Quasi-Poisson Double Structures

• Quasi-Poisson doubles are structures where a doubled object carries a bracket with a controlled Jacobi defect governed by a canonical 3-tensor.
• They extend Van den Bergh’s double bracket formalism to contexts like quasi-Hamiltonian and quasi-Drinfeld doubles, enabling applications in fusion, twist, and symmetry reductions.
• Recent advancements generalize these constructions to types B, C, and D, impacting quiver theory, character varieties, and integrable noncommutative systems.

Searching arXiv for the supplied paper and closely related work on quasi-Poisson doubles, double quasi-Poisson algebras, and quasi-Poisson geometry on representation spaces. {"query":"(Arthamonov et al., 22 May 2026) Quasi-Poisson varieties from double quasi-Poisson algebras in types B,C,D; double quasi-Poisson brackets fusion and new examples; Quasi-Poisson structures on representation spaces of surfaces; Euler continuants in noncommutative quasi-Poisson geometry","max_results":10} Quasi-Poisson double denotes a family of constructions in which a “double” object carries a bracket whose Jacobi identity fails in a controlled way governed by a canonical $3$-tensor rather than by zero. In noncommutative Poisson geometry, the most systematic meaning is a double quasi-Poisson bracket on an associative algebra, introduced so that representation spaces inherit quasi-Poisson structures. In adjacent literatures, the same expression is used for quasi-Hamiltonian doubles such as $D(G)=G\times G$, for quasi-Drinfeld doubles of Lie quasi-bialgebras, and for categorical doubles realized as double symplectic groupoids. A 2026 extension places the noncommutative theory over an arbitrary semisimple base and upgrades it from type $\mathtt A$ to twisted representation spaces of types $\mathtt B,\mathtt C,\mathtt D$, with applications to quivers, orthogonal and symplectic character varieties, and integrable systems (Arthamonov et al., 22 May 2026).

1. Terminological scope and recurrent meanings

The expression is not completely uniform across the literature. In the associative-algebraic setting, a quasi-Poisson double is a double quasi-Poisson algebra in the sense of Van den Bergh, meaning an algebra equipped with an $A\otimes A$-valued double bracket whose triple Jacobiator is prescribed by gauge or Cartan terms. In quasi-Hamiltonian geometry, the phrase also refers to the double $D(G)$ of a Lie group, usually $G\times G$, endowed with a quasi-Poisson bivector or a quasi-Hamiltonian $2$-form and a group-valued moment map. In Lie quasi-bialgebra theory, it refers to the quasi-Drinfeld double $\mathfrak d^a=\mathfrak k\oplus\mathfrak k^*$. Several papers explicitly note that the phrase is being used retrospectively or implicitly rather than as a single universal definition (Álvarez, 2020, Klimcik, 2018).

Sense	Underlying object	Representative source
Noncommutative double	Associative algebra with double quasi-Poisson bracket	(Arthamonov et al., 22 May 2026)
Quasi-Hamiltonian double	$D(G)=G\times G$ with moment map	(Huebschmann, 2022)
Quasi-Drinfeld double	$D(G)=G\times G$0	(Klimcik, 2018)
Categorical double	Double symplectic groupoid from reduction	(Álvarez, 2020)

A persistent source of confusion is the word “double.” In the Van den Bergh formalism it refers to tensor-valued brackets on associative algebras; in quasi-Hamiltonian geometry it refers to a doubled group object; in the groupoid setting it refers to a double Lie or double symplectic groupoid. Likewise, “quasi-Poisson” never means an arbitrary Jacobi defect: in every one of these settings the defect is fixed by a canonical trivector or by an explicitly prescribed gauge term.

2. Double quasi-Poisson algebras and representation spaces

Let $D(G)=G\times G$1 be a finitely generated associative unital algebra over a semisimple base $D(G)=G\times G$2, with orthogonal idempotents $D(G)=G\times G$3 and $D(G)=G\times G$4. The tensor square $D(G)=G\times G$5 carries commuting outer and inner $D(G)=G\times G$6-bimodule structures, and a double bracket is a $D(G)=G\times G$7-bilinear map

$D(G)=G\times G$8

that is $D(G)=G\times G$9-linear, cyclically antisymmetric,

$\mathtt A$0

and satisfies the derivation rules

$\mathtt A$1

The associated triple bracket

$\mathtt A$2

measures the Jacobi defect. A double Poisson bracket is characterized by $\mathtt A$3. A double quasi-Poisson bracket is characterized by the fact that the double Jacobiator is governed by the Cartan trivector of the symmetry group, equivalently by the explicit idempotent-controlled $\mathtt A$4-term expression of Eq. (2.11) (Arthamonov et al., 22 May 2026).

Moment maps occur in additive and multiplicative forms. In the double Poisson case, $\mathtt A$5 with $\mathtt A$6 is a noncommutative moment map if

$\mathtt A$7

In the double quasi-Poisson case, an invertible element $\mathtt A$8, $\mathtt A$9, is a multiplicative moment map if

$\mathtt B,\mathtt C,\mathtt D$0

These formulas are the noncommutative analogues of additive and group-valued moment map identities (Arthamonov et al., 22 May 2026).

For a dimension vector $\mathtt B,\mathtt C,\mathtt D$1, the representation space $\mathtt B,\mathtt C,\mathtt D$2 parametrizes algebra maps $\mathtt B,\mathtt C,\mathtt D$3 compatible with the idempotent blocks. Van den Bergh’s formula

$\mathtt B,\mathtt C,\mathtt D$4

induces a bracket on $\mathtt B,\mathtt C,\mathtt D$5. In type $\mathtt B,\mathtt C,\mathtt D$6, this yields a $\mathtt B,\mathtt C,\mathtt D$7-invariant Poisson bracket in the double Poisson case and a $\mathtt B,\mathtt C,\mathtt D$8-invariant quasi-Poisson bracket in the double quasi-Poisson case, with Jacobiator equal to $\mathtt B,\mathtt C,\mathtt D$9 (Arthamonov et al., 22 May 2026). The same mechanism underlies the surface-group construction of Massuyeau and Turaev, where a quasi-Poisson double algebra on the group algebra $A\otimes A$0 induces a canonical quasi-Poisson bracket on representation spaces and extends Goldman’s bracket on trace functions (Massuyeau et al., 2012).

The algebraic theory is not restricted to differential double brackets. Fusion was originally available for a large differential class, but it was extended to arbitrary double quasi-Poisson brackets, including non-differential examples such as $A\otimes A$1 with $A\otimes A$2 and

$A\otimes A$3

which is quasi-Poisson but not differential (Fairon, 2019).

3. Twisted representation spaces and types $A\otimes A$4

The major 2026 development is the extension of the Van den Bergh–Olshanski–Safonkin mechanism from type $A\otimes A$5 to types $A\otimes A$6. The starting datum is an involutive algebra $A\otimes A$7, where $A\otimes A$8 is a $A\otimes A$9-linear anti-automorphism with $D(G)$0 and $D(G)$1. One also fixes an involution $D(G)$2 on $D(G)$3, either orthogonal,

$D(G)$4

or symplectic,

$D(G)$5

The twisted representation space is

$D(G)$6

In the orthogonal case, the defining relations are $D(G)$7; in the symplectic case they become $D(G)$8 (Arthamonov et al., 22 May 2026).

If $D(G)$9 is a double Poisson algebra and $G\times G$0 is compatible with the bracket in the sense that

$G\times G$1

then the induced twisted brackets are

$G\times G$2

in the orthogonal case and

$G\times G$3

in the symplectic case. These are Poisson brackets, and $G\times G$4 or $G\times G$5 acts by Poisson automorphisms. If $G\times G$6, then $G\times G$7 is a moment map with values in $G\times G$8 or $G\times G$9 (Arthamonov et al., 22 May 2026).

The quasi-Poisson analogue is formally parallel. If $2$0 is double quasi-Poisson and $2$1 is compatible, then the same formulas define quasi-Poisson brackets for the $2$2- and $2$3-actions. If $2$4 is a multiplicative moment map satisfying

$2$5

then $2$6 is a group-valued moment map into $2$7 or $2$8. Conceptually, the Jacobiator of the induced bracket equals $2$9 for the symmetry group $\mathfrak d^a=\mathfrak k\oplus\mathfrak k^*$0 or $\mathfrak d^a=\mathfrak k\oplus\mathfrak k^*$1; in the Poisson case the double Jacobiator vanishes, and so does the induced Jacobiator (Arthamonov et al., 22 May 2026).

A further refinement is the mixed-type theory. Each vertex may be assigned type $\mathfrak d^a=\mathfrak k\oplus\mathfrak k^*$2 or $\mathfrak d^a=\mathfrak k\oplus\mathfrak k^*$3, encoded by $\mathfrak d^a=\mathfrak k\oplus\mathfrak k^*$4 and a typed anti-involution $\mathfrak d^a=\mathfrak k\oplus\mathfrak k^*$5 with $\mathfrak d^a=\mathfrak k\oplus\mathfrak k^*$6 on $\mathfrak d^a=\mathfrak k\oplus\mathfrak k^*$7. The induced bracket on $\mathfrak d^a=\mathfrak k\oplus\mathfrak k^*$8 is then of mixed orthogonal-symplectic type, and its Jacobiator matches the sum over vertex-types. This is the mechanism by which a single quiver can carry orthogonal symmetry at some vertices and symplectic symmetry at others (Arthamonov et al., 22 May 2026).

4. Quivers, fusion, and multiplicative quiver geometry

Quivers supply the main computational laboratory for quasi-Poisson doubles. For a quiver $\mathfrak d^a=\mathfrak k\oplus\mathfrak k^*$9, the doubled path algebra carries Van den Bergh’s canonical double Poisson bracket

$D(G)=G\times G$0

with the other generator pairs zero, and additive moment map

$D(G)=G\times G$1

After localization, the multiplicative analogue carries a double quasi-Poisson bracket with

$D(G)=G\times G$2

modulo the ordering corrections of Eqs. (4.6)–(4.8) (Arthamonov et al., 22 May 2026).

Fusion is the operation that glues idempotents and produces new quasi-Hamiltonian algebras from old ones. For arbitrary double quasi-Poisson brackets, the fused bracket is

$D(G)=G\times G$3

and the fused multiplicative moment map is

$D(G)=G\times G$4

This extension of fusion beyond the differential setting is the central structural result of “Double quasi-Poisson brackets: fusion and new examples” (Fairon, 2019).

The quiver theory has several notable consequences. First, Van den Bergh’s quiver brackets and the Massuyeau–Turaev surface brackets admit alternative constructions by fusion (Fairon, 2019). Second, for the one-arrow quiver $D(G)=G\times G$5 and dimension vector $D(G)=G\times G$6, the twisted localized multiplicative representation spaces are identified with the orthogonal and symplectic doubles: $D(G)=G\times G$7 depending on whether $D(G)=G\times G$8 is transposition or symplectic transposition (Arthamonov et al., 22 May 2026). Third, the canonical double (quasi-)Poisson structure attached to a quiver depends only on the underlying undirected graph, up to isomorphism, and this orientation-independence can be exploited representation-theoretically in action-angle duality for integrable systems (Fairon, 2020).

A particularly explicit multiplicative theory appears for the two-vertex quiver $D(G)=G\times G$9 with $D(G)=G\times G$00 equioriented arrows. The localized algebra $D(G)=G\times G$01 carries a Hamiltonian double quasi-Poisson bracket whose multiplicative moment map is given by the two $D(G)=G\times G$02-th Euler continuants,

$D(G)=G\times G$03

After further localization, this algebra factors, via fusion, into $D(G)=G\times G$04 copies of the $D(G)=G\times G$05 algebra of Van den Bergh (Fairon et al., 2021). On $D(G)=G\times G$06-dimensional representation spaces, the induced Poisson brackets coincide with the Flaschka–Newell bracket on the Sibuya varieties.

5. Surfaces, character varieties, and the classical group double

For a compact oriented surface $D(G)=G\times G$07 with nonempty boundary and a base point on the boundary, Massuyeau and Turaev construct a natural structure of quasi-Poisson double algebra on the group algebra $D(G)=G\times G$08. The construction begins with the homotopy intersection Fox pairing $D(G)=G\times G$09, whose skew-symmetrization $D(G)=G\times G$10 yields a double bracket. On group elements $D(G)=G\times G$11, represented by transverse loops $D(G)=G\times G$12, the induced tensor-valued bracket is

$D(G)=G\times G$13

and the quasi-Poisson bracket is

$D(G)=G\times G$14

For each $D(G)=G\times G$15, this induces a canonical quasi-Poisson bracket on the representation algebra of $D(G)=G\times G$16, and the trace map sends $D(G)=G\times G$17 times Goldman’s bracket to the induced bracket on $D(G)=G\times G$18-invariants (Massuyeau et al., 2012).

The boundary loop provides the multiplicative moment map. If $D(G)=G\times G$19 is the loop along the boundary component containing the base point, then $D(G)=G\times G$20 satisfies

$D(G)=G\times G$21

which is exactly the multiplicative moment map identity. Reduction by $D(G)=G\times G$22 recovers the Poisson geometry of the corresponding closed-surface moduli (Massuyeau et al., 2012).

Marked-surface generalizations replace $D(G)=G\times G$23 by the twisted fundamental group of a marked oriented surface $D(G)=G\times G$24 with boundary, where every boundary component contains at least one marked point. The resulting surface algebra $D(G)=G\times G$25 carries a double quasi-Poisson bracket built locally from how paths meet the decoration curves at marked points, and its necklace reduction again reproduces Goldman’s bracket. The same paper constructs induced double brackets on spaces of decorated twisted $D(G)=G\times G$26-, symplectic, and indefinite orthogonal local systems (Gekhtman et al., 2024).

In the Lie-group language, the classical quasi-Poisson double is $D(G)=G\times G$27. Huebschmann’s formulation distinguishes an external double, with $D(G)=G\times G$28-action and momentum map $D(G)=G\times G$29, from the internally fused double, with diagonal $D(G)=G\times G$30-action and group-valued moment map

$D(G)=G\times G$31

The corresponding quasi-Poisson bivector on the external double is

$D(G)=G\times G$32

while the weakly quasi-Hamiltonian $D(G)=G\times G$33-form is

$D(G)=G\times G$34

After fusion, one obtains $D(G)=G\times G$35 and $D(G)=G\times G$36, which serve as the basic building blocks for extended moduli spaces and their reductions to Poisson structures on character varieties (Huebschmann, 2022).

6. Related doubles, cohomology, and integrable systems

A different but closely related meaning is provided by the quasi-Drinfeld double of a Lie quasi-bialgebra $D(G)=G\times G$37. Here the double is the Lie algebra

$D(G)=G\times G$38

equipped with a bracket in which $D(G)=G\times G$39 enters explicitly in the $D(G)=G\times G$40-component and with the canonical ad-invariant pairing

$D(G)=G\times G$41

In Klimčík’s affine quasi-Poisson $D(G)=G\times G$42-duality, this quasi-double organizes the quasi-Poisson data, while the associated left and right Poisson-Lie structures admit ordinary Drinfeld doubles $D(G)=G\times G$43 and $D(G)=G\times G$44 used in the $D(G)=G\times G$45-model description (Klimcik, 2018).

The categorical counterpart arises in the integrability theory of quasi-Poisson quotients. For a quasi-Poisson manifold $D(G)=G\times G$46 with free and proper symmetry, the conormal bundle $D(G)=G\times G$47 to the orbits is a Lie algebroid, and the reduced Poisson manifold $D(G)=G\times G$48 is integrable if and only if $D(G)=G\times G$49 is integrable. In the Poisson-group action case, the gauge Poisson groupoid $D(G)=G\times G$50 is integrated by a double symplectic groupoid

$D(G)=G\times G$51

This construction is explicitly described as a categorical analogue of a “double” in quasi-Poisson geometry (Álvarez, 2020).

Double quasi-Poisson structures also have higher-algebraic and cohomological avatars. Every double quasi-Poisson algebra naturally gives rise to a pre-Calabi–Yau algebra on $D(G)=G\times G$52; unlike the double Poisson case, the even higher multiplications do not vanish, and their coefficients are governed by Bernoulli numbers, with $D(G)=G\times G$53 and $D(G)=G\times G$54 among the first examples (Fernández et al., 2020). Cohomologically, one can define completed double Poisson cohomology without assuming the existence of a noncommutative bivector, and the theory extends to quasi-Poisson and gauged Poisson settings while remaining compatible with representation functors (Fairon et al., 25 Sep 2025).

Finally, quasi-Poisson doubles support concrete Hamiltonian systems. In the noncommutative setting, the 2026 type-$D(G)=G\times G$55 theory yields a modified Kontsevich system on $D(G)=G\times G$56 with Hamiltonian

$D(G)=G\times G$57

and equations

$D(G)=G\times G$58

The system is Hamiltonian, $D(G)=G\times G$59 is a first integral, and the trace functions $D(G)=G\times G$60 commute both on ordinary and on twisted orthogonal or symplectic representation spaces (Arthamonov et al., 22 May 2026). In the Lie-group setting, an extended quasi-Poisson double of $D(G)=G\times G$61, built from the internally fused double $D(G)=G\times G$62 together with $D(G)=G\times G$63 exponentiated quasi-Poisson balls, carries a $D(G)=G\times G$64-parameter pencil of compatible quasi-Poisson bivectors whose reductions produce multi-Hamiltonian degenerate integrable systems and a new real form of the trigonometric spin Ruijsenaars–Schneider model (Fairon et al., 2023).

Taken together, these constructions show that “quasi-Poisson double” is best understood as a structural theme rather than a single object. Across associative algebras, Lie groups, Lie quasi-bialgebras, and symplectic groupoids, the common principle is the same: a doubled object carries a bracket or bivector whose Jacobiator is fixed by canonical symmetry data, and whose reductions recover ordinary Poisson geometry on invariants, character varieties, or quotient moduli.