow of the operator path $A$.

Figures

Figures reproduced from arXiv: 2412.16332 by the authors.

**Figure 1.** Spectral flow ς(A) = 2 along [−T, T] Lemma 3.2. The map A∗ IT → Z, A 7→ ς(A), has the following properties. (Homotopy) ς is constant on the connected components of A∗ IT . 12 [PITH_FULL_IMAGE:figures/full_fig_p012_1.png] view at source ↗

**Figure 2.** Spectral flow ς(A) = 0 along R+ Half-infinite backward interval Definition 3.4 (Spectral flow - half-infinite backward interval). Given a backward path A ∈ A∗ I− , we define a forward path A˜ ∈ A∗ I+ by A˜(s) := A(−s). Then we define the spectral flow of the backward path as the spectral flow of the negative forward path, in symbols ς(A) := ς(−A˜). Real line Similarly as in the case of half infinite intervals the s… view at source ↗

**Figure 3.** Approximate DA via finitely many invertible DA λ(σj ) , (4.31), (4.32) Step 5: But invertibility is an open property, thus A µσ (τ ) := A(τ ) − µσι is still invertible in a sufficiently narrow interval about σ, in symbols ∀τ ∈ Iσ := (σ − εσ, σ + εσ). The compact interval [−T3, T3] is covered by finitely many intervals Iσ1 , Iσ2 , . . . , IσN , see [PITH_FULL_IMAGE:figures/full_fig_p016_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: DA = ∂s + A(s) on P1(IT ) and its restriction D +− A to P +− 1 Proof of Theorem 4.11. Pick A ∈ A∗ IT , then A−T := A(−T) and AT := A(T) are invertible. By Corollary 4.10 the operator D +− A (and also D +− −A∗ ) has finite dimensional kernel and closed image. It remains…

**Figure 5.** Figure 5: Step 2: Invertibility shifts λi and intervals [tj , tj+1] Proof. For each σ ∈ [−T, T] choose µσ ∈ R\spec A(σ). So A(σ)−µσι: H1 → H0 is invertible. Since invertibility is an open condition, there exists εσ > 0 such that A(τ ) − µσι is invertible for every τ ∈ Iσ := (σ −…

**Figure 6.** Figure 6: DA = ∂s + A(s) on P1 and its restriction D + A to P + 1 Theorem 4.24 (Fredholm). D + A : P + 1 (R+, A0; H1, H0) → P0(R+; H0) is a Fredholm operator for any Hessian path A ∈ A∗ R+ . Corollary 4.25. The operator DA : P1(R+; H1, H0) → P0(R+; H0) in (4.72) has closed imag…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

17 extracted references · 12 canonical work pages

[1]

Transversal mappings and flows

Ralph Abraham and Joel Robbin. Transversal mappings and flows . An appendix by Al Kelley. W. A. Benjamin, Inc., New York-Amsterdam, 1967

work page 1967
[2]

o ran Bergh and J\

J\" o ran Bergh and J\" o rgen L\" o fstr\" o m. Interpolation spaces. A n introduction , volume No. 223 of Grundlehren der Mathematischen Wissenschaften . Springer-Verlag, Berlin-New York, 1976

1976
[3]

Functional analysis, S obolev spaces and partial differential equations

Ha \" m Brezis. Functional analysis, S obolev spaces and partial differential equations . Universitext. Springer, New York, 2011

work page 2011
[4]

First steps in the geography of scale Hilbert structures

Urs Frauenfelder . First steps in the geography of scale Hilbert structures . arXiv e-prints , page arXiv:0910.3980, Oct 2009

work page Pith review arXiv 2009
[5]

The shift map on Floer trajectory spaces

Urs Frauenfelder and Joa Weber. The shift map on Floer trajectory spaces . J. Symplectic Geom. , 19(2):351--397, 2021. arXiv:1803.03826 https://arxiv.org/abs/1803.03826

arXiv 2021
[6]

Growth of eigenvalues of Floer Hessians

Urs Frauenfelder and Joa Weber . Growth of eigenvalues of Floer Hessians . viXra e-prints https://vixra.org/author/joa_weber science, freedom, dignity , pages 1--50, 2024. viXra:2411.0060 https://vixra.org/abs/2411.0060

arXiv 2024
[7]

Hofer, K

H. Hofer, K. Wysocki, and E. Zehnder. The dynamics on three-dimensional strictly convex energy surfaces. Ann. of Math. (2) , 148(1):197--289, 1998

work page 1998
[8]

Sobolev Spaces of Vector-Valued Functions

Marcel Kreuter . Sobolev Spaces of Vector-Valued Functions . Master's thesis, Universit \"a t Ulm, April 2015

work page 2015

Show all 17 references

[9]

J -holomorphic curves and symplectic topology , volume 52 of American Mathematical Society Colloquium Publications

Dusa McDuff and Dietmar Salamon. J -holomorphic curves and symplectic topology , volume 52 of American Mathematical Society Colloquium Publications . American Mathematical Society, Providence, RI, 2004

2004
[10]

u ller. Spectral Theory of Linear Operators -- and Spectral Systems in Banach Algebras , volume 139 of Operator Theory: Advances and Applications . Birkh\

Vladimir M \" u ller. Spectral Theory of Linear Operators -- and Spectral Systems in Banach Algebras , volume 139 of Operator Theory: Advances and Applications . Birkh\" a user Verlag, Basel, 2nd edition, 2007

2007
[11]

The curve shrinking flow, compactness and its relation to scale manifolds (access pdf http://www.ime.unicamp.br/ joa/P/Neumeister-(2020)-master-thesis.pdf)

Oliver Neumeister. The curve shrinking flow, compactness and its relation to scale manifolds (access pdf http://www.ime.unicamp.br/ joa/P/Neumeister-(2020)-master-thesis.pdf). Master's thesis, Universit \"a t Augsburg, July 2020

2020
[12]

Patrick J. Rabier. The R obbin- S alamon index theorem in B anach spaces with UMD . Dyn. Partial Differ. Equ. , 1(3):303--337, 2004

2004
[13]

Roubi c ek

T. Roubi c ek. Nonlinear Partial Differential Equations with Applications , volume 153 of International Series of Numerical Mathematics . Birkh\"auser, Basel, 2nd ed. edition, 2013

2013
[14]

Methods of modern mathematical physics

Michael Reed and Barry Simon. Methods of modern mathematical physics . I . F unctional analysis . Academic Press, Inc. [Harcourt Brace Jovanovich, Publishers], New York, second edition, 1980. Functional analysis

1980
[15]

The spectral flow and the Maslov index

Joel Robbin and Dietmar Salamon . The spectral flow and the Maslov index . Bull. Lond. Math. Soc. , 27(1):1--33, 1995

1995
[16]

Functional analysis

Walter Rudin . Functional analysis. 2nd ed. International Series in Pure and Applied Mathematics. New York, NY: McGraw-Hill, 2nd ed. edition, 1991

1991
[17]

A Hardy Space Approach to Lagrangian Floer Gluing

Tatjana Sim c evi\' c . A Hardy Space Approach to Lagrangian Floer Gluing . PhD thesis, ETH Z\"urich, October 2014. https://doi.org/10.3929/ethz-a-010271531 https://doi.org/10.3929/ethz-a-010271531

2014 doi

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [1]

Transversal mappings and flows

Ralph Abraham and Joel Robbin. Transversal mappings and flows . An appendix by Al Kelley. W. A. Benjamin, Inc., New York-Amsterdam, 1967

work page 1967

[2] [2]

o ran Bergh and J\

J\" o ran Bergh and J\" o rgen L\" o fstr\" o m. Interpolation spaces. A n introduction , volume No. 223 of Grundlehren der Mathematischen Wissenschaften . Springer-Verlag, Berlin-New York, 1976

1976

[3] [3]

Functional analysis, S obolev spaces and partial differential equations

Ha \" m Brezis. Functional analysis, S obolev spaces and partial differential equations . Universitext. Springer, New York, 2011

work page 2011

[4] [4]

First steps in the geography of scale Hilbert structures

Urs Frauenfelder . First steps in the geography of scale Hilbert structures . arXiv e-prints , page arXiv:0910.3980, Oct 2009

work page Pith review arXiv 2009

[5] [5]

The shift map on Floer trajectory spaces

Urs Frauenfelder and Joa Weber. The shift map on Floer trajectory spaces . J. Symplectic Geom. , 19(2):351--397, 2021. arXiv:1803.03826 https://arxiv.org/abs/1803.03826

arXiv 2021

[6] [6]

Growth of eigenvalues of Floer Hessians

Urs Frauenfelder and Joa Weber . Growth of eigenvalues of Floer Hessians . viXra e-prints https://vixra.org/author/joa_weber science, freedom, dignity , pages 1--50, 2024. viXra:2411.0060 https://vixra.org/abs/2411.0060

arXiv 2024

[7] [7]

Hofer, K

H. Hofer, K. Wysocki, and E. Zehnder. The dynamics on three-dimensional strictly convex energy surfaces. Ann. of Math. (2) , 148(1):197--289, 1998

work page 1998

[8] [8]

Sobolev Spaces of Vector-Valued Functions

Marcel Kreuter . Sobolev Spaces of Vector-Valued Functions . Master's thesis, Universit \"a t Ulm, April 2015

work page 2015