A family of algebraic operations extending the Turaev cobracket

Toyo Taniguchi

REVIEW 2 major objections 3 minor 1 cited by

A family of algebraic operations extending the Turaev cobracket

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

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

Pith's one-line read The k-divergence maps on cyclic words coincide with bivalent ribbon-graph operations, and flatness makes them Lie algebra cocycles.

desk verdict A genuinely new family of operations that plausibly extends the Turaev cobracket to ribbon graph operations, but the proof of the main identification has a clear overcounting error that needs fixing before publication. read the letter →

arxiv 2502.04806 v2 pith:YTCLVQB7 submitted 2025-02-07 math.QA math.ATmath.RA

Toyo Taniguchi This is my paper · ORCID

classification math.QAmath.ATmath.RA MSC 16D2017B5617B6553C0557K2058B34

keywords non-commutativegeometrydivergencemapsflatconnectionsribbongraphsTuraevcobracketloopoperationsLiealgebracohomologyfreeassociative

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 constructs a family of algebraic operations, the k-divergences, that extend the Turaev cobracket. The central claim is that on the free associative algebra T(W), the k-th k-divergence coincides, up to a sign, with the operation encoded by a specific bivalent connected ribbon graph with k vertices and k edges; the proof is a direct trace computation using the canonical flat connection on noncommutative 1-forms. When the underlying connection is flat, the skew-symmetrised k-divergences are Lie algebra k-cocycles on the derivation algebra, and for odd k on T(W) they give non-zero cohomology classes that restrict to the standard generators of the cohomology ring of gl_n. Applied to the group algebra of a surface group, the k=1 case reproduces the framed Turaev cobracket, so the family is an algebraic generalisation of that topological operation.

What carries the argument

The central object is the k-divergence map $\mathrm{Div}^{\nabla}_k(f_1,\dots,f_k)=\operatorname{Tr}(c_\nabla(f_1)\circ\cdots\circ c_\nabla(f_k))$, where $c_\nabla(f)=(i_{\phi(f)}\otimes\mathrm{id})\circ\nabla-\rho(f)$ measures the failure of the connection to be invariant under the derivation action. For the tensor algebra, $\nabla_W$ is the canonical flat connection defined by $\nabla_W(dw)=0$ for $w\in W$, and $\mathrm{Ham}^{\langle\cdot,\cdot\rangle}$ is the Hamiltonian flow associated with the skew pairing viewed as a double bracket. The trace lands in the cyclic quotient $|B|=B/[B,B]$, so the whole construction is an algebraic analogue of taking divergence of vector fields; the ribbon graph operation in Theorem 3.4 is the pairwise-contraction-and-boundary-tracing recipe coming from the prop of ribbon graphs.

What would settle it

Compute $\nabla_C^2$ on a noncommutative 1-form for a free generating system $C$ of a surface group; if the result is non-zero for some such system, the claim that the family generalises the framed Turaev cobracket loses its foundation. Alternatively, evaluate both sides of Theorem 3.4 for k=2 and W=$K^{2}$ with a fixed skew pairing: if the trace computation gives a cyclic word not equal to the $L_2$ ribbon-graph contraction, the identification fails.

Watch

Extended reading notes

Core claim

At the heart of the paper is the statement (Theorem 3.4) that for a finite-dimensional K-vector space W with a skew-symmetric pairing $\langle\cdot,\cdot\rangle$, the map $$(-1)^k \$delta^{{\mathrm{Ham}}$^{\langle\cdot,\cdot\rangle},\nabla_W}_k \colon |T(W)|^{\otimes k}\to |T(W)|^{\otimes 2}$$ coincides with the operation associated with the unique bivalent connected ribbon graph $L_k$ having $k$ vertices and $k$ edges. Here $|T(W)|$ is the cyclic quotient of the tensor algebra (the space of cyclic words), $\nabla_W$ is the canonical flat connection defined by $\nabla_W(dw)=0$ for $w\in W$, and $\mathrm{Ham}^{\langle\cdot,\cdot\rangle}$ is the Hamiltonian flow induced by the pairing. The equality is proved by expanding the trace of a product of endomorphisms and observing that each term is exactly a contraction of letters along the edges of $L_k$. A second theorem (Theorem 4.14) states that for any algebra with a flat connection the skew-symmetrised k-divergence $\mathrm{Div}^{\nabla}_k\circ \mathrm{alt}$ is a Lie algebra k-cocycle on the derivation algebra, vanishing for even k; and for $A=T(W)$ with odd k the class of $\operatorname{Tr}(c_{\nabla_W}^k)$ does not vanish, since it restricts to the standard generator of the cohomology ring of $\mathfrak{gl}_n$.

Load-bearing premise

The surface-group interpretation depends on the existence of a flat connection $\nabla_C$ on $\Omega^1K\pi$ satisfying $\nabla_C((dc)c^{-1})=0$ for each free generator $c$; that existence is quoted from the author's previous paper rather than proved here.

Editorial extensions

If this is right

For the tensor algebra, the identification with $L_k$ upgrades the necklace space $|T(W)|$ from an involutive Lie bialgebra to a full representation of the ribbon-graph prop, for every $k$.
Whenever a flat connection exists, each odd $k$ gives a Lie algebra cocycle on $\mathrm{Der}_K(A)$ with values in $|A|$; in particular these are algebraic analogues of the divergence cocycle of classical differential geometry.
On $T(W)$, the non-vanishing classes $\operatorname{Tr}(c_{\nabla_W}^k)$ for odd $k$ restrict to the standard generators of $H^*_{\mathrm{CE}}(\mathfrak{gl}_n,K)$, so the construction detects the full cohomology of the general linear Lie algebra inside the derivation cohomology.
For a compact oriented surface with boundary, the $k=1$ member is the framed Turaev cobracket, so the whole family is an algebraic extension of that loop operation to higher 'simultaneous intersection' operations.

Reading between the lines

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

A topological interpretation of the higher $\delta^{\sigma,\nabla_C}_k$ for surface groups is not given; if one existed, it would amount to resolving several curve intersections simultaneously, which the author notes earlier smoothing-type attempts failed to do.
Because the operations are not mapping-class-group equivariant, any such topological lift would need extra data beyond the surface itself; testing whether the $\delta^{\sigma,\nabla_C}_k$ satisfy the higher relations of ribbon-graph props would be a concrete check of how close the algebraic family is to a true loop operation.
The even-k vanishing comes from cyclic symmetry of the trace; dropping the anti-symmetrisation or allowing non-flat connections could produce non-vanishing even cocycles, which would test whether flatness is the essential hypothesis.

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

The k-divergence family is a real contribution. The construction is elegant: a cyclic trace of a composition of connection operators, which for k=1 recovers the divergence that gives the framed Turaev cobracket, and for general k lands in the ribbon graph prop of Merkulov–Willwacher. The identification in Theorem 3.4 is the kind of result people will want to cite, and Theorems 4.14 and 4.15 (cocycle and nonvanishing via Fuks) are solid and independently useful. The non-commutative de Rham framework with curved DGLA structures is well motivated.

The soft spot is the proof of Theorem 3.4. The stress-test note is right: the displayed formula after taking the trace contains a sum over i whose terms are completely independent of i. For k=2 that would multiply the claimed value by 2. A direct computation with W=span{a,b} and w1=w2=ab gives δ_2 = 2(1⊗1), matching the ribbon graph operation, so the theorem statement is likely correct—but the proof as written overcounts. This is the central theorem, so the proof needs to be fixed, either by removing the extraneous sum or by cyclically shifting the indices so that the sum actually reflects the k possible choices of where the final d lands. Without that fix, the identification is not rigorously established.

A second, smaller issue: Definition 2.3 says the k-divergence takes values in |B|, but Theorem 3.4 and the proof clearly use a double trace into |B|⊗|B|. That mismatch should be clarified, since the reader has to infer the intended trace convention.

The surface-group interpretation relies on a flat connection from the author's previous paper [Tan24b]; that's acceptable, but it means the geometric generalization is only as solid as that cited construction. Theorem 4.10's proof is terse but appears correct. The citation pattern is fine; self-citations are for context and the k=1 base case, not for the new claims.

This paper deserves a serious referee. The ideas are good, the theorems are plausible, and the flaws are local and fixable. I would send it to peer review and ask for a corrected proof of Theorem 3.4 plus a clear statement of the trace conventions.

Referee Report

2 major / 3 minor

Summary. The paper introduces a family of maps, the k-divergences Div^∇_k, associated to an algebra A, a derivation action, and a connection ∇, together with their restrictions δ^{ψ,∇}_k to a Lie subalgebra. The main structural results are: (i) Theorem 3.4, which identifies (−1)^k δ^{Ham,∇_W}_k on the trace space of the free associative algebra T(W) with the operation of the bivalent connected ribbon graph L_k from Merkulov–Willwacher; (ii) Theorem 4.14, which states that for a flat connection the skew-symmetrized k-divergence is a Lie algebra k-cocycle; and (iii) Theorem 4.15, which proves nonvanishing of certain cohomology classes via Fuks’ theorem. The paper also discusses the surface-group case, where the k=1 map recovers the framed Turaev cobracket.

Significance. If the results are correct, the paper offers a genuinely new algebraic family of loop operations that interpolates between the Turaev cobracket and the Merkulov–Willwacher ribbon graph operations, with applications to Lie algebra cohomology of derivation algebras. The paper is commendably concrete: the definitions are explicit, the tensor-algebra computations are carried out by hand, and the nonvanishing theorem uses a standard external result (Fuks) rather than a fitted or circular argument. The acknowledgement of a previous error and the correction of the proof are also positive signs. However, the proof of the central Theorem 3.4, as written, contains a serious computational error that prevents the main ribbon-graph identification from being regarded as established.

major comments (2)

[§3, Proof of Theorem 3.4] The displayed formula after “by taking the trace” contains a summation ∑_{1≤i≤k} whose summand is independent of i. For each i, the product of pairings is the same up to permutation, and the two output cyclic words are cyclic rotations of those for i=1; since the outputs are taken in the trace space |T(W)|, the terms are identical. Thus the displayed right-hand side carries an extra factor k. For k=2, with W=span{a,b}, ⟨a,b⟩=1, and w1=w2=ab, a direct computation from Definition 2.3 gives c_{∇W}(Ham(ab)) = diag(1,−1) in the basis {da,db}, so δ_2(w1,w2)=Tr(id)=2(1⊗1). This contradicts the factor-k version of the formula. The theorem statement may nevertheless be correct, but the proof as written does not establish it.
[§3, Proof of Theorem 3.4] Independently of the extraneous factor k, the output words in the displayed trace formula do not correctly implement the deletion of the letters contracted by the trace. In the same k=2 example with w1=w2=ab, take s1=2,t1=1,s2=2,t2=1. The formula produces a second output word |w1_{t1+1}⋯w1_{s1-1} w2_{t2+1}⋯w2_{s2-1}| = |baba| = |abab|, although the letters at positions t1,s1,t2,s2 are precisely the ones paired and traced, so they should not survive. The direct trace computation gives 2(1⊗1), with no such word. The displayed computation therefore has an indexing error that is independent of the factor-k issue, and the proof of Theorem 3.4 needs to be rewritten.

minor comments (3)

[Remark 4.16] The cross-reference to “Proposition 3.4” should be to “Theorem 3.4”; the cited result is a theorem, not a proposition.
[Example 2.8] The table would benefit from a sentence explaining the convention for the empty word in the output (e.g., whether |1| is denoted by 1), since some rows involve traces of empty words.
[Remark 2.7] The existence and flatness of the connection ∇_C on Ω^1 Kπ is cited from [Tan24b]; since this is a load-bearing ingredient for the surface interpretation, a brief reminder of the construction or a precise reference to the statement in [Tan24b] would help the reader.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the tensor-algebra theorem is a direct computation from the definitions, and the self-citations supply background context rather than load-bearing inputs.

full rationale

The paper's central derivation chain is self-contained. Theorem 3.4 compares two independently defined objects: the k-divergence δ^{Ham,∇_W}_k, defined in Definitions 2.2 and 2.3 as a trace of a product of operators c_∇(Ham(w_i)), and the ribbon-graph operation L_k, defined in Section 3 by summing pairings along the edges of a bivalent ribbon graph. The proof is an explicit computation from those definitions, not a renaming or a fitted parameter. The canonical connection ∇_W and the Hamiltonian map Ham are definitions (Definitions 3.1 and 3.2), and no data-dependent parameter is adjusted. The self-citations to [Tan24b] supply the k=1 surface-group interpretation and the flat connection on Ω^1Kπ; even if those citations were absent, Theorems 3.4, 4.14, and 4.15 would remain derived within the paper, so the self-citations are not load-bearing for the main new assertions. A prior result cited with stated assumptions is independent support rather than circularity. One correctness concern, not a circularity: in the proof of Theorem 3.4, the displayed sum over 1≤i≤k appears to have an i-independent summand, which would overcount by a factor k; if real, this would make the proof as written unsound, but it would not make the claimed equality equivalent to its input by definition.

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

The central claims rest on standard background (trace spaces, ribbon graph props, Fuks' theorem) and on two cited results from the author's prior work and Alekseev et al. No free parameters are tuned; no new entities are postulated.

assumptions (7)

standard math Trace space |B| = B/[B,B] and the cyclic trace map for dualisable modules are well-defined.
Used throughout Section 2 and 4 to define Div^∇_k and to prove its cyclic symmetry (Definition 2.1, Definition 2.3, Proposition 2.4).
domain assumption The representation of the prop RGra1 on |T(W)| defined in [MW15] computes the operation of a ribbon graph by pairing letters according to the graph's edges.
Assumed in Section 3 when interpreting δ as L_k; the computation of Γ(w_1,...,w_n) in Section 3 is taken as the definition of the ribbon graph operation.
standard math Fuks' theorem: the cohomology ring H^*_CE(gl(W),K) is the exterior algebra on classes φ_k for 1≤k≤2dim(W)-1, k odd.
Used in the proof of Theorem 4.15 to conclude j^*(Tr(c^k_∇_W)) = (-1)^k φ_k is nonzero.
domain assumption The 'centre kills' argument: H^*_CE(gl(W), j^*V) is isomorphic to H^*_CE(gl(W), K) because the Euler operator acts by multiplication by degree on V.
Invoked in the proof of Theorem 4.15 to reduce the coefficient module to K; this is a standard equivariant cohomology technique but is not proved in the paper.
domain assumption The flat connection ∇_C on Ω^1Kπ defined by ∇_C((dc)c^{-1})=0 for a generating set C exists and is flat.
Stated in Remark 2.7 with a reference to [Tan24b]; needed for the surface-group case to identify -δ^{σ,∇_C}_1 with the framed Turaev cobracket.
standard math The identification |A^e| ≅ |A|^{⊗2} of trace spaces.
Used implicitly to view δ as a map |A|^{⊗k} → |A|^{⊗2}; follows from standard Hochschild homology results for tensor products of algebras.
domain assumption The result of [AKKN23] and [Tan24b] that -δ^{σ,∇_C}_1 equals the framed Turaev cobracket.
This is the k=1 base case that motivates calling the family a generalisation of the Turaev cobracket; it is an external theorem.

how reviews work

0 comments

Cite this review

Pith. "Pith review of A family of algebraic operations extending the Turaev cobracket." pith.science (2026). https://pith.science/paper/YTCLVQB7

@misc{pith2026250204806,
  author       = {Pith},
  title        = {Pith review of: A family of algebraic operations extending the Turaev cobracket},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YTCLVQB7}},
  note         = {Machine review of arXiv:2502.04806}
}

abstract

We introduce a family of maps parametrised by certain ribbon graphs. It is based on a connection in non-commutative geometry and contains the double divergence as a special case. Applying the construction to the case of the group algebra of the fundamental group of a compact connected oriented surface with boundary, we obtain an algebraic generalisation of the Turaev cobracket. If the connection is flat, they define classes in the Lie algebra cohomology of the space of derivations. In the case of the free associative algebra, we show that they are canonically identified with the standard generators of the cohomology ring of the matrix Lie algebra $\mathfrak{gl}_n$.

Figures

Figures reproduced from arXiv: 2502.04806 by the authors.

**Figure 1.** The ribbon graph Lk ∈ Rk 2,k. The labelling of the edges is suppressed, as any choice of these gives the same element in RGra1(2, k). Theorem 3.4. Suppose that the pairing h·, ·i: W ⊗ W → K is skew-symmetric. Then, the map (−1)k δ Hamh·,·i,∇W k : |T (W)| ⊗k → |T (W)| ⊗2 coincides with the operation given by the ribbon graph Lk in [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

The formality of the Goldman-Turaev Lie bialgebra on a closed surface
math.QA 2025-02 accept novelty 8.0 of 10

The pro-unipotent automorphism group of the associated graded Goldman-Turaev Lie bialgebra on a closed surface is explicitly described in terms of a divergence map and the kernel of a reduced coproduct.

Reference graph

Works this paper leans on

16 extracted references · 4 canonical work pages · cited by 1 Pith paper

[5]

Stable homology of Lie algebras of derivations and homotopy invariants of wheeled operads

arXiv:2311.18594. 4 [Fuk86] Dmitry Borisovich Fuks. Cohomology of Inﬁnite-Dimensional Lie Algebras . Springer US,

work page Pith review arXiv
[8]

1 [KK14] Nariya Kawazumi and Yusuke Kuno

doi:10.1007/BF01389091. 1 [KK14] Nariya Kawazumi and Yusuke Kuno. The logarithms of De hn twists. Quantum Topology, 5(3):347–423,

doi:10.1007/bf01389091
[13]

3.3 [Tan24b] Toyo Taniguchi

arXiv:2410.24064. 3.3 [Tan24b] Toyo Taniguchi. Non-commutative divergence and t he Turaev cobracket. To appear in Algebraic & Geometric Topology,

arXiv
[14]

1, 2, 2.5, 2.7, 4, A, A [Tur91] Vladimir G

arXiv:2403.16566. 1, 2, 2.5, 2.7, 4, A, A [Tur91] Vladimir G. Turaev. Skein quantization of Poisson a lgebras of loops on surfaces. Annales scientiﬁques de l’´Ecole Normale Sup´ erieure, Ser. 4, 24(6):635–704,

arXiv
[1986]

4, 4 14 [Gin05] Victor Ginzburg

doi:10.1007/978-1-4684-8765-7 . 4, 4 14 [Gin05] Victor Ginzburg. Lectures on noncommutative geome try

work page doi:10.1007/978-1-4684-8765-7
[1991]

1 [vdB08] Michel van den Bergh

doi:10.24033/asens.1639. 1 [vdB08] Michel van den Bergh. Double Poisson algebras. Transactions of the American Mathematical Society , 360(11):5711–5769,

doi:10.24033/asens.1639
[2004]

2.8 [CK21] Moira Chas and Arpan Kabiraj

arXiv:math/0105178. 2.8 [CK21] Moira Chas and Arpan Kabiraj. The Lie bracket of undir ected closed curves on a surface. Transactions of the American Mathematical Society , 375(4):2365–2386,

arXiv
[2005]

Lectures on Noncommutative Geometry

arXiv:math/0506603. 2 [Gol86] William M. Goldman. Invariant functions on Lie grou ps and Hamiltonian ﬂows of surface group representa- tions. Inventiones mathematicae , 85(2):263–302,

work page Pith review arXiv

Show all 16 references

[2008]

arXiv:math/0410528. 3.3 15

arXiv
[2009]

4.16 [MW15] Sergei Merkulov and Thomas Willwacher

arXiv:0707.0889v4, doi:10.1515/crelle.2009.084. 4.16 [MW15] Sergei Merkulov and Thomas Willwacher. Props of ribb on graphs, involutive Lie bialgebras and moduli spaces of curves

2009 arXiv
[2014]

1 [KK16] Nariya Kawazumi and Yusuke Kuno

arXiv:1008.5017v1, doi:10.4171/QT/54. 1 [KK16] Nariya Kawazumi and Yusuke Kuno. The Goldman–Turaev Lie bialgebra and the Johnson homomor- phisms, volume V of Handbook of Teichm¨ uller Theory, pages 97–165. EMS Press,

arXiv
[2015]

1, 3, 3, 3, 3, 3.6, 4.16 [Tan24a] Toyo Taniguchi

arXiv:1511.07808. 1, 3, 3, 3, 3, 3.6, 4.16 [Tan24a] Toyo Taniguchi. Modular vector ﬁelds in non-commu tative geometry

arXiv
[2016]

1, 3.6 [MV09] Sergei Merkulov and Bruno Vallette

arXiv:1304.1885, doi:10.4171/160. 1, 3.6 [MV09] Sergei Merkulov and Bruno Vallette. Deformation the ory of representations of prop(erad)s II. Jour- nal f¨ ur die reine und angewandte Mathematik (Crelles Journ al), 2009(636),

2009 arXiv
[2021]

2.8 [Dot23] Vladimir Dotsenko

arXiv:1910.08991, doi:10.1090/tran/8541. 2.8 [Dot23] Vladimir Dotsenko. Stable homology of Lie algebras of derivations and homotopy invariants of wheeled operads

1910 arXiv
[2023]

1, 2.7 [ANPˇS24] Anton Alekseev, Florian Naef, J´ an Pulmann, and Pavol ˇSevera

arXiv:1804.09566v3. 1, 2.7 [ANPˇS24] Anton Alekseev, Florian Naef, J´ an Pulmann, and Pavol ˇSevera. Batalin-Vilkovisky structures on moduli spaces of ﬂat connections. Advances in Mathematics , 443:109580,

arXiv
[2024]

1 [Cha04] Moira Chas

arXiv:2210.08944, doi:10.1016/j.aim.2024.109580. 1 [Cha04] Moira Chas. Combinatorial Lie bialgebras of curves on surfaces. Topology, 43(3):543–568,

2024

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.

[1] [5]

Stable homology of Lie algebras of derivations and homotopy invariants of wheeled operads

arXiv:2311.18594. 4 [Fuk86] Dmitry Borisovich Fuks. Cohomology of Inﬁnite-Dimensional Lie Algebras . Springer US,

work page Pith review arXiv

[2] [8]

1 [KK14] Nariya Kawazumi and Yusuke Kuno

doi:10.1007/BF01389091. 1 [KK14] Nariya Kawazumi and Yusuke Kuno. The logarithms of De hn twists. Quantum Topology, 5(3):347–423,

doi:10.1007/bf01389091

[3] [13]

3.3 [Tan24b] Toyo Taniguchi

arXiv:2410.24064. 3.3 [Tan24b] Toyo Taniguchi. Non-commutative divergence and t he Turaev cobracket. To appear in Algebraic & Geometric Topology,

arXiv

[4] [14]

1, 2, 2.5, 2.7, 4, A, A [Tur91] Vladimir G

arXiv:2403.16566. 1, 2, 2.5, 2.7, 4, A, A [Tur91] Vladimir G. Turaev. Skein quantization of Poisson a lgebras of loops on surfaces. Annales scientiﬁques de l’´Ecole Normale Sup´ erieure, Ser. 4, 24(6):635–704,

arXiv

[5] [1986]

4, 4 14 [Gin05] Victor Ginzburg

doi:10.1007/978-1-4684-8765-7 . 4, 4 14 [Gin05] Victor Ginzburg. Lectures on noncommutative geome try

work page doi:10.1007/978-1-4684-8765-7

[6] [1991]

1 [vdB08] Michel van den Bergh

doi:10.24033/asens.1639. 1 [vdB08] Michel van den Bergh. Double Poisson algebras. Transactions of the American Mathematical Society , 360(11):5711–5769,

doi:10.24033/asens.1639

[7] [2004]

2.8 [CK21] Moira Chas and Arpan Kabiraj

arXiv:math/0105178. 2.8 [CK21] Moira Chas and Arpan Kabiraj. The Lie bracket of undir ected closed curves on a surface. Transactions of the American Mathematical Society , 375(4):2365–2386,

arXiv

[8] [2005]

Lectures on Noncommutative Geometry

arXiv:math/0506603. 2 [Gol86] William M. Goldman. Invariant functions on Lie grou ps and Hamiltonian ﬂows of surface group representa- tions. Inventiones mathematicae , 85(2):263–302,

work page Pith review arXiv