Orbifold Inertia Stack

• Orbifold Inertia Stack is a derived enhancement that records points with isotropy automorphisms, serving as the receptacle for twisted sectors and age shifts.
• It refines the classical inertia by linking free loop space constructions with Hochschild theory, ensuring correct interactions in derived Deligne–Mumford stacks.
• Sector decompositions indexed by cyclic automorphisms underpin its utility in orbifold cohomology, Riemann–Roch corrections, and moduli classifications.

An orbifold inertia stack is the stack-theoretic object that records points together with isotropy automorphisms. For a Deligne–Mumford stack $X$, the classical inertia stack is

$IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$

and its objects are pairs $(x,g)$ consisting of an object $x$ and an automorphism $g\in \mathrm{Aut}(x)$. In this form, inertia is the basic receptacle for twisted sectors, age shifts, orbifold products, and Riemann–Roch correction terms. In derived geometry, the same idea requires a finer replacement: the orbifold inertia stack $X^{\mathrm{orb}}$ introduced for derived Deligne–Mumford stacks is a derived enhancement of inertia that is designed to retain the correct stacky information and to interact correctly with free loop spaces and Hochschild theory (Lim et al., 2021, Fu et al., 30 Aug 2025).

1. Classical inertia and twisted sectors

The classical inertia stack of a Deligne–Mumford stack $\mathcal S$ may be presented either as the self-intersection of the diagonal,

$$I\mathcal{S}=\mathcal{S}\times_{\Delta,\;\mathcal{S}\times \mathcal{S},\;\Delta}\mathcal{S},$$

or concretely as the stack whose objects are pairs $(x,g)$, where $x$ is an object of $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$0 and $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$1 is an automorphism of $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$2. This formulation makes explicit that inertia is not merely the underlying point-set geometry of $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$3, but the geometry together with isotropy data (Lim et al., 2021).

A complementary description decomposes inertia by cyclotomic order. For a stack $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$4,

$IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$5

where $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$6 parametrizes pairs $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$7 with $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$8 and an injective homomorphism $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$9. In this decomposition, $(x,g)$0 is the untwisted sector, while the union of the connected components with $(x,g)$1 is the twisted part (Pagani, 2010, Pagani et al., 2011).

For quotient stacks, inertia admits the familiar fixed-locus decomposition. If $(x,g)$2 with $(x,g)$3 finite, then

$(x,g)$4

hence

$(x,g)$5

The connected components are the twisted sectors, indexed by conjugacy classes of stabilizer elements and realized geometrically by fixed loci modulo centralizers (Lim et al., 2021).

This classical structure is the standard orbifold enhancement of a stack. It replaces the ordinary space of points by a space of points-with-symmetry, and that replacement is exactly what later enters orbifold cohomology, orbifold Gromov–Witten theory, and Grothendieck–Riemann–Roch.

2. Derived Deligne–Mumford stacks and the orbifold inertia stack

For derived Deligne–Mumford stacks, the classical inertia stack is not the correct derived object. The 2025 Hochschild–Kostant–Rosenberg theorem for derived DM stacks introduces the orbifold inertia stack precisely to remedy this failure. The starting point is the $(x,g)$6-th orbifold inertia stack

$(x,g)$7

where $(x,g)$8 is the cyclic group of order $(x,g)$9. The orbifold inertia stack itself is then defined by

$x$0

equivalently

$x$1

This construction is presented as a derived enhancement of the classical inertia stack, but explicitly not as the classical truncation of the free loop space (Fu et al., 30 Aug 2025).

The distinction from the free loop space is essential. For a classical underived stack $x$2, one has

$x$3

and if $x$4 is underived then $x$5 agrees with the classical inertia stack. For a genuinely derived DM stack, however, $x$6 is designed so that one still sees $x$7 itself as a connected component, whereas this would generally fail if one simply took $x$8. The orbifold inertia stack is therefore a finely tuned derived enhancement of inertia rather than a truncation of loops (Fu et al., 30 Aug 2025).

There are natural maps

$x$9

induced by the maps $g\in \mathrm{Aut}(x)$0 and $g\in \mathrm{Aut}(x)$1. These maps exhibit orbifold inertia as a geometrically meaningful subobject of the loop theory of a derived DM stack, but not as a naive replacement for it (Fu et al., 30 Aug 2025).

3. Loop spaces, HKR, and the stacky filtered circle

In characteristic $g\in \mathrm{Aut}(x)$2, the central geometric statement is that the shifted tangent bundle of the orbifold inertia stack recovers the free loop space: $g\in \mathrm{Aut}(x)$3 The paper presents this as the derived-DM analogue of the Ben-Zvi–Nadler exponential map $g\in \mathrm{Aut}(x)$4, with the important difference that in the DM setting it is a genuine equivalence rather than only a formal-completion statement (Fu et al., 30 Aug 2025).

This equivalence yields the multiplicative Hochschild–Kostant–Rosenberg theorem. The Hochschild homology of a derived DM stack $g\in \mathrm{Aut}(x)$5 is identified with functions on the free loop space,

$g\in \mathrm{Aut}(x)$6

and hence, in characteristic $g\in \mathrm{Aut}(x)$7,

$g\in \mathrm{Aut}(x)$8

The identification is an isomorphism of graded algebras, with multiplication on the right induced by the symmetric algebra structure on $g\in \mathrm{Aut}(x)$9. The corresponding Hochschild cohomology statement is

$X^{\mathrm{orb}}$0

and in the finite type lci case,

$X^{\mathrm{orb}}$1

For smooth DM stacks this recovers decompositions indexed by connected components of the inertia stack, while in the global quotient case it becomes a sum over conjugacy classes (Fu et al., 30 Aug 2025).

The filtered version of the theory is organized by the stacky filtered circle. The construction defines

$X^{\mathrm{orb}}$2

Its generic fiber is $X^{\mathrm{orb}}$3, and its central fiber is $X^{\mathrm{orb}}$4. Passing to global functions gives a natural filtration on $X^{\mathrm{orb}}$5, with associated graded

$X^{\mathrm{orb}}$6

In characteristic $X^{\mathrm{orb}}$7 the filtration collapses and recovers the HKR isomorphism; in mixed or positive characteristic it remains nontrivial and interpolates between Hochschild homology and de Rham theory (Fu et al., 30 Aug 2025).

4. Sector decompositions and representative examples

For a global quotient stack $X^{\mathrm{orb}}$8 with $X^{\mathrm{orb}}$9 finite, the orbifold inertia decomposes as

$\mathcal S$0

where $\mathcal S$1 is the genuine fixed locus and $\mathcal S$2 is the centralizer. In this setting, the derived HKR theorem recovers the Arinkin–Căldăraru–Hablicsek formula

$\mathcal S$3

The same discussion emphasizes that the genuine fixed locus is the correct derived notion, distinguished from a naive derived fixed locus, and for a finite-order automorphism $\mathcal S$4 one has

$\mathcal S$5

The theorem also applies beyond global quotients, including weighted projective lines such as “Thurston’s football” and “teardrop” (Fu et al., 30 Aug 2025).

In moduli theory, inertia organizes components by automorphism type. For $\mathcal S$6, twisted sectors are classified by smooth genus-$\mathcal S$7 curves with a finite-order automorphism preserving an unordered set of $\mathcal S$8 marked points; the resulting sectors are expressed in terms of quotients of genus-$\mathcal S$9 moduli spaces. This analysis controls the inertia of the hyperelliptic moduli stack $$I\mathcal{S}=\mathcal{S}\times_{\Delta,\;\mathcal{S}\times \mathcal{S},\;\Delta}\mathcal{S},$$0, where automorphisms of the quotient $$I\mathcal{S}=\mathcal{S}\times_{\Delta,\;\mathcal{S}\times \mathcal{S},\;\Delta}\mathcal{S},$$1 must be lifted through the hyperelliptic double cover (Pagani, 2010).

A similar mechanism governs the inertia of $$I\mathcal{S}=\mathcal{S}\times_{\Delta,\;\mathcal{S}\times \mathcal{S},\;\Delta}\mathcal{S},$$2. There, points of the inertia stack are smooth genus-$$I\mathcal{S}=\mathcal{S}\times_{\Delta,\;\mathcal{S}\times \mathcal{S},\;\Delta}\mathcal{S},$$3 curves equipped with cyclic automorphisms, and the connected components of $$I\mathcal{S}=\mathcal{S}\times_{\Delta,\;\mathcal{S}\times \mathcal{S},\;\Delta}\mathcal{S},$$4 are identified with moduli stacks $$I\mathcal{S}=\mathcal{S}\times_{\Delta,\;\mathcal{S}\times \mathcal{S},\;\Delta}\mathcal{S},$$5 of cyclic covers indexed by admissible data

$$I\mathcal{S}=\mathcal{S}\times_{\Delta,\;\mathcal{S}\times \mathcal{S},\;\Delta}\mathcal{S},$$6

subject to the Riemann–Hurwitz condition and a congruence condition. This converts the classification of twisted sectors into the geometry of cyclic covers of lower-genus curves (Pagani et al., 2011).

These examples show that orbifold inertia is simultaneously local and global: local because it is built from stabilizers, global because its connected components often parametrize moduli of objects carrying specified symmetry.

5. Cohomological, enumerative, and product structures

In orbifold Gromov–Witten theory, the inertia stack is the state space and the target of evaluation maps. For a smooth Deligne–Mumford stack $$I\mathcal{S}=\mathcal{S}\times_{\Delta,\;\mathcal{S}\times \mathcal{S},\;\Delta}\mathcal{S},$$7,

$$I\mathcal{S}=\mathcal{S}\times_{\Delta,\;\mathcal{S}\times \mathcal{S},\;\Delta}\mathcal{S},$$8

with grading shifted by the age of each twisted sector. Stable maps to $$I\mathcal{S}=\mathcal{S}\times_{\Delta,\;\mathcal{S}\times \mathcal{S},\;\Delta}\mathcal{S},$$9 carry evaluation maps

$(x,g)$0

so insertions in orbifold Gromov–Witten invariants live naturally on $(x,g)$1, not on $(x,g)$2 itself. In root-stack and gerbe constructions, the change in Gromov–Witten theory is described precisely through the way the inertia stack acquires or reorganizes twisted sectors (Tseng, 2017).

The same principle persists in higher $(x,g)$3-theory and motivic cohomology. For a quotient orbifold $(x,g)$4, the inertia stack is

$(x,g)$5

and the orbifold products are defined on inertia, double inertia, and triple inertia through pull-push constructions with obstruction classes. In the finite-group case the product on stringy $(x,g)$6-theory is

$(x,g)$7

while the higher Chow product is

$(x,g)$8

In the general quotient-stack setting these formulas are expressed using the twisted pullback obstruction bundle $(x,g)$9 on the inertia stack (Fu et al., 2018).

Orbifold localization on smooth GKM stacks is likewise formulated in inertia-theoretic terms. The inertia stack is decomposed into connected components $x$0, age shifts enter the virtual dimension formula, and evaluation maps of twisted stable maps land in $x$1. The stacky GKM graph records inertia groups at fixed points and along one-dimensional orbits, together with injective maps from edge stabilizers into vertex stabilizers, thereby encoding the sector data needed for localization (Liu et al., 2018).

Grothendieck–Riemann–Roch for DM stacks is another canonical appearance of inertia. Toën’s theorem replaces the ordinary cohomology target by cohomology of the inertia stack,

$x$2

and for $x$3 yields

$x$4

with

$x$5

The correction term is therefore exactly the contribution of the nontrivial twisted sectors (Lim et al., 2021).

6. Mapping-stack formulations, cyclification, and structural extensions

The inertia stack has an intrinsic mapping-stack interpretation. In the smooth higher-stack framework,

$x$6

is the inertia stack, while the smooth loop stack is $x$7. For a good orbifold $x$8, this gives the classical decomposition

$x$9

Cyclification then adds the rotation quotient: $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$00 In this picture, the inertia stack is the essentially constant-loop part of the loop theory, while cyclification is the homotopy quotient by loop rotation (Sati et al., 2022).

At the level of differentiable stacks, this mapping interpretation has a concrete groupoid model. For a compact manifold $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$01 and a Lie groupoid $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$02, the Hom-stack $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$03 is presented by a Fréchet-Lie groupoid $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$04. If $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$05 presents an orbifold, then $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$06 is proper étale and hence presents an infinite-dimensional orbifold. The inertia-type case arises when $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$07, so the loop or inertia mapping stack inherits an orbifold groupoid presentation (Roberts et al., 2016).

In birational and logarithmic geometry, inertia also controls how much stack structure remains after coarsening. For toroidal Deligne–Mumford stacks with finite diagonalizable inertia, the maximal toroidal coarsening $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$08 is characterized by the relative inertia $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$09, whose fibers are the toroidal stabilizers. In the destackification setting, the basic exact sequence

$IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$10

measures how stabilizers change under morphisms, and blow-ups or rigidifications are designed to make this inertia simpler, diagonalizable, or abelian (Abramovich et al., 2017, 1905.00872).

Arithmetic and orbifold variants exhibit the same stabilizer package in a different guise. For Campana-type quotient constructions, a relative inertia computation gives

$IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$11

displaying an identity sector together with finite sectors supported on divisor strata. For hyperbolic Deligne–Mumford curves, closed-point inertia groups are characterized group-theoretically as maximal finite nontrivial closed subgroups of the geometric fundamental group, while the generic inertia is their intersection and the unique maximal finite closed normal subgroup (Bartsch et al., 30 Mar 2026, Collas et al., 4 May 2026).

Taken together, these formulations show that the orbifold inertia stack is not a specialized auxiliary construction. It is the mechanism by which stacky isotropy becomes geometric: in classical orbifold theory as twisted sectors, in derived geometry as the correct enhancement of inertia for HKR, in mapping-stack language as loops of type $IX=X\times_{\Delta,\;X\times X,\;\Delta}X,$12, and in birational or arithmetic settings as the residual stabilizer structure that survives coarsening, rigidification, or group-theoretic reconstruction.