Home Locally finite groups and configurations
Article Publicly Available

Locally finite groups and configurations

  • Mahdi Meisami , Ali Rejali EMAIL logo , Meisam Soleimani Malekan and Akram Yousofzadeh
Published/Copyright: August 3, 2021

Abstract

Let 𝐺 be a discrete group. In 2001, Rosenblatt and Willis proved that 𝐺 is amenable if and only if every possible system of configuration equations admits a normalized solution. In this paper, we show independently that 𝐺 is locally finite if and only if every possible system of configuration equations admits a strictly positive solution. Also, we give a procedure to get equidecomposable subsets 𝐴 and 𝐵 of an infinite finitely generated or a locally finite group 𝐺 such that A B , directly from a system of configuration equations not having a strictly positive solution.

1 Introduction

The concept of configuration originated in the work of Willis and Rosenblatt, cf. [10] on locally compact (in particular discrete) groups. They used it to construct the first example of a net of positive, normalized functions in L 1 ( G ) such that the net is weak convergent to invariance but does not converge strongly to invariance. This concept was then applied to characterize amenability of groups in the same paper. Some other theoretical properties of groups have been studied through configurations, see [2, 3, 7, 6].

Definition 1.1

Let 𝐺 be a group. Two subsets 𝐴 and 𝐵 of 𝐺 are called equidecomposable if there exist partitions { A 1 , , A n } and { B 1 , , B n } of 𝐴 and 𝐵 respectively and elements g 1 , , g n of 𝐺 such that

B i = g i A i ( 1 i n ) .

In such a case, we use the notation A B .

Note that is an equivalence relation on P ( G ) . It is clear that if 𝐴 and 𝐵 are two finite equidecomposable sets, then | A | = | B | . Thus no finite set is equidecomposable with its proper subsets. In point of fact, this property characterizes exactly the family of finite groups, see Theorem 3.5 below.

A group 𝐺 is said to be paradoxical if it admits a paradoxical decomposition, i.e., if there exist disjoint subsets A 1 , , A n , B 1 , , B m of 𝐺 and elements g 1 , , g n , h 1 , , h m G such that

G = i = 1 n g i A i = j = 1 m h j B j .

In other words, 𝐺 is paradoxical if it is equidecomposable with two disjoint subsets.

The Tarski number of 𝐺, which is denoted by τ ( G ) , is the smallest number of pieces needed for a paradoxical decomposition. If the group admits no paradoxical decompositions, it is usual to write τ ( G ) = . It is easily checked that the smallest possible Tarski number of any group is 4. In fact, one has τ ( G ) = 4 if and only if 𝐺 contains a non-abelian free subgroup. It is well known that a discrete group is amenable if and only if it admits no paradoxical decompositions.

Given a non-amenable group 𝐺, in [9, 16], it is explained how to construct paradoxical decompositions of 𝐺 by using configurations on 𝐺. In [8], the authors defined an extended version of configuration of groups, known as two-sided configuration. This new definition enabled the authors to investigate more aspects of discrete groups (see also [11]). Configurations were defined also beyond the class of groups. They are extended to more algebraic structures such as semigroups and hypergroups [1, 14, 15].

We use the first definition of configurations throughout this paper. In this note, our aim is to use this notion in order to characterize locally finite groups, that is, groups such that every finitely generated subgroup is finite. In the sequel, we provide a simple procedure for finding special equidecomposable subsets of an finitely generated infinite group or a non-locally compact group.

2 Notations and basic definitions

Our notation and terminology are standard. For a set 𝐴, | A | is the cardinality of 𝐴, and by P ( A ) we mean the power set of 𝐴, i.e., the family of all subsets of 𝐴. If A X , then we use A for the complement of the set 𝐴. The notation is applied for disjoint union of sets. Throughout this paper, 𝐺 denotes a discrete group. Suppose that 𝐺 is a finitely generated group and g = ( g 1 , , g n ) is an ordered string of elements of 𝐺 and E = { E 1 , , E m } is a finite partition of 𝐺. A configuration corresponding to ( g , E ) is an ( n + 1 ) -tuple C = ( C 0 , C 1 , , C n ) , where 1 C j m , such that there exists x G satisfying x E C 0 and g j x E C j for all j = 1 , 2 , , n . The set of all such ( n + 1 ) -tuples is denoted by Con ( g , E ) and the family of all possible such sets is denoted by Con ( G ) . This notion was defined in [10]. The pair ( g , E ) is called a configuration pair for 𝐺. If 𝔤 is taken as a generating string for 𝐺, we use the notation Con * ( g , E ) and Con * ( G ) instead of Con ( g , E ) and Con ( G ) . Two groups G 1 and G 2 are called configuration equivalent (*-configuration equivalent) if Con ( G 1 ) = Con ( G 2 ) ( Con * ( G 1 ) = Con * ( G 2 ) ). The best general references for the study of *-configuration equivalent groups are [2, 3, 8, 7, 6]. See also the following example.

Example 2.1

The equations Con ( G 1 H 1 ) = Con ( G 2 H 2 ) and Con ( H 1 ) = Con ( H 2 ) do not imply the equality Con ( G 1 ) = Con ( G 2 ) . For a counterexample, let G 1 = Z , H 1 = 2 Z , G 2 = Z 2 × Z and H 2 = Z , see [2].

Remark 2.2

Let 𝐺 be a group, and let H G be a subgroup. Then, by [17, Theorem 3.8], the following holds.

  1. Con ( H ) Con ( G ) . In particular, τ ( G ) τ ( H ) .

  2. If the subgroup 𝐻 is normal in 𝐺, then Con ( G / H ) Con ( G ) . In particular, τ ( G ) τ ( G / H ) .

For a configuration pair ( g , E ) , consider the following finite system of m n ( n + 1 ) 2 equations in real variables ( f C ) C Con ( g , E ) :

(2.1) C Con ( g , E ) C j = i f C = C Con ( g , E ) C k = i f C ( 0 j k n ,  1 i m ) .

If g = ( g 1 , , g n ) , E = { E 1 , , E m } , C = ( C 0 , C 1 , , C n ) Con ( g , E ) , the following notations are also used:

x 0 ( C ) = E C 0 ( j = 1 n g j - 1 E C j ) ,

which is the set of elements of E C 0 that give rise to 𝐶 and

x j ( C ) = g j x 0 ( C ) ( 1 j n ) .

Note that, for all j = 0 , 1 , , n and all C 1 , C 2 Con ( g , E ) , two sets x j ( C 1 ) and x j ( C 2 ) intersect if and only if C 1 = C 2 . Thus the family { x j ( C ) : x j ( C ) E i } is actually a partition of E i for all i = 1 , 2 , , m , as 𝐶 ranges over all configurations.

So one can rephrase system (2.1) as follows:

C Con ( g , E ) x j ( C ) E i f C = C Con ( g , E ) x k ( C ) E i f C ( 0 j k n ,  1 i m ) .

This system is called the system of configuration equations corresponding to the pair ( g , E ) and is denoted by Eq ( g , E ) . In [10], the authors proved that a group 𝐺 is amenable if and only if every possible instance of configuration equations has a normalized solution, where by a normalized solution for (2.1) we mean a solution ( f C ) C Con ( g , E ) such that each f C 0 and f C = 1 .

Besides the normalized solutions, strictly positive solutions will also play an important role in our work. We say that a solution ( f C ) C Con ( g , E ) is strictly positive if each f C > 0 . We will prove that a discrete group 𝐺 is not locally finite if and only if every possible instance of configuration equations has a strictly positive solution. As every non-amenable discrete group is paradoxical, one may ask how to construct a paradoxical decomposition directly from a system of configuration equations admitting no normalized solutions. Some attempts at such a construction are given in [16]. Now a similar question arises for the case of non-locally finite groups: “How can we construct two equidecomposable subsets A B of a non-locally finite group 𝐺, directly from a system Eq ( g , E ) not having a strictly positive solution?” Theorem 2.4 along with Theorem 3.9 will answer this question.

We now present the following lemma and theorem as prerequisites for the next section.

Lemma 2.3

Let A , B , C , D be subsets of 𝐺 such that A B = . If A C and B D , then there exist two disjoint subsets 𝑀 and 𝑁 of A B such that ( C D ) M and ( C D ) N .

Proof

Let A = i = 1 n A i , B = j = 1 m B j , C = i = 1 n C i , D = j = 1 m D j , and for some g i G and h j G , we have

C i = g i A i and D j = h j B j .

Then ( C D ) M and ( C D ) N , where

M = A ( j = 1 m h j - 1 [ D j C ] ) and N = ( j = 1 m h j - 1 [ D j C ] ) .

Theorem 2.4

Let ( g , E ) be a configuration pair for 𝐺, where g = ( g 1 , , g n ) is a generating string and E = { E 1 , , E m } . Suppose 1 i 1 < i 2 < < i s m , 0 j 1 , , j s n and 0 k 1 , , k s n are such that, for every C Con ( g , E ) ,

(2.2) | { t : 1 t s , x k t ( C ) E i t } | - | { t : 1 t s , x j t ( C ) E i t } | 0

and, for some C * Con ( g , E ) ,

(2.3) | { t : 1 t s , x k t ( C * ) E i t } | - | { t : 1 t s , x j t ( C * ) E i t } | > 0 .

Then there exists B G such that

B t = 1 s E i t B .

Proof

As mentioned before, for every 1 i m and 0 j , k n ,

C Con ( g , E ) { x 0 ( C ) : x k ( C ) E i } = g k - 1 g j C Con ( g , E ) { x 0 ( C ) : x j ( C ) E i } .

In particular,

E i = g j C Con ( g , E ) { x 0 ( C ) : x j ( C ) E i } .

Setting M j C = g j x 0 ( C ) , where x j ( C ) E i , we have

E i = x j ( C ) E i M j C .

Consequently, for every 1 i m and 0 j n ,

(2.4) E i C Con ( g , E ) { x 0 ( C ) : x j ( C ) E i } .

Let D = t = 1 s { C : x k t ( C ) E i t } . For C D , define

p C = C Con ( g , E ) | { t : x k t ( C ) E i t } | .

Since E i 1 , , E i s are pairwise disjoint, by Lemma 2.3, there exist disjoint families of pairwise disjoint subsets { M 1 C , M 2 C , , M p C C } C D of t = 1 s E i t such that, for each C D ,

x 0 ( C ) M l C ( 1 l p C ) .

Now, by (2.2) and (2.3), there exist disjoint subsets M 1 , M 2 , , M s , M * of t = 1 s E i t such that x 0 ( C * ) M * and

{ x 0 ( C ) : C Con ( g , E ) , x j t ( C ) E i t } M t .

On the other hand, by (2.4), for every 1 t s ,

{ x 0 ( C ) : C Con ( g , E ) , x k t ( C ) E i t } E i t .

Therefore, t = 1 s M t t = 1 s E i t , and however,

t = 1 s M t t = 1 s E i t .

3 Locally finite groups and configurations

Let G 1 and G 2 be two groups. The equivalence of the configurations of these groups implies that, for each pair ( g , E ) of a finite ordered string 𝔤 and a finite partition ℰ of G 1 , there exists a similar pair ( h , F ) for G 2 such that

Con ( g , E ) = Con ( h , F ) .

This implies that G 1 and G 2 interact with each other in many aspects. For example, commutativity, amenability, finiteness, being torsion and nilpotency of one of them can be carried to the other one; for the explicit study of this topic, see [2, 3, 8, 7, 6, 17]. In this direction, the following question was asked in [2].

Question 3.1

Is there an infinite group 𝐺 such that, for any Con ( g , E ) in Con ( G ) , there exists a finite group 𝐹 with Con ( g , E ) Con ( F ) ?

It is obvious that if the answer is affirmative, then 𝐺 could not be non-amenable or non-torsion because single configuration pairs can confirm these two properties, which are not the properties of finite groups [2]. We now answer this question negatively in general. Before proceeding to the answer, let us give some preliminaries.

Definition 3.2

Suppose 𝐺 is a finitely generated group, and let g = ( g 1 , , g n ) be a generating string. The Cayley graph Γ = Cay ( G , g ) is the directed graph with vertex set 𝐺 and edges the pairs ( x , g i x ) G × G , where x G , i = 1 , 2 , , n .

It is easily seen that if g = ( g 1 , , g n ) is a symmetric generating set for 𝐺, then the corresponding Cayley graph is a connected locally finite graph.

The following lemma was obtained after a discussion in MathOverflow [13].

Lemma 3.3

Let 𝐺 be finitely generated by the ordered set g = { g 1 , , g n } . Then 𝐺 is infinite if and only if it possesses two subsets 𝐴 and 𝐵 such that A B , which are equidecomposable.

Proof

The “if” part of the lemma is clear. Conversely, assume that 𝐺 is infinite. By the König lemma [4], the Cayley graph Cay ( G , g ) contains a ray (a path with no repeated vertices) that starts at a vertex and continues from it through infinitely many vertices. It means that there exists an infinite subset B = { x 0 , x 1 , x 2 , } of 𝐺 such that x j + 1 x j - 1 g , j N { 0 } . Set A := B { x 0 } , and for 1 k n , define

B k = { x j B : x j + 1 x j - 1 = g k } and A k = g k B k .

Clearly, A B . On the other hand, A = k = 1 n A k and B = k = 1 n B k . Therefore, 𝐴 and 𝐵 are equidecomposable. This proves the lemma. ∎

Recall that a group is locally finite if every finitely generated subgroup is finite. As a corollary of Lemma 3.3 we have the following theorem.

Theorem 3.4

Let 𝐺 be a discrete group. Then 𝐺 is not locally finite if and only if it possesses two subsets 𝐴 and 𝐵 such that A B , which are equidecomposable.

Proof

Suppose that 𝐺 has two equidecomposable subsets 𝐴 and 𝐵 such that A B . In other words, there exist partitions { A 1 , , A n } and { B 1 , , B n } of 𝐴 and 𝐵 respectively and elements g 1 , , g n of 𝐺 such that

B i = g i A i ( 1 i n ) .

Let 𝐻 be the subgroup generated by the finite set { g 1 , , g n , x } , x B A . It is easy to see that A 1 := A H and B 1 := B H are equidecomposable and A 1 B 1 . Consequently, 𝐻 is not finite by Lemma 2.4. This implies that 𝐺 is not locally finite. Suppose conversely that 𝐺 is not locally finite, and let 𝐻 be one of its infinite finitely generated subgroups. Then, again by Lemma 3.3, 𝐻 admits a subset 𝐵 equidecomposable with a proper subset 𝐴 of itself. Clearly, these sets are also equidecomposable subsets of the whole group 𝐺. This completes the proof.∎

Theorem 3.5

Let 𝐺 be a finitely generated discrete group. Then the following conditions are equivalent.

  1. 𝐺 is finite.

  2. Every configuration equation Eq ( g , E ) of 𝐺 admits a strictly positive solution.

Proof

Let 𝐺 be finite, and let ( g , E ) be a configuration pair for 𝐺. Defining

f C := | x 0 ( C ) |

for every C Con ( g , E ) , one achieves a strictly positive solution for Eq ( g , E ) . Suppose on the other hand that 𝐺 is not finite. Then, by Lemma 3.3, there are two equidecomposable subsets 𝐴 and 𝐵 of 𝐺 such that A B . In other words, D := B A , and there are partitions { A 1 , , A n } and { B 1 , , B n } of 𝐴 and 𝐵 respectively and elements g 1 , , g n of 𝐺 such that B i = g i A i ( 1 i n ). Set E = { A 1 , , A n , D , B c } and g = ( g 1 - 1 , , g n - 1 ) . Suppose that Eq ( g , E ) has a strictly positive solution ( f C ) C Con ( g , E ) . Then

i = 1 n x 0 ( C ) A i f C = i = 1 n x i ( C ) A i f C = i = 1 n g i - 1 x 0 ( C ) A i f C = i = 1 n x 0 ( C ) B i f C = i = 1 n x 0 ( C ) A i f C + x 0 ( C ) D f C ,

and the first equation holds because ( f C ) C Con ( g , E ) is a solution to the system Eq ( g , E ) . This implies that x 0 ( C ) D f C = 0 , which contradicts the strict positivity of ( f C ) since D . ∎

Now we answer Question 3.1.

Theorem 3.6

Let 𝐺 be an infinite group. Then there exists a configuration pair ( g , E ) of 𝐺 such that Con ( g , E ) Con ( F ) for any finite group 𝐹.

Proof

It is clear by Theorem 3.5. ∎

Corollary 3.7

Let 𝐺 be a discrete group. Then the following conditions are equivalent.

  1. 𝐺 is locally finite.

  2. Every configuration equation Eq ( g , E ) of 𝐺 admits a strictly positive solution.

Proof

Let 𝐺 be a locally finite group, and let ( g , E ) be a configuration pair for 𝐺. We shall prove that Eq ( g , E ) admits a strictly positive solution. Suppose that, for each C Con ( g , E ) , x C is a fixed element of x 0 ( C ) and 𝐻 is the group generated by the finite set g { x C : C Con ( g , E ) } . Then, by assumption, 𝐻 is a finite group. Let E = { E 1 , , E m } and g = { g 1 , , g n } . For each 1 i m , set

F i = H E i and F = { F i : 1 i m } .

Note that each F i is a non-empty set and ℱ is a partition for 𝐻. Consequently, ( g , F ) is a configuration pair for 𝐻. We prove that Con ( g , E ) = Con ( g , F ) . Let C = ( C 0 , C 1 , , C n ) Con ( g , E ) . Then

x 0 ( C ) E C 0 and x j ( C ) E C j .

So we have

x C H E C 0 = F C 0 and g j x C H E C j = F C j .

This means that

C = ( C 0 , C 1 , , C n ) Con ( g , F ) .

The inclusion Con ( g , F ) Con ( g , E ) is clear since each F i E i . Now, by Theorem 3.5, Eq ( g , E ) = Eq ( g , F ) admits a strictly positive solution.

For the converse, suppose that every configuration equation Eq ( g , E ) of 𝐺 admits a strictly positive solution. Then, by Theorem 3.4 and an argument like that in the proof of Theorem 3.5, 𝐺 is a locally finite group. ∎

It is well known that every locally finite group is amenable. On the other hand, the condition of amenability is equivalent to not having any paradoxical decomposition. Therefore, we also have the following result.

Proposition 3.8

Let 𝐺 be a discrete group. Then 𝐺 is not locally finite if and only if it is paradoxical or it admits an amenable infinite finitely generated subgroup.

In order to make precise the relation between configurations and equidecomposable sets in a more general case, we now recall the following well-known theorem proved by Stiemke [12] in 1915. Suppose 𝐴 is an n × m matrix, and for z = ( z 1 , , z k ) R k , write z > 0 when z j > 0 for each 𝑗, z 0 when z j 0 for each 𝑗, and write z 0 to mean z 0 and z 0 .

Theorem 3.9

Theorem 3.9 (Stiemke’s theorem [5])

Exactly one of the following systems has a solution:

  1. y T A 0 for some y R n ,

  2. A x = 0 , x > 0 , for some x R m .

Let 𝐺 be an infinite discrete group, let g = ( g 1 , , g n ) be an ordered string of elements of 𝐺 and let E = { E 1 , , E m } be a partition for 𝐺. If 1 i 1 , i 2 m and 0 j 1 , j 2 , k 1 , k 2 n , we call the equations

C Con ( g , E ) C j 1 = i 1 f C - C Con ( g , E ) C k 1 = i 1 f C = 0 and C Con ( g , E ) C j 2 = i 2 f C - C Con ( g , E ) C k 2 = i 2 f C = 0 .

Let Eq ( g , E ) have no strictly positive solutions. Write this homogeneous system as A X = 0 , in which 𝑋 is the column vector of variables and

A = ( A 1 A ( n + 1 ) × n m )

is the coefficient matrix of the system. Then, by Stiemke’s theorem, there exists a vector y R ( n + 1 ) × n m such that y T = ( y 1 , , y ( n + 1 ) × n m ) and y T A 0 . By the nature of Eq ( g , E ) , we can assume that each y i 0 . Suppose in addition that y i { 0 , 1 } and y i = y j = 1 only if i = j or A i and A j are the coefficient vectors of equations from different classes. Under these conditions, y T A 0 means that, for some 1 i 1 < i 2 < < i s m , 0 j 1 , , j s n and 0 k 1 , , k s n ,

t = 1 s A i t 0 ,

where A i t is the coefficient vector of

C Con ( g , E ) C j t = i t f C - C Con ( g , E ) C k t = i t f C .

Hence, for every C Con ( g , E ) ,

| { t : x k t ( C ) E i t } | - | { t : x j t ( C ) E i t | 0 ,

and for some C * Con ( g , E ) ,

| { t : x k t ( C * ) E i t | - | { t : x j t ( C * ) E i t | > 0 .

These are exactly the conditions of Theorem 2.4. This way, the process of constructing equidecomposable subsets 𝐴 and 𝐵 of 𝐺 such that A B is given by the proof of that theorem.

Award Identifier / Grant number: 99010762

Funding statement: The research of the third author was in part supported by a grant from Iran National Science Foundation (INSF) (No. 99010762).

Acknowledgements

The authors wish to thank professor George A. Willis for his valuable comments on the paper. They would also like to thank the referee for their careful reading of the manuscript and for various comments.

  1. Communicated by: George A. Willis

References

[1] A. Abdollahi and A. Rejali, Paradoxical decomposition of semigroups, Semigroup Forum 71 (2005), no. 3, 471–480. 10.1007/s00233-005-0535-8Search in Google Scholar

[2] A. Abdollahi, A. Rejali and G. A. Willis, Group properties characterised by configurations, Illinois J. Math. 48 (2004), no. 3, 861–873. 10.1215/ijm/1258131056Search in Google Scholar

[3] A. Abdollahi, A. Rejali and A. Yousofzadeh, Configuration of nilpotent groups and isomorphism, J. Algebra Appl. 8 (2009), no. 3, 339–350. 10.1142/S0219498809003333Search in Google Scholar

[4] D. König, Über eine Schlussweise aus dem Endlichen ins Unendliche, Acta Sci. Math. (Szeged) 3 (1927), no. 2–3, 121–130. Search in Google Scholar

[5] C.-T. Perng, On a class of theorems equivalent to Farkas’s lemma, Appl. Math. Sci. 11 (2017), no. 44, 2175–2184. 10.12988/ams.2017.77218Search in Google Scholar

[6] A. Rejali and M. Soleimani Malekan, Configuration equivalence is not equivalent to isomorphism, Internat. J. Algebra Comput. 27 (2017), no. 8, 1073–1085. 10.1142/S0218196717500503Search in Google Scholar

[7] A. Rejali and M. Soleimani Malekan, Solubility of groups can be characterized by configuration, New York J. Math. 23 (2017), 1427–1445. Search in Google Scholar

[8] A. Rejali and A. Yousofzadeh, Group properties characterized by two-sided configurations, Algebra Colloq. 17 (2010), no. 4, 583–594. 10.1142/S1005386710000568Search in Google Scholar

[9] A. Rejali and A. Yousofzadeh, Configuration of groups and paradoxical decompositions, Bull. Belg. Math. Soc. Simon Stevin 18 (2011), no. 1, 157–172. 10.36045/bbms/1299766495Search in Google Scholar

[10] J. M. Rosenblatt and G. A. Willis, Weak convergence is not strong convergence for amenable groups, Canad. Math. Bull. 44 (2001), no. 2, 231–241. 10.4153/CMB-2001-023-xSearch in Google Scholar

[11] M. Soleimani Malekan and A. Rejali, Two-sided configuration equivalence and isomorphism, J. Algebra Appl. 19 (2020), no. 10, 2050189. 10.1142/S0219498820501893Search in Google Scholar

[12] E. Stiemke, Über positive Lösungen homogener linearer Gleichungen, Math. Ann. 76 (1915), no. 2–3, 340–342. 10.1007/BF01458147Search in Google Scholar

[13] R. Tucker-Drob, Re: Can infiniteness of finitely generated groups be read by a paradoxical decomposition? (Online forum comment), https://mathoverflow.net/questions/298045/. Search in Google Scholar

[14] B. Willson, Configurations and invariant nets for amenable hypergroups and related algebras, Trans. Amer. Math. Soc. 366 (2014), no. 10, 5087–5112. 10.1090/S0002-9947-2014-05731-0Search in Google Scholar

[15] A. Yousofzadeh, Paradoxical partition of unity for hypergroups, Publ. Res. Inst. Math. Sci. 53 (2017), no. 3, 419–429. 10.4171/PRIMS/53-3-4Search in Google Scholar

[16] A. Yousofzadeh, A constructive way to compute the Tarski number of a group, J. Algebra Appl. 17 (2018), no. 7, Article ID 1850139. 10.1142/S0219498818501396Search in Google Scholar

[17] A. Yousofzadeh, A. Tavakoli and A. Rejali, On configuration graph and paradoxical decomposition, J. Algebra Appl. 13 (2014), no. 2, Article ID 1350086. 10.1142/S0219498813500862Search in Google Scholar

Received: 2021-08-27
Revised: 2021-05-27
Published Online: 2021-08-03
Published in Print: 2022-01-01

© 2021 Walter de Gruyter GmbH, Berlin/Boston

Downloaded on 10.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/jgth-2020-0178/html
Scroll to top button