Double of the Orlik–Solomon Algebra

• Double of the Orlik–Solomon algebra is a bi-graded extension that integrates classical OS constructions with geometric or colored data from manifold and bi-arrangements.
• It recovers the cohomology ring of arrangement complements by combining a combinatorial horizontal differential with a geometric vertical one.
• The construction extends to spectral sequences and mixed Hodge complexes, providing insights into relative cohomology motives and examples like multiple zeta values.

Searching arXiv for the cited work on manifold arrangements and bi-arrangements. The double of the Orlik–Solomon algebra refers to a family of bi-graded constructions that extend the classical Orlik–Solomon algebra beyond ordinary hyperplane-complement cohomology. In one direction, for a manifold arrangement $\mathcal{A}$ in a smooth manifold $M$, the generalized Orlik–Solomon algebra with coefficients in a monoidal presheaf of cochain complexes produces a double complex $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$ whose total differential graded algebra recovers the cohomology ring $H^*(\mathcal{M}(\mathcal{A}))$ of the complement $\mathcal{M}(\mathcal{A})$ (Chen et al., 2020). In another direction, for a bi-arrangement $\mathscr{B}$ of hypersurfaces endowed with a coloring, one obtains an Orlik–Solomon bi-complex $A_{\bullet,\bullet}(\mathscr{B})$ whose totalization governs relative cohomology motives and, in exact cases, computes them through an explicit spectral sequence (Dupont, 2014). The shared principle is that the classical Orlik–Solomon differential is supplemented by a second grading and differential reflecting additional geometric data: strata cohomology in the manifold-arrangement setting, and $\lambda/\mu$-colored relative geometry in the bi-arrangement setting.

1. Classical Orlik–Solomon theory as the underlying combinatorial core

For an arrangement $\mathcal{A}=\{K_1,\dots,K_k\}$ in $\mathbb{C}^n$, the classical Orlik–Solomon algebra is constructed from the exterior algebra $M$0 with $M$1 and derivation $M$2 of degree $M$3 defined by $M$4, hence

$M$5

If $M$6 is the ideal generated by $M$7 for dependent sets $M$8, then the Orlik–Solomon algebra is

$M$9

It decomposes over strata and, in the affine case, the Brieskorn–Orlik–Solomon theorem identifies $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$0 with $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$1 (Dupont, 2014).

The manifold-arrangement framework reformulates the same combinatorics on a locally geometric poset $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$2. If $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$3 is the set of atoms, $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$4 is the exterior algebra on generators $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$5, and $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$6 is generated by $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$7 for dependent sets $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$8, then

$A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$9

This algebra is $H^*(\mathcal{M}(\mathcal{A}))$0-graded,

$H^*(\mathcal{M}(\mathcal{A}))$1

and $H^*(\mathcal{M}(\mathcal{A}))$2 descends to $H^*(\mathcal{M}(\mathcal{A}))$3, lowering the lattice grade by one (Chen et al., 2020).

The “double” constructions preserve this OS core rather than replacing it. In both frameworks, the original differential remains one axis of the bi-graded object. What changes is the coefficient system and the interpretation of the second axis.

2. The manifold-arrangement double complex

Let $H^*(\mathcal{M}(\mathcal{A}))$4 be a smooth manifold without boundary and $H^*(\mathcal{M}(\mathcal{A}))$5 a finite family of smooth, closed submanifolds meeting cleanly in the sense of Bott. The associated quasi-intersection poset $H^*(\mathcal{M}(\mathcal{A}))$6 consists of quasi-layers, ordered by reverse inclusion, with minimal element $H^*(\mathcal{M}(\mathcal{A}))$7, and each interval $H^*(\mathcal{M}(\mathcal{A}))$8 is assumed to be a geometric lattice; such $H^*(\mathcal{M}(\mathcal{A}))$9 is called locally geometric (Chen et al., 2020). For $\mathcal{M}(\mathcal{A})$0, one writes $\mathcal{M}(\mathcal{A})$1 for the corresponding quasi-layer and

$\mathcal{M}(\mathcal{A})$2

The complement of the arrangement is

$\mathcal{M}(\mathcal{A})$3

The generalized Orlik–Solomon algebra with coefficients in a presheaf $\mathcal{M}(\mathcal{A})$4 on $\mathcal{M}(\mathcal{A})$5 is

$\mathcal{M}(\mathcal{A})$6

with differential

$\mathcal{M}(\mathcal{A})$7

where $\mathcal{M}(\mathcal{A})$8 in $\mathcal{M}(\mathcal{A})$9 and each $\mathscr{B}$0 covers $\mathscr{B}$1. The construction uses the negative lattice grading

$\mathscr{B}$2

The naive coefficient system $\mathscr{B}$3 is a presheaf of cochain complexes, but its cup product does not in general define a monoidal presheaf. The key modification is the supported-cochain quotient

$\mathscr{B}$4

leading to

$\mathscr{B}$5

The cup product descends to

$\mathscr{B}$6

and for quasi-layers $\mathscr{B}$7,

$\mathscr{B}$8

which makes $\mathscr{B}$9 a monoidal presheaf (Chen et al., 2020).

The double complex then consists of the horizontal OS differential and the vertical cochain differential: $A_{\bullet,\bullet}(\mathscr{B})$0 with

$A_{\bullet,\bullet}(\mathscr{B})$1

Because $A_{\bullet,\bullet}(\mathscr{B})$2, the pair $A_{\bullet,\bullet}(\mathscr{B})$3 is a double complex, and its total complex carries total differential

$A_{\bullet,\bullet}(\mathscr{B})$4

of cohomological degree $A_{\bullet,\bullet}(\mathscr{B})$5 (Chen et al., 2020).

A concise description given in the source is that the construction “doubles” OS by combining local Orlik–Solomon algebras $A_{\bullet,\bullet}(\mathscr{B})$6, a presheaf of cochain complexes recording the topology of the strata, and a multiplication governed by joins $A_{\bullet,\bullet}(\mathscr{B})$7. This identifies the horizontal direction with combinatorics and the vertical direction with stratum topology.

3. Algebra structure, totalization, and recovery of the complement ring

If $A_{\bullet,\bullet}(\mathscr{B})$8 is monoidal, then $A_{\bullet,\bullet}(\mathscr{B})$9 is a graded algebra with product

$\lambda/\mu$0

for $\lambda/\mu$1, $\lambda/\mu$2, $\lambda/\mu$3, and $\lambda/\mu$4. Here $\lambda/\mu$5 is the canonical inclusion and $\lambda/\mu$6 is the projection to the summand $\lambda/\mu$7 (Chen et al., 2020).

For $\lambda/\mu$8 and $\lambda/\mu$9, the horizontal and vertical differentials satisfy Leibniz rules: $\mathcal{A}=\{K_1,\dots,K_k\}$0

$\mathcal{A}=\{K_1,\dots,K_k\}$1

where $\mathcal{A}=\{K_1,\dots,K_k\}$2. Consequently,

$\mathcal{A}=\{K_1,\dots,K_k\}$3

is a differential graded algebra (Chen et al., 2020).

The main structural theorem states that

$\mathcal{A}=\{K_1,\dots,K_k\}$4

as algebras. In the corresponding non-hat model one first obtains a module isomorphism

$\mathcal{A}=\{K_1,\dots,K_k\}$5

via the quotient $\mathcal{A}=\{K_1,\dots,K_k\}$6 sending all summands with $\mathcal{A}=\{K_1,\dots,K_k\}$7 to zero; the hat-model upgrades this to an algebra isomorphism (Chen et al., 2020).

This construction specializes to the classical Orlik–Solomon algebra when the coefficient presheaf is the constant presheaf $\mathcal{A}=\{K_1,\dots,K_k\}$8 concentrated in cochain degree $\mathcal{A}=\{K_1,\dots,K_k\}$9. Then $\mathbb{C}^n$0, and $\mathbb{C}^n$1 reduces to the classical OS algebra with its $\mathbb{C}^n$2-complex. The additional vertical differential is therefore not an auxiliary decoration but the mechanism by which the topology of arrangement strata enters the model.

4. Bi-arrangements and the Orlik–Solomon bi-complex

A second, distinct use of the term “double of the Orlik–Solomon algebra” arises for bi-arrangements. A bi-arrangement of hyperplanes $\mathbb{C}^n$3 in $\mathbb{C}^n$4 consists of an arrangement $\mathbb{C}^n$5 and a coloring of strict strata

$\mathbb{C}^n$6

satisfying the Künneth condition: for any non-trivial decomposition $\mathbb{C}^n$7, one has $\mathbb{C}^n$8 or $\mathbb{C}^n$9. Equivalently, one can write $M$00, where $M$01 and $M$02 are disjoint sub-arrangements and $M$03, $M$04 (Dupont, 2014).

The Orlik–Solomon bi-complex $M$05 is defined by vector spaces $M$06 indexed by strata $M$07, together with maps for inclusions $M$08,

$M$09

uniquely characterized by four conditions: $M$10; for each stratum $M$11, $M$12 is a bi-complex; $M$13-colored strict strata impose exact sequences along rows; and $M$14-colored strict strata impose exact sequences along columns (Dupont, 2014).

The bi-complex identities are encoded by

$M$15

together with mixed commutation $M$16. The latter reflects the incidence geometry of strata: it is expressed through identities associated to diagrams $M$17 and to the unique stratum $M$18 when such a diagram exists (Dupont, 2014).

This bi-complex reduces to the ordinary Orlik–Solomon algebra in the pure-color limits. For $M$19, one has

$M$20

For $M$21,

$M$22

Accordingly, one grading and differential $M$23 reflect the $M$24-side, while the second grading and differential $M$25 reflect the $M$26-side. This is the sense in which the Orlik–Solomon bi-complex is a “double” of OS (Dupont, 2014).

For tame bi-arrangements, the doubling becomes completely explicit. One sets

$M$27

with $M$28 and $M$29, then imposes relations for circuits with $M$30 and co-relations for circuits with $M$31. The theorem states: “All tame bi-arrangements are exact. Furthermore, $M$32 as above is the Orlik–Solomon bi-complex of $M$33” (Dupont, 2014).

5. Spectral sequences, mixed Hodge theory, and motives

In the manifold-arrangement setting, the total DGA carries a natural column filtration

$M$34

This yields a spectral sequence with

$M$35

$M$36

converging to

$M$37

The filtration is multiplicative, and if the $M$38-algebra $M$39 is generated in degrees $M$40 and $M$41, the spectral sequence collapses at $M$42 (Chen et al., 2020).

When $M$43 and all $M$44 are smooth complex algebraic varieties, the construction admits a mixed Hodge refinement. Writing $M$45 for Deligne’s mixed Hodge complex, one defines a presheaf $M$46 as a mixed telescope cone built from the geometric filtration $M$47. Then

$M$48

is a mixed Hodge complex with weight filtration

$M$49

and its cohomology identifies with $M$50 endowed with its canonical mixed Hodge structure (Chen et al., 2020). If $M$51 is projective, $M$52 is pure and

$M$53

Moreover, for projective $M$54 with $M$55-coefficients, the column-wise filtration $M$56 induces the same filtration on cohomology as the weight filtration $M$57 (Chen et al., 2020).

In the bi-arrangement setting, the geometric Orlik–Solomon bi-complex is defined, for a fixed integer $M$58, by

$M$59

with differentials obtained by combining the combinatorial maps $M$60 with Gysin and restriction maps. Its total complex $M$61 is the geometric Orlik–Solomon complex of index $M$62 (Dupont, 2014).

For an exact bi-arrangement of hypersurfaces in $M$63, there is a spectral sequence

$M$64

that is,

$M$65

If $M$66 is algebraic with $M$67 divisors, this is a spectral sequence in mixed Hodge structures; if $M$68 is smooth projective, it degenerates at $M$69, and

$M$70

Thus both theories connect the doubled OS object to mixed Hodge-theoretic filtrations, although one computes ordinary complement cohomology and the other computes relative cohomology motives (Dupont, 2014).

6. Examples, interpretations, and limitations

A primary example in the manifold-arrangement framework is the chromatic configuration space. For a simple graph $M$71 on $M$72,

$M$73

If $M$74 is the bond lattice and $M$75 is the collection of diagonals $M$76 indexed by atoms of $M$77, then the poset of quasi-layers is $M$78 and $M$79. Applying the main theorem gives

$M$80

as algebras, together with a multiplicative spectral sequence whose $M$81 and $M$82 pages are expressed in terms of $M$83 (Chen et al., 2020).

If $M$84 and coefficients are $M$85, the Poincaré polynomial is

$M$86

where $M$87 and $M$88 is the chromatic polynomial. In the special case where the diagonal cohomology class of $M$89 vanishes and $M$90-coefficients are used, the spectral sequence collapses at $M$91, and $M$92 as algebras; under additional cohomological vanishing hypotheses this upgrades to $M$93 itself (Chen et al., 2020).

In the bi-arrangement setting, one of the motivating examples is the computation of motives associated with multiple zeta values. For the multizeta bi-arrangement $M$94 in $M$95, the motive $M$96 is mixed Tate over $M$97, and the integral

$M$98

arises as a period pairing de Rham and Betti classes defined by the bi-arrangement (Dupont, 2014). The $M$99 example is obtained from a blow-up of $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$00 at four corner points to reach a normal crossing configuration, after which the relevant relative cohomology carries the de Rham class of $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$01 and the Betti class of the lifted domain (Dupont, 2014).

Two misconceptions are precluded by the cited constructions. First, the “double” is not a universal single object: in (Chen et al., 2020) it is a double complex $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$02 attached to a locally geometric manifold arrangement, while in (Dupont, 2014) it is an Orlik–Solomon bi-complex $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$03 attached to a colored bi-arrangement. Second, the doubling is not merely formal duplication of generators. In the manifold-arrangement case it incorporates supported cochains and a vertical differential $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$04; in the bi-arrangement case it separates $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$05-type and $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$06-type incidence data through $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$07 and $A^*(\mathfrak{L}, \hat{\mathcal{C}}(\mathcal{A}))$08.

The constructions also have explicit hypotheses. For manifold arrangements, the submanifolds must be smooth, closed, and meet cleanly, and the quasi-intersection poset must be locally geometric; the coefficient ring is a commutative ring with unit, while mixed Hodge statements require smooth complex algebraic varieties and projectivity for purity (Chen et al., 2020). For bi-arrangements, exactness is a central condition for the spectral sequence computing the motive, and it is preserved under blowups of good strata; tame bi-arrangements form a subclass where exactness and explicit presentations are available (Dupont, 2014).

Taken together, these two theories establish a precise mathematical meaning for the double of the Orlik–Solomon algebra: a bi-graded extension of the classical OS formalism in which one differential remains combinatorial and the second records geometric or relative information. In one case the outcome is a DGA model for the cohomology ring of a complement; in the other it is a bi-complex computing relative cohomology motives and their weight-graded structure.