Pfaffian Format in Mathematics and Physics

• Pfaffian Format is a framework where skew-symmetric matrices or differential forms yield a canonical invariant that encodes key algebraic, geometric, and combinatorial data.
• It structures representations in algebraic geometry, combinatorics, and mathematical physics, enabling efficient expression of invariants like Schubert classes, amplitudes, and Lie-theoretic constructs.
• In complexity theory, Pfaffian format quantifies the definitional size of functions and sets, influencing bounds on real solution counts and applications in neural-network robustness.

Pfaffian format denotes a family of constructions in which a skew-symmetric object is taken as primary data and the Pfaffian is used as the organizing invariant. In the literature surveyed here, the term has two main meanings. In algebra, geometry, combinatorics, and mathematical physics, it usually refers to an explicit skew-symmetric matrix, block matrix, or differential-form matrix whose Pfaffian reproduces a polynomial, a central element, an overlap, a Schubert class, or an amplitude contribution. In Pfaffian geometry and o-minimality, it denotes a complexity package attached to Pfaffian chains and the functions or sets defined from them, most commonly the triple $(\alpha,\beta,s)$ or a related structural parameter $\mathcal F$ together with a degree $D$ (Oscari, 2018, Molev, 2011, Bickerton et al., 23 Jun 2026, Binyamini et al., 2020).

1. Algebraic core and terminological scope

At the algebraic level, the Pfaffian is attached to an even-dimensional skew-symmetric matrix. For a skew-symmetric matrix $M$, several papers in this corpus use the standard facts that $\operatorname{Pf}(M)$ is defined by a signed sum over pairings and that $\det(M)=\operatorname{Pf}(M)^2$ (Aluffi et al., 2019, Oscari, 2018). This is the common formal source of all subsequent “Pfaffian formats”: the matrix or differential-form input is chosen so that its skew-symmetry is intrinsic, and the Pfaffian becomes the distinguished scalar or top-degree output.

In the matrix-based usage, “format” means the architecture of the skew-symmetric input together with the ambient algebra in which its entries live. Typical instances are the matrix of affine generators $F[-1]$ for $\widehat{\mathfrak{o}_{2n}}$, a $2d\times 2d$ linear matrix $xA+yB+zC$ for a degree-$\mathcal F$0 plane curve, or a $\mathcal F$1 block matrix built from Bogoliubov data and a basis-overlap matrix in Hartree–Fock–Bogoliubov theory (Molev, 2011, Oscari, 2018, Robledo, 30 Mar 2025).

In the complexity-theoretic usage, the same phrase is detached from a single skew matrix and instead records how a Pfaffian object is defined relative to a Pfaffian chain. For a chain $\mathcal F$2, a Pfaffian function $\mathcal F$3 is said to have format $\mathcal F$4, where $\mathcal F$5 is the order of the chain, $\mathcal F$6 bounds the degrees of the defining differential equations, and $\mathcal F$7 bounds the polynomial degree of $\mathcal F$8 (Bickerton et al., 23 Jun 2026). In restricted sub-Pfaffian geometry, the revised $\mathcal F$9-format of a set is defined from a representation as a union of projections of connected components of semi-Pfaffian sets, while $D$0-degree is the sum of the degrees of those pieces (Binyamini et al., 2020).

A concise way to organize these usages is the following.

Domain	Pfaffian object	Meaning of format
Algebra and geometry	Skew-symmetric matrix or block matrix	Explicit input architecture for $D$1
Differential topology and graph forms	Curvature or graph-valued skew form	Pfaffian top form or transgression form
Pfaffian geometry and o-minimality	Pfaffian chain and derived function/set	Complexity data such as $D$2 or $D$3

This distribution of meanings suggests that “Pfaffian format” is best understood as a structural principle rather than a single definition: skew-symmetry is encoded first, and the Pfaffian then packages the resulting data into a scalar, a top-degree form, or a complexity class.

2. Linear and finite-dimensional matrix realizations

One of the most concrete realizations of Pfaffian format is the linear Pfaffian representation of algebraic hypersurfaces. For a homogeneous polynomial $D$4 of degree $D$5, a linear Pfaffian representation over a commutative ring $D$6 is a skew-symmetric matrix $D$7 of size $D$8 with $D$9. In the plane-curve case this becomes $M$0, and explicit constructions were given for all homogeneous polynomials of degree $M$1, with the representation verified by solving a linear system in the unknown matrix entries (Oscari, 2018). A related result establishes explicit linear Pfaffian $M$2-representations for every homogeneous polynomial in $M$3 of degree at most $M$4, using universal $M$5 skew-symmetric templates (Oscari, 2017).

For cubic surfaces, the same idea takes a more structured form. If $M$6 is cubic, a linear Pfaffian $M$7-representation is a $M$8 skew-symmetric matrix of linear forms with $M$9 for some nonzero scalar $\operatorname{Pf}(M)$0. The constructive algorithm proceeds through five points in general position, a $\operatorname{Pf}(M)$1 skew-symmetric matrix whose principal $\operatorname{Pf}(M)$2 Pfaffians generate their ideal, and a final $\operatorname{Pf}(M)$3 block matrix $\operatorname{Pf}(M)$4 (Tanturri, 2012). Here the Pfaffian format is simultaneously geometric, via the five-point Gorenstein configuration, and algebraic, via Buchsbaum–Eisenbud-type skew-symmetric structure.

A different finite-dimensional use appears in many-body theory. For overlaps of Hartree–Fock–Bogoliubov vacua written in different non-equivalent one-body bases, the overlap is expressed as

$\operatorname{Pf}(M)$5

where $\operatorname{Pf}(M)$6 is the restricted basis-overlap matrix and $\operatorname{Pf}(M)$7 are Thouless matrices (Robledo, 30 Mar 2025). In this setting the Pfaffian format is valued because it gives the overlap including its sign or phase and avoids the square-root ambiguity of determinant formulas.

Across these examples, the common invariant is not merely the presence of a Pfaffian, but the existence of a canonical skew-symmetric linear or block-linear input whose size is dictated by the object being represented.

3. Affine, combinatorial, and symmetric-function constructions

In representation theory, Pfaffian format acquires a canonical Lie-theoretic meaning in type $\operatorname{Pf}(M)$8. For the affine Kac–Moody algebra $\operatorname{Pf}(M)$9, the skew-symmetric matrix $\det(M)=\operatorname{Pf}(M)^2$0 of degree $\det(M)=\operatorname{Pf}(M)^2$1 currents has the property that

$\det(M)=\operatorname{Pf}(M)^2$2

is a Segal–Sugawara vector in the critical-level vacuum module (Molev, 2011). This identifies the missing type-$\det(M)=\operatorname{Pf}(M)^2$3 generator in the explicit construction of the Feigin–Frenkel center and realizes the classical Pfaffian invariant directly at the affine level.

A later combinatorial refinement rewrites the same affine Pfaffian in a mixed Pfaffian–determinant–Pfaffian form indexed by partitions. The top exterior coefficient of $\det(M)=\operatorname{Pf}(M)^2$4 yields $\det(M)=\operatorname{Pf}(M)^2$5, and the resulting formula decomposes the affine object into left and right Pfaffian-type blocks together with a central determinant-type block involving $\det(M)=\operatorname{Pf}(M)^2$6 (Rozhkovskaya, 2013). This form is designed so that the Harish–Chandra image becomes immediate: $\det(M)=\operatorname{Pf}(M)^2$7

In algebraic combinatorics, the same structural idea governs Schur $\det(M)=\operatorname{Pf}(M)^2$8-functions. Okada established Pfaffian analogues of the Cauchy–Binet formula and of the Ishikawa–Wakayama minor-summation formula, turning sums of products of subpfaffians into a single Pfaffian of a block skew-symmetric matrix (Okada, 2017). This provides a direct mechanism for proving identities for Schur $\det(M)=\operatorname{Pf}(M)^2$9-functions, which themselves are naturally indexed by strict partitions and expressed by Pfaffians.

A recent reformulation extends this Pfaffian presentation beyond strict partitions to arbitrary compositions, including negative parts. For a composition $F[-1]$0,

$F[-1]$1

where $F[-1]$2 is made skew-symmetric at the matrix level by setting $F[-1]$3 for $F[-1]$4, $F[-1]$5 on the diagonal, and $F[-1]$6 for $F[-1]$7 (Graf et al., 2024). This removes the need for the ad hoc convention $F[-1]$8, restores identities such as $F[-1]$9, and aligns the Pfaffian construction with tableau and vertex-operator formalisms.

These examples show that in algebra and combinatorics, Pfaffian format often means more than “take the Pfaffian of a matrix”: it is a carefully chosen skew-symmetric packaging that exposes hidden symmetries, simplifies central or character-theoretic calculations, and converts large summations into single Pfaffian identities.

4. Degeneracy loci, Schubert calculus, and Pfaffian systems

In Schubert calculus, Pfaffian format appears as a Giambelli-type mechanism for expressing global classes from special ones. For odd maximal orthogonal Grassmannians, the connective $\widehat{\mathfrak{o}_{2n}}$0-theory class of the degeneracy locus $\widehat{\mathfrak{o}_{2n}}$1 associated with a strict partition $\widehat{\mathfrak{o}_{2n}}$2 is given by a Pfaffian formula whose entries are built from deformed special classes $\widehat{\mathfrak{o}_{2n}}$3: $\widehat{\mathfrak{o}_{2n}}$4 The deformation by factors $\widehat{\mathfrak{o}_{2n}}$5 is a distinctive orthogonal feature arising from the nonreduced quadric intersection $\widehat{\mathfrak{o}_{2n}}$6 (Hudson et al., 2016).

For symplectic Grassmannians, the corresponding equivariant Schubert class is expressed not by a single Pfaffian but by a sum of multi Schur-Pfaffians: $\widehat{\mathfrak{o}_{2n}}$7 Here $\widehat{\mathfrak{o}_{2n}}$8 is determined by the characteristic index $\widehat{\mathfrak{o}_{2n}}$9, and the formula interpolates between determinantal and single-Pfaffian extremes depending on $2d\times 2d$0 (Ikeda et al., 2014). The Pfaffian format therefore acts as an organizing principle for equivariant Giambelli formulas in types $2d\times 2d$1 and $2d\times 2d$2.

A nearby but distinct usage appears in the theory of hypergeometric functions of type $2d\times 2d$3. There the relevant object is a Pfaffian equation

$2d\times 2d$4

derived from the Gauss–Manin connection on a bundle of twisted cohomology groups (Goto et al., 2016). The connection matrix has the explicit logarithmic form

$2d\times 2d$5

with each residue $2d\times 2d$6 expressed through the intersection matrix of twisted cocycles. Although this is not a Pfaffian formula in the matrix-Pfaffian sense, it belongs to the same terminological field: a highly structured first-order system is extracted from geometry and written in a canonical matrix format suitable for contiguity relations and numerical evaluation.

Taken together, these papers show that in intersection-theoretic settings, Pfaffian format is the natural analogue of determinantal format whenever isotropic or skew-symmetric geometry is fundamental.

5. Differential forms, topology, graphs, and amplitudes

A second major branch of the subject replaces skew-symmetric matrices of algebra elements by skew-symmetric matrices of differential forms. For a singular complex projective variety $2d\times 2d$7 with smooth locus $2d\times 2d$8, the curvature of the Levi-Civita connection on $2d\times 2d$9 yields the Pfaffian form $xA+yB+zC$0, and the normalized integral

$xA+yB+zC$1

computes the degree of the zero-dimensional Mather–Chern class: $xA+yB+zC$2 It also admits a stratified Euler-obstruction expansion and, in isolated hypersurface singularities, localizes to Milnor-number expressions (Aluffi et al., 2019). Here Pfaffian format means the Chern–Weil Euler integrand of the underlying real tangent bundle, continued from the smooth to the singular setting.

In odd dimensions, an explicit analogue of the Pfaffian is defined by

$xA+yB+zC$3

a top-degree form on an oriented $xA+yB+zC$4-manifold (Cibotaru et al., 2018). This odd Pfaffian appears as the transgression form for the cone metric and supplies the boundary term in Chern–Gauss–Bonnet formulas for incomplete edge metrics and complete fibered-boundary metrics.

A graph-theoretic version of the same phenomenon was identified in 2025. For a connected graph $xA+yB+zC$5, the “topological form” $xA+yB+zC$6 from one-dimensional topological QFT and the “Pfaffian form” $xA+yB+zC$7 from the odd commutative graph complex satisfy

$xA+yB+zC$8

where $xA+yB+zC$9 is a cycle incidence matrix, $\mathcal F$00 is a path matrix, and $\mathcal F$01 is the loop number (Balduf et al., 12 Mar 2025). The Pfaffian form itself is built from the cycle Laplacian $\mathcal F$02 by

$\mathcal F$03

This exact identification transfers closedness, subdivision compatibility, and Maurer–Cartan-type relations between the two contexts.

In scattering amplitudes, the reduced Pfaffian of the CHY matrix $\mathcal F$04 is reorganized into “Pfaffian diagrams.” The reduced Pfaffian is expanded into an open cycle $\mathcal F$05 and opened branches $\mathcal F$06, producing a root-plus-branches diagrammatic format adapted to direct integration. At four points, the paper reports 19 terms in Pfaffian diagrams versus 35 terms in Feynman diagrams (Lam, 2018). This suggests that Pfaffian format can serve not only as a compact algebraic encoding but also as a reorganization principle that makes gauge-invariant structure more transparent.

6. Complexity-theoretic format for Pfaffian functions and sets

In Pfaffian geometry proper, “format” is an explicit quantitative complexity invariant. A Pfaffian chain on an open set $\mathcal F$07 is a sequence $\mathcal F$08 satisfying triangular polynomial differential equations

$\mathcal F$09

with $\mathcal F$10 of degree at most $\mathcal F$11. A Pfaffian function relative to that chain has the form

$\mathcal F$12

with $\mathcal F$13, and is said to have format $\mathcal F$14 (Bickerton et al., 23 Jun 2026). Khovanskii’s Bézout-type bound then controls the number of isolated real solutions of a Pfaffian system in terms of exactly these parameters.

Recent sharpness results show that these parameters are not merely bookkeeping. For every $\mathcal F$15, there exists a Pfaffian function of format $\mathcal F$16 with at least $\mathcal F$17 nondegenerate real zeros, showing asymptotic sharpness in the chain-degree parameter $\mathcal F$18. For fixed $\mathcal F$19 and $\mathcal F$20, there also exist systems with $\mathcal F$21 regular common zeros, matching the order of growth predicted by Khovanskii’s theorem in the degree parameter $\mathcal F$22 (Bickerton et al., 23 Jun 2026). Thus Pfaffian format is a genuine quantitative complexity invariant rather than a formal label.

The same format governs recent geometric estimates. Tube-volume bounds for smooth Pfaffian hypersurfaces are expressed in terms of $\mathcal F$23, and these bounds are used to derive tail estimates for condition numbers of neural-network classifiers with Pfaffian activation functions. In the special case of single-hidden-layer sigmoid networks with rational first-layer weights, the paper derives polynomial-in-width bounds for the tubular neighborhood of the decision boundary, improving the generic dependence that arises from the Khovanskii factor $\mathcal F$24 (Lezeau et al., 9 Jul 2026). In this setting, the format of the activation function propagates through network composition and becomes a bridge between architecture and robustness estimates.

For restricted sub-Pfaffian sets, a different but related complexity language is used. If a sub-Pfaffian set is represented as

$\mathcal F$25

where each $\mathcal F$26 is a connected component of a semi-Pfaffian set $\mathcal F$27, then the $\mathcal F$28-format of $\mathcal F$29 is the maximum format of the $\mathcal F$30, and the $\mathcal F$31-degree is the sum of their degrees (Binyamini et al., 2020). With this revised notion, cylindrical cell decomposition preserves polynomial dependence on the degree $\mathcal F$32: for a collection of $\mathcal F$33 sub-Pfaffian sets of $\mathcal F$34-format $\mathcal F$35 and $\mathcal F$36-degree $\mathcal F$37, there exists a compatible cell decomposition with $\mathcal F$38 cells, each of $\mathcal F$39-format $\mathcal F$40 and $\mathcal F$41-degree $\mathcal F$42 (Binyamini et al., 2020). One consequence is a polynomial-in-$\mathcal F$43 upper bound for the sum of Betti numbers of sets defined by quantified formulas in the restricted sub-Pfaffian structure.

This complexity-theoretic branch shifts the meaning of Pfaffian format from an explicit skew-symmetric matrix to an effective notion of definitional size. A plausible implication is that the two uses of the term are structurally related: in both cases, Pfaffian format isolates the minimal data needed to control an otherwise large combinatorial expansion.