Dual Amplituhedron: Geometric Structure in N=4 SYM

• The dual amplituhedron is defined by minimal positivity (zero sign-flip) conditions in the Grassmannian, encoding scattering amplitudes as positive volumes in configuration space.
• It offers a combinatorial and topological framework with polyhedral interpretations and parity duality via twist maps, linking convex duality in the Plücker embedding.
• This construction bridges geometric, algebraic, and physical properties in N=4 SYM theory, providing insights for exploring multi-loop and worldsheet interpretations.

The dual amplituhedron is a geometric avatar of scattering amplitudes in planar $\mathcal{N}=4$ supersymmetric Yang-Mills theory, defined by replacing the maximal positivity or winding/top-cell condition (characteristic of the original amplituhedron) with a minimal positivity (minimal winding, or zero sign-flip) condition. It provides a purely bosonic, combinatorial, and topological structure—sometimes with direct polyhedral interpretation—whose canonical differential form encodes the amplitude as the positive volume of a region in configuration or Grassmannian space. The dual amplituhedron appears in several guises: as minimal sign-flip spaces, dual convex bodies in the Plücker embedding, explicit dual regions in momentum twistor or dual-momentum space, and parity-dual or twisted versions of the original amplituhedron itself.

1. Definition and Characterizations

The dual amplituhedron is most fundamentally characterized by minimal winding (or zero sign-flip) conditions in the configuration of external data projected through candidate subspaces. For integer parameters $(n,k,m)$ and even $m$ (physical case $m=4$), given external bosonized momentum-twistor data $Z_1,\ldots, Z_n$ with all ordered $(k+m)\times (k+m)$ minors positive, the dual amplituhedron $𝒜^{dual}_{n,k,m}$ is the locus of $Y\in G(k,k+m)$ (the Grassmannian of $k$-planes) satisfying

• All boundary minors: $\langle Y, i\;i+1\;\cdots\;i+m-1 \rangle > 0$ for all $(n,k,m)$0;
• The minimal possible winding: for all $(n,k,m)$1-dimensional projections, the sequence $(n,k,m)$2 has zero sign flips, i.e., is strictly positive or strictly negative throughout.

Equivalently, the dual region may be defined in terms of convex duality in projective space: it is the intersection of a Grassmannian (or positive configuration space) with the convex dual to the convex hull of configuration points embedded (the "exterior cyclic polytope"), or as a locus in momentum-twistor or dual-momentum space obeying nonnegativity and minimal winding on projected data (Arkani-Hamed et al., 2017, Mazzucchelli et al., 23 Jul 2025, Herrmann et al., 2020, Ferro et al., 2016).

For $(n,k,m)$3, the original amplituhedron is a cyclic polytope, and its dual is the projective dual polytope; for higher $(n,k,m)$4 and $(n,k,m)$5, explicit "twist" constructions map the amplituhedron to its dual (Mazzucchelli et al., 23 Jul 2025, Galashin et al., 2018).

2. Mathematical Structures and Duality

In the Plücker embedding, the amplituhedron $(n,k,m)$6, a semialgebraic subset of $(n,k,m)$7, is associated with its convex hull $(n,k,m)$8. The dual amplituhedron arises as the intersection of $(n,k,m)$9 with the convex dual $m$0:

$m$1

For $m$2, the dual amplituhedron is again an amplituhedron, now with twisted data $m$3, where $m$4 is the Marsh–Scott or Plücker twist map. This construction exploits the extendably convex nature of the amplituhedron in this case, and the dual is again cut out by explicit inequalities in Plücker coordinates, corresponding to Schubert-type inequalities tied to the twisted data (Mazzucchelli et al., 23 Jul 2025, Galashin et al., 2018).

For $m$5 or $m$6 (no loops), the canonical form of the amplituhedron is the volume of its dual polytope; for general $m$7 and $m$8, it is conjectured that an analog exists, possibly involving pushforward/pullback of canonical forms and parity-dual triangulations (Arkani-Hamed et al., 2014, Ferro et al., 2015).

3. Canonical Forms, Positivity, and Volume

The differential form assigned to the dual amplituhedron is uniquely characterized by having only logarithmic singularities along (and only along) its boundaries and is invariant under GL symmetries. For $m$9, this is the well-understood volume form (e.g., dlog of side variables for polygons, GKZ hypergeometric for higher $m=4$0), with the geometric formula:

$m=4$1

for $m=4$2. This representation realizes the canonical form as the projective volume of the dual region.

For higher $m=4$3, one expects a similar integral over a dual Grassmannian (possibly $m=4$4), but the explicit measure is not known except in trivial cases. Positivity of the canonical form in the dual region is a hallmark, making "amplitudes" manifestly positive in all explicit cases (Arkani-Hamed et al., 2014, Ferro et al., 2015, Ferro et al., 2016).

The dual amplituhedron also clarifies the structure of "spurious" singularities: the co-dimension one vanishing locus of the numerator in the canonical rational function is conjecturally the boundary of the dual amplituhedron, separating the positive (amplituhedral) region from the outside (Arkani-Hamed et al., 2014).

4. Combinatorial and Topological Descriptions

The dual amplituhedron's construction is fundamentally combinatorial: it is defined by (minimal) winding numbers, i.e., global topological invariants of projected point configurations, or, equivalently, by requiring that all 1-dimensional projections of the data have no sign flips (the "binary code"). This description seamlessly encodes locality (only adjacent minors may vanish) and unitarity (factorization upon vanishing of certain minors) (Arkani-Hamed et al., 2017).

The dual amplituhedron can be triangulated into cells labeled by 0-flip binary patterns, each cell corresponding to a geometric region parametrized positively, with the canonical form expressed as a sum of wedge products of dlogs. These combinatorial features are intimately related to BCFW-type triangulations and positroid cell decompositions in Grassmannian geometry (Galashin et al., 2018).

Parity duality, formalized by the twist map, exchanges the amplituhedron and its dual by mapping affine permutations $m=4$5 labeling triangulations to their inverses $m=4$6, establishing a combinatorial and geometric symmetry between the kinematic sectors $m=4$7.

5. Explicit Realizations at Loop Level and Relations to Momentum Amplituhedra

At loop level, the dual (or momentum) amplituhedron can also be given an explicit polytope-like realization in dual-momentum space. In four-dimensional split signature $m=4$8, the one-loop fiber over a given null polygon is a curvy convex polytope (in $m=4$9-space) cut out by quadratic inequalities (null separation from kinematic dual points), whose canonical form is obtained as a sum over its box and chiral vertices (Ferro et al., 2023, Ferro et al., 2024).

Internally, the dual amplituhedron admits an intrinsic triangulation by chiral pentagons (one-loop MHV), each cell corresponding to a positive region determined by sign-flip conditions. This structure generalizes to higher loops via iterated fibrations of fibers ("fibration of fibration"), with canonical forms decomposed over these fibers and their combinatorial chambers, governed by combinatorics of critical sign patterns and permutation classes (Herrmann et al., 2020, Ferro et al., 2024).

The momentum amplituhedron framework, including dual-momentum configurations, provides a unifying geometric description of scattering amplitudes, naturally yielding prescriptive unitarity: i.e., the canonical forms constructed on the dual space have residues precisely matching physical unitarity cuts, with positive combinatorial decomposition at any multiplicity (Ferro et al., 2023).

6. Parity Duality, Twist Maps, and the Exterior Cyclic Polytope

For $Z_1,\ldots, Z_n$0, and more generally for even $Z_1,\ldots, Z_n$1, deep algebraic and geometric dualities connect the amplituhedron and its dual. The exterior cyclic polytope construction and its dual in Plücker space underlie the explicit realization of the dual region. The twist map converts external data $Z_1,\ldots, Z_n$2 to "twisted" data $Z_1,\ldots, Z_n$3, and the dual amplituhedron is precisely the region corresponding to the twisted data:

$Z_1,\ldots, Z_n$4

with $Z_1,\ldots, Z_n$5 (Mazzucchelli et al., 23 Jul 2025, Galashin et al., 2018).

Parity duality is further reflected at the level of canonical forms. Pullbacks under the twist map map the canonical form of $Z_1,\ldots, Z_n$6 to $Z_1,\ldots, Z_n$7 the form of $Z_1,\ldots, Z_n$8, respecting the triangulation structure given by positroid cells and triangulations labeled by affine permutations and their inverses (Galashin et al., 2018).

7. Open Problems and Future Directions

While explicit descriptions and dual-volume formulas exist for $Z_1,\ldots, Z_n$9, $(k+m)\times (k+m)$0 ("tree NMHV"), and special cases ($(k+m)\times (k+m)$1), the general dual amplituhedron for $(k+m)\times (k+m)$2 and/or $(k+m)\times (k+m)$3 (multi-loop) is not known in closed form. Conjecturally, it is given by a convex body in Grassmannian space with explicit boundaries defined by positivity of certain minors, or as a region selected by minimal winding.

Further research directions include:

• Constructing an explicit dual region for arbitrary $(k+m)\times (k+m)$4, so that the canonical form is, up to normalization, the volume of this dual;
• Understanding the worldsheet or string-theory interpretation of dual amplituhedra, analogous to their role in positive Grassmannian/original amplituhedron approaches;
• Analytic proofs of the new positivity properties and the explicit algebraic characterization of the boundary surfaces separating amplituhedron and non-amplituhedron regions (Arkani-Hamed et al., 2014).

The dual amplituhedron offers a powerful lens for both the combinatorial/topological structure of scattering amplitudes and for the geometric encoding of physical properties such as locality, unitarity, and positivity (Arkani-Hamed et al., 2017, Ferro et al., 2016, Herrmann et al., 2020, Mazzucchelli et al., 23 Jul 2025, Ferro et al., 2015, Ferro et al., 2023, Ferro et al., 2024).