Noncommutative Tagged Clusters

• Noncommutative tagged clusters are noncommutative cluster structures on marked surfaces that use tagged triangulations and noncommutative surface algebras.
• They employ geometric relations like triangle, Ptolemy, and bigon rules to establish a noncommutative Laurent phenomenon with positivity.
• Cluster mutations via flips induce braid-group symmetries, bridging tagged triangulations with noncommutative exchange relations in the underlying algebra.

Noncommutative tagged clusters are noncommutative cluster structures attached to marked surfaces, especially punctured surfaces, in which tagged triangulations index cluster data inside a noncommutative surface algebra $A_\Sigma$. In the 2025 formulation, the aim is “to define noncommutative cluster structure on several algebras ${\mathcal A}$ related to marked surfaces possibly with orbifold points of various orders,” and, for punctured surfaces, to “construct new symmetries, noncommutative tagged clusters and establish a noncommutative Laurent Phenomenon” (Berenstein et al., 27 Jul 2025). This framework extends the 2015 construction of Berenstein and Retakh, where a marked surface $\Sigma$ carries a noncommutative algebra ${\mathcal A}_\Sigma$ generated by “noncommutative geodesics” subject to triangle relations and noncommutative analogues of Ptolemy–Plücker relations, with a noncommutative Laurent Phenomenon with respect to any triangulation (Berenstein et al., 2015).

1. Marked surfaces, curves, and tagged triangulations

A marked surface $\Sigma$ in the noncommutative surface formalism is an oriented marked surface with boundary marked points $I_b$ and interior marked points $I_p$, where interior points may be ordinary punctures, special punctures of order $\ge 2$, or “0-punctures” (Berenstein et al., 27 Jul 2025). In the 2015 version, $\Sigma$ is a connected compact $2$-manifold with possibly empty boundary and a finite set of marked points ${\mathcal A}$0, where ${\mathcal A}$1 are “special” punctures, described there as orbifold points of order ${\mathcal A}$2, and ${\mathcal A}$3 is the complement of ordinary punctures (Berenstein et al., 2015).

The basic geometric objects are directed curves ${\mathcal A}$4 connecting marked points, considered up to isotopy or homotopy relative to endpoints. In the 2025 formulation, ${\mathcal A}$5 denotes the set of directed curves connecting marked points “without running into other marked points in their interiors,” and ${\mathcal A}$6 is the subset of curves admissible with respect to special punctures (Berenstein et al., 27 Jul 2025). In the 2015 formulation, ${\mathcal A}$7 denotes equivalence classes of curves from ${\mathcal A}$8 to ${\mathcal A}$9, and there is an involution $\Sigma$0 reversing orientation (Berenstein et al., 2015).

Tagged triangulations enter through the punctured case. In the standard surface-cluster language, a tagged arc is an ordinary arc together with a “plain” or “notched” tag at each end, subject to the conditions that endpoints on $\Sigma$1 are plain, both ends of a loop carry the same tag, and the underlying curve does not cut out a once-punctured monogon (Irelli et al., 2011). A tagged triangulation is then a maximal collection of pairwise-compatible tagged arcs (Irelli et al., 2011). In the noncommutative marked-surface construction, “a tagged triangulation is an ordinary triangulation $\Sigma$2 where at each ordinary puncture one allows to ‘notch’ arcs; in the noncommutative theory one treats special punctures by $\Sigma$3-gons and ordinary punctures by ‘pending arcs’” (Berenstein et al., 2015). This places tagging at the interface between puncture combinatorics and noncommutative exchange.

2. The noncommutative surface algebra

The noncommutative surface algebra $\Sigma$4 is a unital associative algebra generated by invertible symbols attached to directed curves. In the 2025 generalization, $\Sigma$5 is defined over

$\Sigma$6

and is generated by invertible symbols $\Sigma$7 for $\Sigma$8 and $\Sigma$9 for ${\mathcal A}_\Sigma$0, subject to a family of geometric relations (Berenstein et al., 27 Jul 2025). In the earlier version, one fixes a base field ${\mathcal A}_\Sigma$1 of characteristic ${\mathcal A}_\Sigma$2, with generators ${\mathcal A}_\Sigma$3 for each ${\mathcal A}_\Sigma$4, and relations ${\mathcal A}_\Sigma$5 and ${\mathcal A}_\Sigma$6 (Berenstein et al., 2015).

The defining relations encode local configurations in the surface. For every oriented triangle ${\mathcal A}_\Sigma$7 in ${\mathcal A}_\Sigma$8,

${\mathcal A}_\Sigma$9

If $\Sigma$0 is a loop enclosing a special puncture, then $\Sigma$1. If $\Sigma$2 is a pending arc ending at a $\Sigma$3-puncture and $\Sigma$4 is the loop around that puncture, then $\Sigma$5. In any quadrilateral with diagonal $\Sigma$6 and other diagonal $\Sigma$7,

$\Sigma$8

For a bigon around a special puncture $\Sigma$9, one has the additional relation

$I_b$0

These are the triangle, monogon, zero-puncture, Ptolemy, and bigon-around-special-puncture relations of the 2025 theory (Berenstein et al., 27 Jul 2025).

These relations imply in particular that the “noncommutative angles” $I_b$1 are well defined and central to many constructions (Berenstein et al., 27 Jul 2025). In the 2015 formulation, the same geometric mechanism already appeared through triangle relations and noncommutative Ptolemy–Plücker relations, confirming the “cluster nature” of $I_b$2 via Laurentness (Berenstein et al., 2015).

3. Cluster charts, flips, and tagged mutation

A noncommutative cluster in $I_b$3 is not merely a set of generators. It is the data of a triangle group $I_b$4 associated to a triangulation $I_b$5, an embedding $I_b$6 sending $I_b$7, and a cluster braid group $I_b$8 of automorphisms of the object $I_b$9 in the flip-groupoid $I_p$0, generated by elementary flips $I_p$1 at non-pending internal edges $I_p$2 (Berenstein et al., 27 Jul 2025). The triangle group $I_p$3 is defined by generators $I_p$4 with the same triangle and monogon relations as in $I_p$5, but in the group-theoretic setting (Berenstein et al., 27 Jul 2025).

Mutation is realized as flip of an arc. Given triangulations $I_p$6 related by a flip of an internal edge $I_p$7, there is an induced isomorphism

$I_p$8

called “monomial mutation” (Berenstein et al., 27 Jul 2025). Dually, one gets $I_p$9, and these satisfy the same braid-type relations (Berenstein et al., 27 Jul 2025). On the algebra side, the exchange relations in $\ge 2$0 read

$\ge 2$1

together with tagged versions when flips involve self-folded edges or tagging at punctures (Berenstein et al., 27 Jul 2025).

The 2015 treatment describes the same mechanism by saying that the cluster seed attached to $\ge 2$2 has cluster variables $\ge 2$3 for each tagged arc $\ge 2$4, and that “mutation = flip of one arc” (Berenstein et al., 2015). There the replacement of an arc by the opposite diagonal in a quadrilateral is “in complete analogy with standard (commutative) cluster mutation but now in a noncommutative ring” (Berenstein et al., 2015). This suggests that the tagged formalism is not a peripheral refinement but the punctured-surface realization of mutation itself.

4. Laurent expansions and positivity

The central structural result is a noncommutative Laurent Phenomenon. For any tagged triangulations $\ge 2$5 of $\ge 2$6 and any curve $\ge 2$7, the cluster variable $\ge 2$8 in the chart $\ge 2$9 expands in the chart $\Sigma$0 as a finite $\Sigma$1-linear combination of monomial terms in the generators $\Sigma$2 of $\Sigma$3, and “the leading term under the natural partial order is exactly $\Sigma$4” (Berenstein et al., 27 Jul 2025). Concretely,

$\Sigma$5

where the sequence ranges over admissible sequences of edges of $\Sigma$6 crossing $\Sigma$7, the coefficient is a product of suitable $\Sigma$8 factors for special-puncture crossings, and $\Sigma$9 keeps track of orientation (Berenstein et al., 27 Jul 2025).

The 2015 theorem gives an explicit expansion in terms of admissible edge-walks in the dual $2$0-gon of a triangulation. For each curve $2$1, there exists a finite admissible set of edge-walks such that $2$2 is a sum of alternating products of $2$3 and inverses, lying in

$2$4

so each $2$5 is a noncommutative Laurent polynomial in the $2$6-variables (Berenstein et al., 2015). The proof strategy there reduces to the polygon case by cutting along $2$7 and using a universal cover; on the polygon one constructs the “$2$8-factorization” of a chord and reconstructs $2$9 as a sum over factorization paths (Berenstein et al., 2015).

Positivity appears in the applications to discrete noncommutative integrable systems. For an annulus with one outer marked point and ${\mathcal A}$00 inner points, the primitive flips along a band of ${\mathcal A}$01 edges yield a recurrence for variables ${\mathcal A}$02, and one shows by induction on flips that each ${\mathcal A}$03 lies in the group-algebra ${\mathcal A}$04, “hence is a sum of group-elements (positivity)” (Berenstein et al., 2015). The infinite-strip example similarly yields noncommutative Hirota-type systems whose solutions are “again Laurent (and in fact positive) in the initial data,” and, more generally, any discrete noncommutative dynamical system arising from an infinite sequence of flips in a periodic tagged triangulation yields exchange recurrences whose variables remain Laurent-polynomial and even positive sums of monomials in the initial seed (Berenstein et al., 2015).

5. Braid-group symmetries and model examples

A distinctive feature of the theory is the presence of a braid-like symmetry group acting on clusters. The elementary flips ${\mathcal A}$05 act on ${\mathcal A}$06 by explicit automorphisms, and “one checks that the ${\mathcal A}$07 satisfy the usual Artin braid relations corresponding to the intersection pattern of edges in ${\mathcal A}$08” (Berenstein et al., 27 Jul 2025). The resulting cluster braid group ${\mathcal A}$09 is isomorphic to a mapping-class-type braid group attached to ${\mathcal A}$10; for a polygon one recovers ${\mathcal A}$11, and for a once-punctured ${\mathcal A}$12-gon one recovers ${\mathcal A}$13 (Berenstein et al., 27 Jul 2025).

The classification theorem states that for each tagged ${\mathcal A}$14, the cluster braid group sits in an exact sequence

${\mathcal A}$15

and ${\mathcal A}$16 is isomorphic to the Artin braid group of the same Coxeter–Dynkin type as the commutative cluster structure on ${\mathcal A}$17: type ${\mathcal A}$18 for the unpunctured ${\mathcal A}$19-gon, type ${\mathcal A}$20 or ${\mathcal A}$21 for an ${\mathcal A}$22-gon with one orbifold point or one ${\mathcal A}$23-puncture, type ${\mathcal A}$24 for once-punctured ${\mathcal A}$25-gon, and affine type ${\mathcal A}$26 or ${\mathcal A}$27 for cylinders or twice-punctured disks (Berenstein et al., 27 Jul 2025). The concluding formulation is that the assignments ${\mathcal A}$28, ${\mathcal A}$29, ${\mathcal A}$30, and ${\mathcal A}$31 give a “triangular functor” from the flip-groupoid ${\mathcal A}$32 to the group-isomorphism-groupoid ${\mathcal A}$33, realizing the full noncommutative cluster-groupoid (Berenstein et al., 27 Jul 2025).

Examples make the structure concrete. For the once-punctured triangle, identified as type ${\mathcal A}$34, a tagged triangulation has four edges and the triangle group ${\mathcal A}$35 is free of rank ${\mathcal A}$36; the flip at a side gives

${\mathcal A}$37

while the flip at a radius yields the noncommutative tagged exchange

${\mathcal A}$38

In the rank-${\mathcal A}$39 example, where ${\mathcal A}$40 is a bigon with two marked boundary points and no punctures, the triangle group is ${\mathcal A}$41, and flips recover the rank-${\mathcal A}$42 “Kontsevich” recursion ${\mathcal A}$43, with braid-group action generated by ${\mathcal A}$44 and ${\mathcal A}$45 satisfying the relation of the ${\mathcal A}$46 braid group (Berenstein et al., 27 Jul 2025).

6. Relation to adjacent formalisms and further consequences

A plausible implication is that noncommutative tagged clusters should not be conflated with two adjacent but distinct frameworks. One is the theory of tagged triangulations and quivers with potentials for surfaces with marked points and non-empty boundary, where to each tagged triangulation ${\mathcal A}$47 one associates a quiver with potential ${\mathcal A}$48, flips correspond to QP-mutations, and admissibility yields the proper-Laurent property and linear independence of cluster monomials (Irelli et al., 2011). The other is the toric-frame formalism for quantum cluster algebra structures on quantum nilpotent algebras, where a quantum seed is a pair ${\mathcal A}$49 consisting of a toric frame and a compatible exchange matrix, mutation is defined by the usual Fomin–Zelevinsky rule, and for CGL extensions one obtains equality ${\mathcal A}$50 (Goodearl et al., 2013). This suggests that the marked-surface theory occupies a specific geometric branch of noncommutative cluster theory: it is organized by curves, triangulations, flips, and surface braid groups rather than by path-algebra Jacobians or Ore-extension chains.

The 2015 marked-surface paper also records two consequences beyond mutation theory. First, the algebra ${\mathcal A}$51 yields “a new topological invariant of ${\mathcal A}$52, which is a free or a 1-relator group easily computable in terms of any triangulation” (Berenstein et al., 2015). Second, the surface construction proves Laurentness and positivity for “certain discrete noncommutative integrable systems” (Berenstein et al., 2015). Within the later tagged formalism, the noncommutative Laurent Phenomenon and the explicit expansion formulas are said to “generalize both classical and quantum Laurent expansions for surfaces” (Berenstein et al., 27 Jul 2025). In that sense, noncommutative tagged clusters provide a surface-theoretic mechanism that simultaneously encodes puncture-sensitive mutation, braid-group symmetries, Laurent expansions, and positive noncommutative recurrences.