Home Simplicial and dimension groups with group action and their realization
Article
Licensed
Unlicensed Requires Authentication

Simplicial and dimension groups with group action and their realization

  • Lia Vaš ORCID logo EMAIL logo
Published/Copyright: March 1, 2022

Abstract

We define simplicial and dimension Γ-groups, the generalizations of simplicial and dimension groups to the case when these groups have an action of an arbitrary group Γ. Assuming that the integral group ring of Γ is Noetherian, we show that every dimension Γ-group is isomorphic to a direct limit of a directed system of simplicial Γ-groups and that the limit can be taken in the category of ordered groups with order-units or generating intervals. We adapt Hazrat’s definition of the Grothendieck Γ-group K0Γ(R) for a Γ-graded ring R to the case when Γ is not necessarily abelian. If G is a pre-ordered abelian group with an action of Γ which agrees with the pre-ordered structure, we say that G is realized by a Γ-graded ring R if K0Γ(R) and G are isomorphic as pre-ordered Γ-groups with an isomorphism which preserves order-units or generating intervals. We show that every simplicial Γ-group with an order-unit can be realized by a graded matricial ring over a Γ-graded division ring. If the integral group ring of Γ is Noetherian, we realize a countable dimension Γ-group with an order-unit or a generating interval by a Γ-graded ultramatricial ring over a Γ-graded division ring. We also relate our results to graded rings with involution, which give rise to Grothendieck Γ-groups with actions of both Γ and 2. We adapt the realization problem for von Neumann regular rings to graded rings and concepts from this work and discuss some other questions.


Communicated by Manfred Droste


A Faithfulness and the classification theorem

If Γ is an abelian group and K a Γ-graded field, [14, Theorem 5.2.4] states that the functor K0Γ classifies Γ-graded ultramatricial K-algebras. We show that this result holds without assuming that Γ is abelian (Theorem A.2). In addition, the connecting maps in [14, Theorem 5.2.4] are assumed to be unit-preserving, while we do not make this assumption.

Analogously to the classic proof [10, Theorem 15.26], the proof of [14, Theorem 5.2.4] uses the intertwining method and [14, Theorem 5.1.3] stating that K0Γ is a full and faithful functor on an appropriate category. If Γ is any group and K a Γ-graded field, Proposition 4.9 asserts the fullness of the functor K0Γ on the category of graded matricial K-algebras (if K is commutative, the map ekligkli from the proof extends to a graded algebra homomorphism, not just to a graded ring homomorphism).

We turn to faithfulness next and show Proposition A.1 stating that the function K0Γ is faithful on the category obtained from the category of graded matricial K-algebras by identifying the graded inner automorphism with the identity map. Thus, Propositions 4.9 and A.1 parallel [14, Theorem 5.1.3]. Further, [15, Theorem 3.4] is the involutive version of [14, Theorem 5.1.3, part (1)]: it states that if Γ is abelian and K has an involution which satisfies some additional conditions, then the appropriately modified functor is faithful.

Proposition A.1 (Faithfulness).

Let Γ be any group, let K be a Γ-graded field, let R and S be graded matricial K-algebras, and let ϕ,ψ:RS be graded algebra homomorphisms (not necessarily unit-preserving). The following assertions are equivalent:

  1. K0Γ(ϕ)=K0Γ(ψ).

  2. There exists an invertible element uS1Γ such that ϕ(r)=uψ(r)u-1 for all rR.

Proof.

If the arguments related to the presence of an involution are not considered, the proof of [15, Theorem 3.4] can be used for the current proof: the relations on the standard graded matrix units hold without the requirements that Γ is abelian and Lemma 2.1 can be used instead of [15, Lemma 3.2]. We also emphasize that the maps in [15, Theorem 3.4] are not assumed to be unit-preserving. ∎

Propositions 4.9 and A.1 (fullness and faithfulness) imply the classification result below.

Theorem A.2 (Classification).

If Γ is any group, K is a Γ-graded field, R and S are graded ultramatricial K-algebras, and f:(K0Γ(R),DR)(K0Γ(S),DS) is an isomorphism of the category OGΓD, then there is a graded algebra isomorphism ϕ:RS such that K0Γ(ϕ)=f. If f is order-unit-preserving, then ϕ can be chosen to be unit-preserving.

Proof.

We recall [15, Theorem 4.5] stating that if Γ is abelian and if K has an involution which satisfies some additional conditions, then ϕ can be obtained to be a graded *-algebra isomorphism. If the arguments related to the presence of an involution are not considered and Propositions 4.9 and A.1 are used instead of [15, Theorem 2.7 and Theorem 3.4], the proof of [15, Theorem 4.5] modifies to the proof of the theorem. ∎

One may wonder whether the assumption that K is commutative can be dropped from Theorem A.2 and still have that the homomorphism ϕ is both a graded ring and a graded K-module homomorphism. The example below illustrates that this cannot be required.

Let K be a Γ-graded division ring such that ΓK=Γ. For example, let K=F[Γ] graded by Kγ=F{γ} for any group Γ and any field F trivially graded by Γ. Let γ,δΓ, R=𝕄1(K)(γ) and S=𝕄1(K)(δ). By the assumptions that ΓK=Γ and that K is a graded division ring, there is an invertible element aKδγ-1. Thus, the map xaxa-1 is a graded ring isomorphism RgrS, and so

(K0Γ(R),[R])(K0Γ(S),[S])

in 𝐎𝐆Γu. If we assume that there is ϕ:RS which is both a graded ring and a graded K-module isomorphism, we have that ϕ(x)=xϕ(1) for any xK. In particular, if xRγ=Kγ, then ϕ(x)Sγ=Kδγδ-1 and xϕ(1)Kγ. By considering Γ and K such that Kδγδ-1Kγ, for example Γ=D3=a,ba3=1,b2=1,ba=a2b, K=[Γ], γ=a, and δ=b, we arrive at a contradiction.

B Dimension Γ-group extensions

In the case when Γ is trivial, every dimension group with a generating interval D has a dimension group extension with an order-unit by the group (,1). In this section, we prove a generalization of this statement for an ultrasimplicial Γ-group and, consequently, for a dimension Γ-group in the case when [Γ] is Noetherian.

If G,H and K are ordered Γ-groups, then H is an ordered Γ-group extension of G by K if

is a short exact sequence in the category 𝐎𝐆Γ (thus p, i are order-preserving Γ-group homomorphisms such that i is injective, p is surjective, and i(G)=kerp), i-1(H+)=G+, and p(H+)=K+.

If D is a generating interval of G, u an order-unit of H, and v an order-unit of K, then (H,u) is an ordered Γ-group extension of (G,D) by (K,v) if H is a Γ-group extension of G by K, p is a morphism of 𝐎𝐆Γu, and i-1([0,u])=D (which also implies that i is a morphism of 𝐎𝐆ΓD). This is written as follows:

Proposition B.1.

Let Γ be a group with a subgroup Δ. If G is an ordered and directed Γ-group with a generating interval D, then (G,D) has an ordered Γ-group extension by (Z[Γ/Δ],Δ).

Proof.

Let H=G[Γ/Δ], let i:GH be the natural injection, and let p:H[Γ/Δ] be the natural projection so that the diagram

is a short exact sequence in the category of [Γ]-modules. Let us denote the set

{(x,aΔ)Ha+[Γ] and x+adG+ for some dD}

by H+. It is direct to check that H+ is closed under the action of Γ and that (0,0)H+. Let (x,aΔ),(y,bΔ)H+ and d1,d2D be such that x+ad1,y+bd2G+. Since D is upwards directed, there is dD such that dd1, dd2, so that x+adx+ad10 and y+bdy+bd20. Thus x+y+(a+b)d0, and so H+ is additively closed. Hence, H+ is a cone in H. The cone H+ is strict since the cone G+ is strict. So, H+ defines a partial order on H.

Next, we show that (0,Δ) is an order-unit of H. For any (x,aΔ)H, there is yG+ with xy since G is directed. Then y=i=1nbidi for some n, diD and bi+[Γ] for i=1,,n. Since D is upwards directed, one can find dD such that did for all i=1,,n. Thus,

xyi=1nbid=bdfor b=i=1nbi.

Let c+[Γ] be such that ca+b. This implies that c-ab so that (c-a)dbd. The relation xbd implies that -x+bd0, and so -x+(c-a)d-x+bd0. Thus, we have that (-x,(c-a)Δ)H+, and hence (x,aΔ)(0,cΔ). This demonstrates that (0,Δ) is an order-unit and also implies that H is directed.

By the definition of H+, we have that i and p are order-preserving. Since (0,aΔ) is in H+ for every a+[Γ], we have that p(H+)=+[Γ/Δ]. The relation i-1(H+)=G+ holds since i(x)H+ for xG if and only if x is in G+. The relation p(0,Δ)=Δ holds by the definition of p. So, it remains to show that i-1([(0,0),(0,Δ)])=D. For dD, we have (-d,Δ)H+ by the definition of H+. Hence (0,0)(d,0)(0,Δ), and so i(d)[(0,0),(0,Δ)]. Conversely, if i(x)[(0,0),(0,Δ)] for some xG, then (0,0)(x,0)(0,Δ), which implies that (x,0),(-x,Δ)H+. Hence, xG+ and -x+dG+ for some dD by the definition of H+. Thus 0xd, which implies that xD by the convexity of D. ∎

Proposition B.2.

Let Δ be a normal subgroup of Γ. If G is a simplicial Γ-group with a simplicial Γ-basis stabilized by Δ and a generating interval D, then (G,D) has an ordered Γ-group extension (H,u) by (Z[Γ/Δ],Δ) such that H satisfies (SDPΔ) and that ΔStab(H).

Proof.

Let H=G[Γ/Δ] and u=(0,Δ). Then (H,u) is an ordered Γ-group extension of (G,D) by ([Γ/Δ],Δ) by definition and by Proposition B.1. By Proposition 3.5, Stab(H)=Δ. So, it is sufficient to show that H satisfies (SDPΔ). Let Xi=(xi,biΔ) be in H+ for i=1,,n and assume that i=1naiXi=(0,0) for some ai[Γ]. This implies that i=1naixi=0, that the elements bi can be chosen in +[Γ], that i=1nπ(aibi)=0 where π is the natural map π:[Γ][Γ/Δ], and that there are diD such that xi+bidi0. Since D is upwards directed, there is dD such that did for all i=1,,n, and hence xi+bidxi+bidi0.

Let {y1,,ym}G+ be a simplicial Γ-basis stabilized by Δ. Thus, for xi+bid, dG+, there are bij,cj+[Γ] for i=1,,n, j=1,,m such that

xi+bid=j=1mbijyjandd=j=1mcjyj  for all i=1,,n.

The condition

0=i=1naixi
=i=1nai(xi+bid)-i=1naibid
=i=1nj=1maibijyj-i=1nj=1maibicjyj

implies that

i=1n(π(aibij)-π(aibicj))=0for all j=1,,m.

Since Δ is normal in Γ, the map π is both a left and a right [Γ]-module homomorphism. Therefore, i=1nπ(aibi)=0 implies that i=1nπ(aibicj)=0 for all j=1,,m, and hence

0=i=1nπ(aibij)-i=1nπ(aibicj)=i=1nπ(aibij)-0=i=1nπ(aibij).

Let Yj=(yj,0) for j=1,,m and let Ym+1=(-d,Δ). Then Yj and Ym+1 are in H+ by Proposition B.1. For all i=1,,n, j=1,,m, let Bij=bij+[Γ] and Bi(m+1)=bi+[Γ]. For any i=1,,n,

Xi=(xi,biΔ)
=(j=1mbijyj-bid,biΔ)
=(j=1mbijyj,0)+bi(-d,Δ)
=j=1mbij(yj,0)+bi(-d,Δ)
=j=1mBijYj+Bi(m+1)Ym+1
=j=1m+1BijYj.

For any j=1,,m,

i=1nπ(aiBij)=i=1nπ(aibij)=0andi=1nπ(aiBi(m+1))=i=1nπ(aibi)=0.

This finishes the proof. ∎

Proposition B.2 implies the following proposition.

Proposition B.3.

Let Δ be a normal subgroup of Γ. If (G,D) is a direct limit of a directed system of simplicial Γ-groups with simplicial Γ-bases stabilized by Δ in the category OGΓD, then (G,D) has an ordered Γ-group extension (H,u) by (Z[Γ/Δ],Δ) such that H satisfies (SDPΔ) and such that ΔStab(H).

Proof.

Let I be a directed set, let (G,D) be a direct limit of simplicial Γ-groups ((Gi,Di),gij), i,jI, ij, with simplicial Γ-bases stabilized by Δ, and let gi, iI, be the translational maps. By Proposition B.2, we can find ordered Γ-group extensions (Hi,ui), iI, of (Gi,Di) by ([Γ/Δ],Δ) which satisfy (SDPΔ) and such that ΔStab(Hi). Let ιi denote the inclusion of (Gi,Di) into (Hi,ui) and let pi denote the projection (Hi,ui)([Γ/Δ],Δ). If hij=gij1[Γ/Δ] where the second term is the identity map on [Γ/Δ], then the system ((Hi,ui),hij), i,jI, ij, is a directed system in 𝐎𝐆Γu. Note that hijιi=ιjgij and pjhij=pi for all ij by the proof of Proposition B.2. Let (H,u) be a direct limit of this directed system which exists by Proposition 4.1 and let hi=gi1[Γ/Δ]. Since all Γ-groups Hi satisfy (SDPΔ) and are stabilized by Δ, H satisfies (SDPΔ) and it is stabilized by Δ.

Define the maps ι:GH and p:H[Γ/Δ] by ι(gi(xi))=hiιi(xi) and p(hi(yi))=pi(yi), respectively. We claim that we obtain the required properties from the commutative diagram below:

Indeed, one checks that the maps ι and p are well-defined, that ι is injective, and that p is surjective using properties of a direct limit. Then one checks that kerp is equal to the image of ι, that ι and p are order-preserving, that ι-1(H+)=G+, that p(u)=Δ, and that p(H+)=+[Γ/Δ] using the definitions of the maps. It remains to check that ι-1([0,u])=D. If dD=iIgi(Di), then there are iI and diDi=ιi-1([0,ui]) such that d=gi(di). Hence, the relation 0ιi(di)ui holds in Hi. Applying hi to this relation, we obtain that

0hiιi(di)=ιgi(di)hi(ui)=u.

Hence, d=gi(di)ι-1([0,u]). Conversely, if xι-1([0,u]), let ι(x)=hi(yi) for some yiHi. Since

0=pι(x)=phi(yi)=pi(yi),

we obtain yi=ιi(xi) for some xiGi. Thus,

ι(x)=hi(yi)=hiιi(xi)=ιgi(xi),

which implies x=gi(xi) since ι is injective. The relation 0ι(x)=hiιi(xi)u=hi(ui) implies that there is ji such that

0hijιi(xi)=ιjgij(xi)hij(ui)=uj,

and so gij(xi)ιj-1([0,uj])=Dj. Thus,

x=gi(xi)=gjgij(xi)gj(Dj)iIgi(Di)=D,

as desired. ∎

We show the main result of this section now.

Theorem B.4.

If Γ is such that Z[Γ] is Noetherian and (G,D) is a dimension Γ-group satisfying (SDPΔ) for some subgroup Δ of Γ such that ΔStab(G), then (G,D) has an ordered Γ-group extension (H,u) by (Z[Γ/Δ¯],Δ¯), where Δ¯ is the normal closure of Δ in Γ, such that H satisfies (SDPΔ¯) and such that Δ¯Stab(H).

Proof.

By Theorem 5.12, we can find a directed system of simplicial Γ-groups with simplicial Γ-bases stabilized by Δ¯ such that (G,D) is a direct limit of this directed system in 𝐎𝐆ΓD. By Proposition B.3, an ordered Γ-group extension (H,u) by ([Γ/Δ¯],Δ¯) exists and it satisfies the required properties. ∎

We ask whether the assumption that [Γ] is Noetherian can be dropped from Theorem B.4.

Acknowledgements

The author is very grateful to the referee for a careful read, a thoughtful report, and for making some highly valuable suggestions.

References

[1] P. Ara, The realization problem for von Neumann regular rings, Ring Theory 2007, World Scientific, Hackensack (2009), 21–37. 10.1142/9789812818331_0004Search in Google Scholar

[2] P. Ara, J. Bosa and E. Pardo, The realization problem for finitely generated refinement monoids, Selecta Math. (N. S.) 26 (2020), no. 3, Paper No. 33. 10.1007/s00029-020-00559-5Search in Google Scholar

[3] P. Ara and M. Brustenga, The regular algebra of a quiver, J. Algebra 309 (2007), no. 1, 207–235. 10.1016/j.jalgebra.2006.10.013Search in Google Scholar

[4] P. Ara and K. R. Goodearl, Leavitt path algebras of separated graphs, J. Reine Angew. Math. 669 (2012), 165–224. 10.1515/CRELLE.2011.146Search in Google Scholar

[5] P. Ara, R. Hazrat, H. Li and A. Sims, Graded Steinberg algebras and their representations, Algebra Number Theory 12 (2018), no. 1, 131–172. 10.2140/ant.2018.12.131Search in Google Scholar

[6] P. Ara and E. Pardo, Representing finitely generated refinement monoids as graph monoids, J. Algebra 480 (2017), 79–123. 10.1016/j.jalgebra.2017.02.007Search in Google Scholar

[7] M. Cohen and S. Montgomery, Group-graded rings, smash products, and group actions, Trans. Amer. Math. Soc. 282 (1984), no. 1, 237–258. 10.1090/S0002-9947-1984-0728711-4Search in Google Scholar

[8] K. R. Davidson, C*-Algebras by Example, Fields Inst. Monogr. 6, American Mathematical Society, Providence, 1996. 10.1090/fim/006Search in Google Scholar

[9] K. R. Goodearl, Partially Ordered Abelian Groups with Interpolation, Math. Surveys Monogr. 20, American Mathematical Society, Providence, 1986. Search in Google Scholar

[10] K. R. Goodearl, Von Neumann Regular Rings, 2nd ed., Robert E. Krieger, Malabar, 1991. Search in Google Scholar

[11] K. R. Goodearl and D. E. Handelman, Classification of ring and C*-algebra direct limits of finite-dimensional semisimple real algebras, Mem. Amer. Math. Soc. 69 (1987), no. 372, 1–147. 10.1090/memo/0372Search in Google Scholar

[12] P. Hall, Finiteness conditions for soluble groups, Proc. Lond. Math. Soc. (3) 4 (1954), 419–436. 10.1112/plms/s3-4.1.419Search in Google Scholar

[13] R. Hazrat, The graded Grothendieck group and the classification of Leavitt path algebras, Math. Ann. 355 (2013), no. 1, 273–325. 10.1007/s00208-012-0791-3Search in Google Scholar

[14] R. Hazrat, Graded Rings and Graded Grothendieck Groups, London Math. Soc. Lecture Note Ser. 435, Cambridge University, Cambridge, 2016. 10.1017/CBO9781316717134Search in Google Scholar

[15] R. Hazrat and L. Vaš, K-theory classification of graded ultramatricial algebras with involution, Forum Math. 31 (2019), no. 2, 419–463. 10.1515/forum-2017-0268Search in Google Scholar

[16] C. Năstăsescu and F. Van Oystaeyen, Methods of Graded Rings, Lecture Notes in Math. 1836, Springer, Berlin, 2004. 10.1007/b94904Search in Google Scholar

[17] F. Wehrung, Non-measurability properties of interpolation vector spaces, Israel J. Math. 103 (1998), 177–206. 10.1007/BF02762273Search in Google Scholar

Received: 2020-04-27
Revised: 2021-12-14
Published Online: 2022-03-01
Published in Print: 2022-05-01

© 2022 Walter de Gruyter GmbH, Berlin/Boston

Downloaded on 24.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/forum-2020-0101/html
Scroll to top button