ScalingStacks

5.4. Dual diagonal action

5.4.1. Algebra

Let BB be a differential algebra endowed with two 22-representations (F1,τ1)(F_{1},\tau_{1}) and (E2,τ2)(E_{2},\tau_{2}) together with a closed morphism λ:F1​E2→E2​F1\lambda:F_{1}E_{2}\to E_{2}F_{1} such that the diagrams (4.2.1) commute.

We define the differential algebra

A=Δλ​(B)=⨁i≥0(E2i​F1i)/((Tr⊗1)​x−(1⊗Tr)​x)x∈E2i​F1i, 1≤r<i.A=\Delta_{\lambda}(B)=\bigoplus_{i\geq 0}(E_{2}^{i}F_{1}^{i})/((T_{r}\otimes 1)x-(1\otimes T_{r})x)_{x\in E_{2}^{i}F_{1}^{i},\ 1\leq r<i}.

Its multiplication is given by the maps μi,j=μ(i,i),(j,j)​E2i​F1i​E2j​F1j→E2i+j​F1i+j\mu_{i,j}=\mu_{(i,i),(j,j)}E_{2}^{i}F_{1}^{i}E_{2}^{j}F_{1}^{j}\to E_{2}^{i+j}F_{1}^{i+j} defined in §4.2.1.

Given MM a differential AA-module and given i≥1i\geq 1, we have differential BB-module maps ςi:E2i​F1i⊗BM→M\varsigma_{i}:E_{2}^{i}F_{1}^{i}\otimes_{B}M\to M. These make (M,(ςi)i)(M,(\varsigma_{i})_{i}) into an object of Δλ​(B​−diff)\Delta_{\lambda}(B\operatorname{\!-diff}\nolimits) and provides an isomorphism of differential categories Δλ​(B)​−diff→∼Δλ​(B​−diff)\Delta_{\lambda}(B)\operatorname{\!-diff}\nolimits\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}\Delta_{\lambda}(B\operatorname{\!-diff}\nolimits).

0P73

Remark 5.4.1. As in Remark 4.4.3, we obtain a lax bi-22-representation on BB by setting Ei,j=E2j​F1iE_{i,j}=E_{2}^{j}F_{1}^{i}. We have an injective morphism of differential algebras ΔE​(B)→Δλ​(B)\Delta_{E}(B)\to\Delta_{\lambda}(B).

Assume the morphisms (5.2.1) are isomorphisms for all ii (this holds for example if λ\lambda is an isomorphism). Then we have a canonical isomorphism ΔE​(B)→∼Δλ​(B)\Delta_{E}(B)\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}\Delta_{\lambda}(B). The algebra Δλ​(B)\Delta_{\lambda}(B) is generated by BB and E2​F1E_{2}F_{1}.

The map λ\lambda extends (uniquely) to a morphism of algebras Δλ′​(B)→Δλ​(B)\Delta^{\prime}_{\lambda}(B)\to\Delta_{\lambda}(B) that is the identity on BB. If λ\lambda is an isomorphism, then this map is an isomorphism Δλ′​(B)→∼Δλ​(B)\Delta^{\prime}_{\lambda}(B)\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}\Delta_{\lambda}(B).

5.4.2. Left dual

We assume now that F1F_{1} is left finite and we put E1=∨F1E_{1}={{}^{\vee}F}_{1}. Consider σ∈Z​Hom⁡(E2​E1,E1​E2)\sigma\in Z\operatorname{Hom}\nolimits(E_{2}E_{1},E_{1}E_{2}) defined as in (4.4.1).

Let π:E2⊗BA→E1⊗BA\pi:E_{2}\otimes_{B}A\to E_{1}\otimes_{B}A be the closed morphism of (B,A)(B,A)-bimodules given as a composition

π:E2⊗BA→E2​η1E2​E1​F1⊗BA→σ​F1E1​E2​E1∨⊗BA→E1​multE1⊗BA.\pi:E_{2}\otimes_{B}A\xrightarrow{E_{2}\eta_{1}}E_{2}E_{1}F_{1}\otimes_{B}A\xrightarrow{\sigma F_{1}}E_{1}E_{2}E_{1}^{\vee}\otimes_{B}A\xrightarrow{E_{1}\mathrm{mult}}E_{1}\otimes_{B}A.

We put E=cone⁡(π)E=\operatorname{cone}\nolimits(\pi). Given i≥1i\geq 1, we define a morphism of (B,B)(B,B)-bimodules ςi:E2i​F1i​E→E\varsigma_{i}:E_{2}^{i}F_{1}^{i}E\to E

ςi=(E2mult∘λ(1⋯2i+1)∑r=1iE2mult∘E2iF1i−1ε1∘λ(1⋯r)(2i⋯i+r)0E1mult∘λ(1⋯2i+1))\varsigma_{i}=\left(\begin{matrix}E_{2}\mathrm{mult}\circ\lambda_{(1\cdots 2i+1)}&\sum_{r=1}^{i}E_{2}\mathrm{mult}\circ E_{2}^{i}F_{1}^{i-1}\varepsilon_{1}\circ\lambda_{(1\cdots r)(2i\cdots i+r)}\\ 0&E_{1}\mathrm{mult}\circ\lambda_{(1\cdots 2i+1)}\end{matrix}\right)

The following lemma is a consequence of Lemmas 4.3.5 and 4.3.7 applied to m=Am=A.

0P74

Lemma 5.4.2. The ςi\varsigma_{i}’s define a left action of AA on EE, giving EE a structure of differential (A,A)(A,A)-bimodule.

Note that the isomorphism of differential categories Δλ​(B)​−diff→∼Δλ​(B​−diff)\Delta_{\lambda}(B)\operatorname{\!-diff}\nolimits\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}\Delta_{\lambda}(B\operatorname{\!-diff}\nolimits) commutes with EE.

Assume now σ\sigma is an isomorphism. We define τ\tau a (B,A)(B,A)-bimodule endomorphism of E2E^{2} as in (5.3.4).

Theorem 4.4.15 has the following consequence.

0P75

Theorem 5.4.3. The data (E,τ)(E,\tau) defines a 22-representation on Δλ​(B)\Delta_{\lambda}(B).

Note that we have an isomorphism of 22-representations Δλ​(B)​−diff→∼Δλ​(B​−diff)\Delta_{\lambda}(B)\operatorname{\!-diff}\nolimits\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}\Delta_{\lambda}(B\operatorname{\!-diff}\nolimits).

Consider the (Δσ​(B),Δλ​(B))(\Delta_{\sigma}(B),\Delta_{\lambda}(B))-bimodule Δλ​(B)\Delta_{\lambda}(B), where the right action is given by multiplication and the left action by multiplication preceded by the morphism of algebras Δσ​(B)=Δλ′​(B)→Δλ​(B)\Delta_{\sigma}(B)=\Delta^{\prime}_{\lambda}(B)\to\Delta_{\lambda}(B). It follows from Proposition 4.4.16 that this bimodule induces a morphism of 22-representations from Δλ​(B)\Delta_{\lambda}(B) to Δσ​(B)\Delta_{\sigma}(B).

Original mathematics by the credited authors. Source collection and HTML conversion remain in progress.

Andrew Manion, Raphael Rouquier

Original source: arXiv:2009.09627v2