This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

An Adjunction Between Boolean Algebras and a Subcategory of Stone Algebras

Inigo Incer
( Computing and Mathematical Sciences
California Institute of Technology
[email protected] )
Abstract

We consider Stone algebras with a distinguished element ee satisfying the identity ex=¬¬xe\to x=\neg\neg x for all elements xx of the algebra. We provide an adjunction between the category of such algebras and that of Boolean algebras.

1 Introduction

The algebra of contracts [1, 2] has been an object of attention in computer science and engineering due to its capability to support compositional design. Given a Boolean algebra BB, its contract algebra 𝐂(B)\mathbf{C}(B) has elements (a,g)B2(a,g)\in B^{2} such that ag=1Ba\lor g=1_{B}. 𝐂(B)\mathbf{C}(B) is a Stone algebra with operations (a,g)(a,g)=(aa,gg)(a,g)\land(a^{\prime},g^{\prime})=(a\lor a^{\prime},g\land g^{\prime}), (a,g)(a,g)=(aa,gg)(a,g)\lor(a^{\prime},g^{\prime})=(a\land a^{\prime},g\lor g^{\prime}), and (a,g)(a,g)=((a¬a)(g¬g),¬gg)(a^{\prime},g^{\prime})\to(a,g)=((a\land\neg a^{\prime})\lor(g^{\prime}\land\neg g),\neg g^{\prime}\lor g). The top and bottom elements of this algebra are, respectively, 1=(0B,1B)1=(0_{B},1_{B}) and 0=(1B,0B)0=(1_{B},0_{B}). This note introduces a subcategory of Stone algebras with the property that the contract functor is the left adjoint of the functor that maps these algebras to their closures.

2 Augmented Stone Algebras

First we recall the definition of Stone algebras. Then we introduce the concept of an augmented Stone algebra.

Definition 2.1.

A Stone algebra SS is a Heyting algebra satisfying ¬xClos(x)=1S\neg x\lor\mathrm{Clos}(x)=1_{S} for all xSx\in S, where ¬x=x0\neg x=x\to 0 and Clos(x)\mathrm{Clos}(x) is the closure of xx: Clos(x)=¬¬x\mathrm{Clos}(x)=\neg\neg x. We say that xx is closed if Clos(x)=x\mathrm{Clos}(x)=x. We say xx is dense if Clos(x)=1S\mathrm{Clos}(x)=1_{S}.

Definition 2.2.

We say (S,,,,1,0,e)(S,\land,\lor,\to,1,0,e) is an augmented Stone algebra if (S,,,,1,0)(S,\land,\lor,\to,1,0) is a Stone algebra and eSe\in S, called the closure element, satisfies ex=Clos(x)e\to x=\mathrm{Clos}(x) for all xSx\in S.

Notation 2.3.

Let 𝐁𝐨𝐨𝐥\mathbf{Bool} be the category of Boolean algebras, and 𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞\mathbf{augStone} that of augmented Stone algebras. We define Clos:𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞𝐁𝐨𝐨𝐥\mathrm{Clos}\colon\mathbf{augStone}\to\mathbf{Bool} as the functor that maps an extended Stone algebra to its Boolean algebra of closed elements.

Example 2.4.

Any Boolean algebra BB is an augmented Stone algebra with eB=1Be_{B}=1_{B}. 𝐂(B)\mathbf{C}(B) is an augmented Stone algebra with e=(1B,1B)e=(1_{B},1_{B}).

Let us prove some elementary facts about augmented Stone algebras.

Proposition 2.5.

Let SS be an augmented Stone algebra.

  1. (i)

    Clos(e)=1\mathrm{Clos}(e)=1.

  2. (ii)

    The closure element is unique.

  3. (iii)

    yey\to e is dense for all ySy\in S.

  4. (iv)

    If xSx\in S is closed, xy=¬xyx\to y=\neg x\lor y.

Proof.
  1. (i)

    1=ee=Clos(e)1=e\to e=\mathrm{Clos}(e).

  2. (ii)

    Suppose ee and ee^{\prime} are closure elements. Then ee=Clos(e)=1e\to e^{\prime}=\mathrm{Clos}(e^{\prime})=1, whence eee\leq e^{\prime}. By the same reasoning, eee^{\prime}\leq e, and so e=ee=e^{\prime}.

  3. (iii)

    Clos(ye)=e(ye)=y(ee)=1\mathrm{Clos}(y\to e)=e\to(y\to e)=y\to(e\to e)=1.

  4. (iv)

    For aSa\in S, we have axyaxy(ax)¬xy¬x(ax)¬x(¬xa)y¬xay¬xa\leq x\to y\Leftrightarrow a\land x\leq y\Leftrightarrow(a\land x)\lor\neg x\leq y\lor\neg x\Leftrightarrow(a\land x)\lor\neg x\lor(\neg x\land a)\leq y\lor\neg x\Leftrightarrow a\leq y\lor\neg x.∎

Proposition 2.6.

𝐂\mathbf{C} is a functor 𝐂:𝐁𝐨𝐨𝐥𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞\mathbf{C}\colon\mathbf{Bool}\to\mathbf{augStone}.

Proof.

Given B𝐎𝐛𝐣(𝐁𝐨𝐨𝐥)B\in\mathbf{Obj}(\mathbf{Bool}), we know that 𝐂(B)𝐎𝐛𝐣(𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞)\mathbf{C}(B)\in\mathbf{Obj}(\mathbf{augStone}) from Section 1 and Example 2.4. Given a map fhom𝐁𝐨𝐨𝐥(B,B)f\in\hom_{\mathbf{Bool}}(B,B^{\prime}), we have 𝐂(f)(a,g)=(fa,fg)𝐂(B)\mathbf{C}(f)(a,g)=(fa,fg)\in\mathbf{C}(B^{\prime}) as f(a)f(g)=f(ag)=1Bf(a)\lor f(g)=f(a\lor g)=1_{B^{\prime}}. Moreover, 𝐂(f)(1)=1\mathbf{C}(f)(1)=1, 𝐂(f)(0)=0\mathbf{C}(f)(0)=0, and 𝐂(f)(e)=e\mathbf{C}(f)(e)=e. We also have

𝐂(f)((a,g)(a,g))\displaystyle\mathbf{C}(f)((a,g)\land(a^{\prime},g^{\prime})) =𝐂(f)((aa,gg))=((f(a)f(a),f(g)f(g)))\displaystyle=\mathbf{C}(f)((a\lor a^{\prime},g\land g^{\prime}))=((f(a)\lor f(a^{\prime}),f(g)\land f(g^{\prime})))
=𝐂(f)(a,g)𝐂(f)(a,g),\displaystyle=\mathbf{C}(f)(a,g)\land\mathbf{C}(f)(a^{\prime},g^{\prime}),
𝐂(f)((a,g)(a,g))\displaystyle\mathbf{C}(f)((a,g)\lor(a^{\prime},g^{\prime})) =𝐂(f)((aa,gg))=((f(a)f(a),f(g)f(g)))\displaystyle=\mathbf{C}(f)((a\land a^{\prime},g\lor g^{\prime}))=((f(a)\land f(a^{\prime}),f(g)\lor f(g^{\prime})))
=𝐂(f)(a,g)𝐂(f)(a,g),\displaystyle=\mathbf{C}(f)(a,g)\lor\mathbf{C}(f)(a^{\prime},g^{\prime}),
and
𝐂(f)((a,g)(a,g))\displaystyle\mathbf{C}(f)((a^{\prime},g^{\prime})\to(a,g)) =𝐂(f)(((gg)(a¬a),gg))\displaystyle=\mathbf{C}(f)(((g^{\prime}\to g)\to(a\land\neg a^{\prime}),g\to g^{\prime}))
=(((fgfg)(fa¬fa),fgfg))\displaystyle=(((fg^{\prime}\to fg)\to(fa\land\neg fa^{\prime}),fg\to fg^{\prime}))
=𝐂(f)(a,g)𝐂(f)(a,g).\displaystyle=\mathbf{C}(f)(a^{\prime},g^{\prime})\to\mathbf{C}(f)(a,g).\qed

3 An adjunction between 𝐁𝐨𝐨𝐥\mathbf{Bool} and 𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞\mathbf{augStone}

Let B𝐎𝐛𝐣(𝐁𝐨𝐨𝐥)B\in\mathbf{Obj}(\mathbf{Bool}) and S𝐎𝐛𝐣(𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞)S\in\mathbf{Obj}(\mathbf{augStone}). For (a,g)𝐂(B)(a,g)\in\mathbf{C}(B), we let π1(a,g)=a\pi_{1}(a,g)=a and π2(a,g)=g\pi_{2}(a,g)=g, and for xBx\in B, we let Δx=(¬x,x)\Delta x=(\neg x,x).

Proposition 3.1.

Let fhom𝐁𝐨𝐨𝐥(B,Clos(S))f\in\hom_{\mathbf{Bool}}(B,\mathrm{Clos}(S)). The assignment

αB,S:ffπ2(fπ1eS)\alpha_{B,S}\colon f\mapsto f\pi_{2}\land(f\pi_{1}\to e_{S})

is a set morphism αB,S:hom𝐁𝐨𝐨𝐥(B,Clos(S))hom𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞(𝐂(B),S)\alpha_{B,S}\colon\hom_{\mathbf{Bool}}(B,\mathrm{Clos}(S))\to\hom_{\mathbf{augStone}}(\mathbf{C}(B),S).

Proof.

Let f=αB,Sff^{*}=\alpha_{B,S}f and c,c𝐂(B)c,c^{\prime}\in\mathbf{C}(B), where c=(a,g)c=(a,g) and c=(a,g)c^{\prime}=(a^{\prime},g^{\prime}).

  • f(0)=f(0B)(f(1B)eS)=0Sf^{*}(0)=f(0_{B})\land(f(1_{B})\to e_{S})=0_{S}.

  • f(1)=f(1B)(f(0B)eS)=1Sf^{*}(1)=f(1_{B})\land(f(0_{B})\to e_{S})=1_{S}.

  • f(e)=f(1B)(f(1B)eS)=eSf^{*}(e)=f(1_{B})\land(f(1_{B})\to e_{S})=e_{S}.

  • f(cc)=f(aa,gg)=f(gg)(f(aa)e)=f(g)f(g)(f(a)e)(f(a)e)=fcfc.\begin{aligned} f^{*}&\left(c\land c^{\prime}\right)=f^{*}(a\lor a^{\prime},g\land g^{\prime})=f(g\land g^{\prime})\land(f(a\lor a^{\prime})\to e)\\ &=f(g)\land f(g^{\prime})\land(f(a)\to e)\land(f(a^{\prime})\to e)=f^{*}c\land f^{*}c^{\prime}.\end{aligned}

  • f(cc)=f(aa,gg)=f(gg)(f(aa)e)=(f(g)(f(a)f(a)e))(f(g)(f(a)f(a)e)).\begin{aligned} f^{*}&\left(c\lor c^{\prime}\right)=f^{*}(a\land a^{\prime},g\lor g^{\prime})=f(g\lor g^{\prime})\land(f(a\land a^{\prime})\to e)\\ &=\left(f(g)\land(f(a)\land f(a^{\prime})\to e)\right)\lor\left(f(g^{\prime})\land(f(a)\land f(a^{\prime})\to e)\right).\end{aligned}
    Since ag=1Ba^{\prime}\lor g^{\prime}=1_{B}, we have ¬ag\neg a^{\prime}\leq g^{\prime}. Thus,

    f(cc)=\displaystyle f^{*}\left(c\lor c^{\prime}\right)= (f(g)(f(a)f(a)e))\displaystyle\left(f(g)\land(f(a)\land f(a^{\prime})\to e)\right)\lor
    (f(g)(f(a)f(a)e))(¬f(a)(f(a)f(a)e))\displaystyle\left(f(g^{\prime})\land(f(a)\land f(a^{\prime})\to e)\right)\lor\left(\neg f(a^{\prime})\land(f(a)\land f(a^{\prime})\to e)\right)
    =\displaystyle= (f(g)(f(a)f(a)e))(f(g)(f(a)f(a)e))\displaystyle\left(f(g)\land(f(a)\land f(a^{\prime})\to e)\right)\lor\left(f(g^{\prime})\land(f(a)\land f(a^{\prime})\to e)\right)
    ¬f(a).\displaystyle\lor\neg f(a^{\prime}).

    The last equality follows from the fact that ¬f(a)f(a)f(a)e\neg f(a^{\prime})\leq f(a)\land f(a^{\prime})\to e. We conclude that

    f(cc)=(f(g)(f(a)e))(f(g)(f(a)f(a)e)).f^{*}\left(c\lor c^{\prime}\right)=\left(f(g)\land(f(a)\to e)\right)\lor\left(f(g^{\prime})\land(f(a)\land f(a^{\prime})\to e)\right). (1)

    By applying an analogous procedure, we obtain

    f(cc)=(f(g)(f(a)f(a)e))(f(g)(f(a)e)).f^{*}\left(c\lor c^{\prime}\right)=\left(f(g)\land(f(a)\land f(a^{\prime})\to e)\right)\lor\left(f(g^{\prime})\land(f(a^{\prime})\to e)\right).

    Conjoining this expression with (1) yields

    f(cc)\displaystyle f^{*}\left(c\lor c^{\prime}\right) =(f(g)(f(a)e))(f(g)(f(a)e))\displaystyle=\left(f(g)\land(f(a)\to e)\right)\lor\left(f(g^{\prime})\land(f(a^{\prime})\to e)\right)
    =fcfc.\displaystyle=f^{*}c\lor f^{*}c^{\prime}.
  • f(cc)=f(gg)(f((a¬a)¬(gg))e)=f(gg)(¬f(a¬a)f(gg)e)=f(gg)(f(a¬a)e)=f(gg)f(¬ag)(f(a¬a)e)\begin{aligned} f^{*}&\left(c^{\prime}\to c\right)\\ &=f(g^{\prime}\to g)\land\left(f\left((a\land\neg a^{\prime})\lor\neg(g^{\prime}\to g)\right)\to e\right)\\ &=f(g^{\prime}\to g)\land\left(\neg f(a\land\neg a^{\prime})\land f(g^{\prime}\to g)\lor e\right)\\ &=f(g^{\prime}\to g)\land\left(f(a\land\neg a^{\prime})\to e\right)\\ &=f(g^{\prime}\to g)\land f(\neg a^{\prime}\to g)\land\left(f(a\land\neg a^{\prime})\to e\right)\end{aligned}

    The last equality follows from the fact that ¬ag\neg a^{\prime}\leq g^{\prime}. We have

    f\displaystyle f^{*} (cc)\displaystyle\left(c^{\prime}\to c\right)
    =(f(g)ef(g)(f(a)e))(¬f(a)f(g)(f(a)e)),\displaystyle=(f(g^{\prime})\land e\to f(g)\land(f(a)\to e))\land(\neg f(a^{\prime})\to f(g)\land(f(a)\to e)),
    which holds because f(gg)=f(g)(ef(g)(f(a)e))f(g^{\prime}\to g)=f(g^{\prime})\land(e\to f(g)\land(f(a)\to e)). We finally obtain
    f\displaystyle f^{*} (cc)=(¬f(a)f(g)e)(f(g)(f(a)e))\displaystyle\left(c^{\prime}\to c\right)=(\neg f(a^{\prime})\lor f(g^{\prime})\land e)\to(f(g)\land(f(a)\to e))
    =f(g)(f(a)e)(f(g)(f(a)e))\displaystyle=f(g^{\prime})\land(f(a^{\prime})\to e)\to(f(g)\land(f(a)\to e))
    =fcfc.\displaystyle=f^{*}c^{\prime}\to f^{*}c.\qed
Proposition 3.2.

Let fhom𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞(𝐂(B),S)f^{*}\in\hom_{\mathbf{augStone}}(\mathbf{C}(B),S). The assignment

βB,S:ffΔ\beta_{B,S}\colon f^{*}\mapsto f^{*}\Delta

is a set morphism βB,S:hom𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞(𝐂(B),S)hom𝐁𝐨𝐨𝐥(B,Clos(S))\beta_{B,S}\colon\hom_{\mathbf{augStone}}(\mathbf{C}(B),S)\to\hom_{\mathbf{Bool}}(B,\mathrm{Clos}(S)). Moreover, αB,S\alpha_{B,S} and βB,S\beta_{B,S} are inverses.

Proof.

Let b,bBb,b^{\prime}\in B and set f=βB,Sff=\beta_{B,S}f^{*}. We have f(0B)=fΔ0B=f0=0f(0_{B})=f^{*}\Delta 0_{B}=f^{*}0=0. Now we study the Boolean operations: f(bb)=f(¬(bb),bb)=fbfb=fbfbf(b\land b^{\prime})=f^{*}(\neg(b\land b^{\prime}),b\land b^{\prime})=f^{*}b\land f^{*}b^{\prime}=fb\land fb^{\prime}. f(bb)=f(¬(bb),bb)=fbfb=fbfbf(b\lor b^{\prime})=f^{*}(\neg(b\lor b^{\prime}),b\lor b^{\prime})=f^{*}b\lor f^{*}b^{\prime}=fb\lor fb^{\prime}. f(¬b)=f(¬(b0B),b0B)=f((¬b,b)0)=fb0=¬fbf(\neg b)=f^{*}(\neg(b\to 0_{B}),b\to 0_{B})=f^{*}((\neg b,b)\to 0)=fb\to 0=\neg fb. Regarding the second part, let (a,g)𝐂(B)(a,g)\in\mathbf{C}(B). We have (αB,SβB,Sf)(a,g)=(αB,SfΔ)(a,g)=(fΔπ2(fΔπ1eS))(a,g)=f(¬g,g)(f(¬a,a)eS)=f((¬g,g)((¬a,a)e))=f((¬g,g)((a,1)))=f(a,g)(\alpha_{B,S}\beta_{B,S}f^{*})(a,g)=(\alpha_{B,S}f^{*}\Delta)(a,g)=(f^{*}\Delta\pi_{2}\land(f^{*}\Delta\pi_{1}\to e_{S}))(a,g)=f^{*}(\neg g,g)\land(f^{*}(\neg a,a)\to e_{S})=f^{*}\left((\neg g,g)\land((\neg a,a)\to e)\right)=f^{*}\left((\neg g,g)\land((a,1))\right)=f^{*}(a,g) and (βB,SαB,Sf)(b)=(fπ2Δ(fπ1ΔeS))(b)=f(b)(f(¬b)eS)=f(b)(f(b)eS)=f(b)(\beta_{B,S}\alpha_{B,S}f)(b)=(f\pi_{2}\Delta\land(f\pi_{1}\Delta\to e_{S}))(b)=f(b)\land(f(\neg b)\to e_{S})=f(b)\land(f(b)\lor e_{S})=f(b). ∎

Corollary 3.3.

𝐂\mathbf{C} is fully faithful.

Proof.

Let BB be a Boolean algebra. The maps Δ:BClos(𝐂(B))\Delta\colon B\to\mathrm{Clos}(\mathbf{C}(B)) and π2:Clos(𝐂(B))B\pi_{2}\colon\mathrm{Clos}(\mathbf{C}(B))\to B form an isomorphism in 𝐁𝐨𝐨𝐥\mathbf{Bool}. Therefore,

hom𝐁𝐨𝐨𝐥(B,B)hom𝐁𝐨𝐨𝐥(B,Clos(𝐂(B)))hom𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞(𝐂(B),𝐂(B)),\hom_{\mathbf{Bool}}(B,B)\cong\hom_{\mathbf{Bool}}(B,\mathrm{Clos}(\mathbf{C}(B)))\cong\hom_{\mathbf{augStone}}(\mathbf{C}(B),\mathbf{C}(B)),

where the second isomorphism follows from Proposition 3.2. ∎

Theorem 3.4.

𝐂\mathbf{C} is the left adjoint of Clos\mathrm{Clos}.

Proof.

Given B𝐎𝐛𝐣(𝐁𝐨𝐨𝐥)B\in\mathbf{Obj}(\mathbf{Bool}) and S𝐎𝐛𝐣(𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞)S\in\mathbf{Obj}(\mathbf{augStone}), Propositions 3.1 and 3.2 show that αB,S:hom𝐁𝐨𝐨𝐥(B,Clos(S))hom𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞(𝐂(B),S)\alpha_{B,S}\colon\hom_{\mathbf{Bool}}(B,\mathrm{Clos}(S))\to\hom_{\mathbf{augStone}}(\mathbf{C}(B),S) and βB,S:hom𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞(𝐂(B),S)hom𝐁𝐨𝐨𝐥(B,Clos(S))\beta_{B,S}\colon\hom_{\mathbf{augStone}}(\mathbf{C}(B),S)\to\hom_{\mathbf{Bool}}(B,\mathrm{Clos}(S)) are inverses. It remains to show this isomorphism is natural in both arguments. Let ρhom𝐁𝐨𝐨𝐥(B,B)\rho\in\hom_{\mathbf{Bool}}(B^{\prime},B) and σhom𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞(S,S)\sigma\in\hom_{\mathbf{augStone}}(S,S^{\prime}). We want to show that the following diagram commutes:

hom𝐁𝐨𝐨𝐥(B,Clos(S)){\hom_{\mathbf{Bool}}(B,\mathrm{Clos}(S))}hom𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞(𝐂(B),S){\hom_{\mathbf{augStone}}(\mathbf{C}(B),S)}hom𝐁𝐨𝐨𝐥(B,Clos(S)){\hom_{\mathbf{Bool}}(B^{\prime},\mathrm{Clos}(S^{\prime}))}hom𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞(𝐂(B),S){\hom_{\mathbf{augStone}}(\mathbf{C}(B^{\prime}),S^{\prime})}αB,S\scriptstyle{\alpha_{B,S}}βB,S\scriptstyle{\beta_{B,S}}σ()ρ\scriptstyle{\sigma\circ()\circ\rho}σ()𝐂(ρ)\scriptstyle{\sigma\circ()\circ\mathbf{C}(\rho)}αB,S\scriptstyle{\alpha_{B^{\prime},S^{\prime}}}βB,S\scriptstyle{\beta_{B^{\prime},S^{\prime}}}

Let fhom𝐁𝐨𝐨𝐥(B,Clos(S))f\in\hom_{\mathbf{Bool}}(B,\mathrm{Clos}(S)) and (a,g)𝐂(B)(a^{\prime},g^{\prime})\in\mathbf{C}(B^{\prime}). We have

(αB,S(σfρ))(a,g)\displaystyle\left(\alpha_{B^{\prime},S^{\prime}}(\sigma f\rho)\right)(a^{\prime},g^{\prime}) =(σfρπ2(σfρπ1eS))(a,g)\displaystyle=(\sigma f\rho\pi_{2}\land(\sigma f\rho\pi_{1}\to e_{S^{\prime}}))(a^{\prime},g^{\prime})
=σfρ(g)(σfρ(a)eS)\displaystyle=\sigma f\rho(g^{\prime})\land(\sigma f\rho(a^{\prime})\to e_{S^{\prime}})
=σ(fρ(g)(fρ(a)eS))\displaystyle=\sigma\left(f\rho(g^{\prime})\land(f\rho(a^{\prime})\to e_{S})\right)
=σ(fπ2(fπ1eS))(ρ(a),ρ(g))\displaystyle=\sigma\left(f\pi_{2}\land(f\pi_{1}\to e_{S})\right)(\rho(a^{\prime}),\rho(g^{\prime}))
=σαB,S(f)𝐂(ρ)(a,g).\displaystyle=\sigma\alpha_{B,S}(f)\mathbf{C}(\rho)(a^{\prime},g^{\prime}).

Let fhom𝐚𝐮𝐠𝐒𝐭𝐨𝐧𝐞(𝐂(B),S)f^{*}\in\hom_{\mathbf{augStone}}(\mathbf{C}(B),S) and bBb^{\prime}\in B^{\prime}. We have

βB,S(σf𝐂(ρ))(b)\displaystyle\beta_{B^{\prime},S^{\prime}}\left(\sigma f^{*}\mathbf{C}(\rho)\right)(b^{\prime}) =σf𝐂(ρ)Δ(b)\displaystyle=\sigma f^{*}\mathbf{C}(\rho)\Delta(b^{\prime})
=σf(¬ρb,ρb)\displaystyle=\sigma f^{*}(\neg\rho b^{\prime},\rho b^{\prime})
=σfΔ(ρb)\displaystyle=\sigma f^{*}\Delta(\rho b^{\prime})
=(σβB,S(f)Δρ)(b).\displaystyle=(\sigma\beta_{B,S}(f^{*})\Delta\rho)(b^{\prime}).\qed

Funding details

This work was supported by NSF and ASEE through an eFellows postdoctoral fellowship.

References

  • [1] Benveniste, A., Caillaud, B., Nickovic, D., Passerone, R., Raclet, J.-B., Reinkemeier, P., Sangiovanni-Vincentelli, A., Damm, W., Henzinger, T. A., and Larsen, K. G. Contracts for system design. Foundations and Trends®{}^{\text{\scriptsize{\textregistered}}} in Electronic Design Automation 12, 2-3 (2018), 124–400.
  • [2] Incer, I. The Algebra of Contracts. PhD thesis, EECS Department, University of California, Berkeley, May 2022.