Arxiv:2001.01058V2 [Math.RA] 28 May 2021 Finite Unitary Rings with A

Home , Finite ring

arXiv:2001.01058v2 [math.RA] 28 May 2021 1]soe htif that showed [10] hog hi ruso nt tegopo nt farn sas calle also if is that ring proved that a [7] such of in fin units Eldridge the 1968, of b in study group have example, (the mathematicians For and units Several theory of groups ring their in view. through interest of of points topics different the many of one are rings Finite Introduction 1 iieuiayrnswt igesbru fprime of subgroup single a with rings unitary Finite -alAddress: E-mail Keywords MSC 2010 p then p Let Abstract. β ∤ k hc soeo h i xlctydsrbdtypes. described explicitly six the of one is which R eso hti h ru fuisof units of group the if that show We . m R eauiayrn ffiiecardinality finite of ring unitary a be eatmnod ae´tc-C-FM 98-0,Manaus Matemática-ICE-UFAM, 69080-900, de Departamento a uefe atrzto,then factorization, free cube has = ∼ A eateto ahmtc,Pym orUiest,Tehran University, Noor Payame Mathematics, of Department iiern,Gopo units. of Group ring, Finite : L 1P0 16U60. :16P10, [email protected] ;[email protected] R B, safiiern n tms n ipecmoeto h semi-simple the of component simple one most at and ring finite a is re ftegopo units of group the of order where otf mn,Mhe Amiri Mohsen Amini, Mostafa B safiiern forder of ring finite a is 1 R R sacmuaiern.I 99 rz in Groza 1989, In ring. commutative a is p a tms n ugopo order of subgroup one most at has β k R where , k safiiern ihui forder of unit with ring finite a is and p A sapienme and number prime a is sarn fcardinality of ring a is A,Brazil -AM, Iran , don Group). Adjoint d e tde from studied een t ntr rings unitary ite p , m R ∗ ring J(R) is a field of order 2, then R (units of R) is a nilpotent group if and only if R is a direct sum of two-sided ideals that are homomorphic images of group algebras of type SP , where S is a particular commutative finite ring and P is a finite p-group for a prime number p. In 2009, Dolzan in [6] improved this result and described the structure of finite rings in which the group of units is nilpotent. J. F. Waters in [13] gave a ring-theoretical characterisation of the class of radical rings with finitely generated nilpotent group of units. J. B. Bateman and D. B. Coleman in [2] determined under what conditions the group of units in the group algebra of a finite group is nilpotent. P. B. Bhattacharya and S. K. Jain in [3] studied certain classes of rings R for which R∗ is nilpotent, supersolvable or solvable. Also, in [4] they proved that if F is a field of characteristic p and G a finite group, then the group of units of the group ring R = FG is nilpotent if and only if G is nilpotent and each q-sylow subgroup of G, q =6 p, is abelian. Recently, in 2018, M. Amiri and M. Ariannejad in [1] classified all finite unitary rings in which all Sylow subgroups of the group of units are cyclic. Here, we improve this result and we classify all finite unitary rings of order pαk, where p is a prime number and p ∤ k in which R∗ has one subgroup of order p. Precisely, we prove the following theorem:

Theorem. Let R be a unitary ring of ﬁnite cardinality pβk, where p is a prime number ∗ and p ∤ k. If R has at most one subgroup of order p, then, R = A L B, where B is a ring of order k and A is a ring of cardinality pβ which is isomorphic to one of the following six types: k c ni ni Zpα , M2(GF (p)), T2(GF (2)), Zpα M(GF (p )), M2(GF (p)) M(GF (p )), i=1 i=1 s ni T2(GF (2)) M(GF (2 )), ni,k,c,s,α ∈ N. i=1 Moreover, if p = 2, then α ≤ 2.

The following are some necessary preliminary concepts and notations.

Let R be a ring with identity 1 =6 0. We denote the Jacobson radical of R by J(R),

2 and the cardinal of a given set X by |X|. The set of unit elements of R (or the group of units of R) is denoted by R∗. For a given prime number p, we denote the set of all ∗ ∗ Sylow p-subgroups of R by Sylp(R ). The prime subring of R, i.e. the subring generated by identity element of R, is denoted by R0. R0[S] is the subring generated by {S ∪ R0} where S is a subset of R. The ring of all n × n matrices over R is denoted by Mn(R). The ring of all n × n upper triangular matrices is denoted by Tn(R), and the ring of integers m modulo m is denoted by Zm. We denote the characteristic of R by Char(R), and GF (p ) denotes the unique ﬁnite ﬁeld of characteristic p and of order pm. The order of a group G is denoted by |G|, and the subgroup generated by g is denoted by hgi. Also, the order of an element g ∈ G denoted by o(g) is the order of hgi.

2 Results

From here onwards we assume that R is a ﬁnite unitary ring of order equal to a power of prime number p, unless we speciﬁcally mention otherwise. We start with the following elementary lemma:

Lemma 1. Let I be an ideal of R such that I ⊆ J(R). If Sylow p-subgroups of R∗ are ∗ R ∗ R ∗ R +I cyclic, then all Sylow p-subgroups of ( I ) are cyclic. In addition, we have ( I ) = I .

∗ R ∗ Proof. Canonical epimorphism f : R −→ ( I ) deﬁned by f(a) = a + I shows that ∗ R ∗ R +I R ∗ every Sylow p-subgroup of ( I ) is cyclic. Clearly, I ⊆ ( I ) . For the other side, let R ∗ R ∗ x + I ∈ ( I ) . Then, there exists y + I ∈ ( I ) such that xy + I =1+ I, and consequently xy − 1 ∈ I. Since I ⊆ J(R), we deduce that xy = xy − 1+1 ∈ R∗. So x ∈ R∗ and hence R∗+I ✷ x + I ∈ I .

Remark 1. Let R = A L B be a ﬁnite ring, where A and B are two ideals of R. ∗ ∗ ∗ Then, R = A L B and 1 = 1A +1B, where 1A and 1B are identity elements of A ∗ ∗ ∗ ∗ and B, respectively. It is clear that A +1B ≤ R and A +1B ∼= A . In addition, if p | gcd(|A∗|, |B∗|) for some prime number p, then by Cauchy Theorem, R∗ has two elements a +1B and 1A + b with the same order p. Clearly, ha +1Bi= 6 h1A + bi and this impleis that Sylow p−subgroups of R∗ are not cyclic.

3 We also need the following lemma, which is a direct consequence of Lemma 1.1 of [10].

∗ Lemma 2. Let R be a ﬁnite unitary ring of odd cardinality. Then, R = R0[R ].

∗ In the ﬁrst step, we characterize all unitary ﬁnite rings R with R = R0[R ] in which all Sylow p−subgroups are cyclic.

∗ ∗ Proposition 3. Suppose that R = R0[R ]. If all Sylow p−subgroups of R are cyclic, ∼ ∼ then R is a commutative ring or R = M2(GF (p)) or R = M2(GF (p)) L B, where B is a direct sum of ﬁnite ﬁelds of characteristic p.

Proof. We proceed by induction on |R| = pβ. If β < 3, then by Lemma of [7] (page 512), R is commutative. So suppose that β ≥ 3. We may assume that R is not commutative. ∗ If there exists proper unitary subring S of R such that S = S0[S ], then by induction ∼ ∼ hypothesis, S is a commutative ring or S = M2(GF (p)) or S = M2(GF (p)) L B, where B is a direct sum of ﬁnite ﬁelds of characteristic p. We have two cases with respect to the Jacobson radical, either J(R)=0or J(R) =6 0:

Case 1. If J(R) = 0, then R is a simple artinian ring, and so by the structure theorem of R ∼ t M D , D n > Artin-Wedderburn, we have = Li=1 ni ( i) where any i is a ﬁnite ﬁeld. If i 2 for some 1 ≤ i ≤ t, then Sylow p−subgroup of R has order bigger than p. Therefore, it is not cyclic, which is a contradiction. Since R is not commutative, we may assume that

n1 =2. If ni > 1 for some 2 ≤ i ≤ t, then Sylow p−subgroup of R has order bigger than p. ∼ Therefore, it is not cyclic, which is a contradiction. Consequently, R = Mn1 (GF (p)) L B, where B is a direct sum of ﬁnite ﬁelds of characteristic p.

Case 2. Suppose that J(R) =6 0. Let I ⊆ J(R) be a minimal ideal of R and let char(I)= pi. If i> 1, then Ip is a non-trivial ideal of R, and so I = Ip. Let s ∈ I. Then, i−1 i i−1 s = P vp for some v ∈ I. It follows that sp = P vp = 0, and hence char(I) = p , a contradiction. Hence, char(I) = p. Clearly, I2 = 0. Thus, for all s ∈ I, we have (1+s)p = 1. Therefore 1+I is an elementary abelian p−group. Since Sylow p−subgroups of R∗ are cyclic, we have |1+ I| = |I| = p. Hence, I = {0, a, 2a, 3a, . . . , (p − 1)a} for any |R| non-zero element a ∈ I. On the other hand, |annR(a)| = p . Then, for each x ∈ R there

4 ∗ exists an integer 1 ≤ i ≤ p such that x + i ∈ annR(a). Let P ∈ Sylp(R ), and let P = hzi. Since 1 + I ✁ R∗, by Normalizer-Centralizer Theorem, there exists a monomorphism f P Aut I ∼ C I Z P from CP (1+a) into (1 + ) = p−1, and consequently, 1 + ≤ ( ). Let x ∈ R∗ such that gcd(o(x),p) = 1. Since x−1(1+a)x ∈ I, there exists 1 ≤ j ≤ p−1 such that x−1(1+ a)x = j(1+ a). So, x−(p−1)(1+ a)xp−1 = jp−1(1+ a)=1+ a, and hence p p ∗ (1 + a) ∈ CR∗ (x ) = CR∗ (x ). Thus, 1 + a ∈ Z(R ), and so a ∈ Z(R). Consequently, ∗ ∗ 1+ I ≤ Z(R ) and I ⊆ Z(R), because R = R0[R ].

First suppose that 1 + I =6 P . Since zj a =6 0 for all integers j and I = {0, a, . . . , (p − 1)a}, there exists an integer 2 ≤ i ≤ p − 1 such that zia = a. Then, (zi − 1)a = 0, and so i z − 1 ∈ annR(a). Since a ∈ Z(R), we deduce that annR(a) is an ideal of R. Also, there i exists an integer 1 ≤ j ≤ p−1 such that z −j ∈ annR(a). It follows that j −1 ∈ annR(a), and so ji − 1=0(mod p). Consequently, p | i, which is a contradiction. Therefore, 1+ I = P . Since 1 + J(R) ≤ P , we have J(R) = I and 1+ J(R)= P . By Wedderburn Structure Theorem, R ∼ t M (D ), where any D is a finite field. By induction J(R) = Li=1 ni i i R R ∼ M GF p R ∼ M GF p B, hypothesis, J(R) is commutative or J(R) = 2( ( )) or J(R) = 2( ( )) L where B is a direct sum of finite fields of characteristic p.

R ∼ M GF p R ∼ M GF p B, B First suppose that J(R) = 2( ( )) or J(R) = 2( ( )) L where is a direct R ∼ some of ﬁnite ﬁelds of characteristic p. Let A be an ideal of R such that A = M2(GF (p)). 4 R ∗ Then, |R| = |A|p . Let z ∈ ( A ) with o(z + A) > 1. We have az ∈ J(R)= {0, a, . . . , (p − R 1)a}. Therefore, az = ja, where 1 ≤ j ≤ p − 1, and so z − j ∈ annR(J(R)). Since A is a

simple ring and annR(a) is a two sided ideal (a ∈ Z(R)), we deduce that annR(a) ⊆ A. |R| On the other hand, we have |annR(a)| = p , and so we obtain that |R| ≤ p|A|, a R contradiction. So, J(R) is commutative. Let x, y ∈ R∗ such that p ∤ o(x) and p ∤ o(y). Since xy − yx ∈ J(R), we have [x, y] − 1 ∈ J(R). Consequently, [x, y] ∈ Z(R∗), and R∗ is a nilpotent group. It follows ∗ ∗ that R is a direct sum of Sylow subgroups. Let x, y ∈ Q ∈ Sylq(R ), where p ∤ q. Also, let xy − yx = ja and xy = cyx, where c ∈ Z(R∗). Then, we have cyx − yx = ja. Hence (c − 1)yx = ja, and so jax−1y−1 = c − 1 ∈{1, 2,...,p}. Thus, o(c) | p, and then xpy = cpyxp = yxp. Since gcd(p,o(x)) = 1, we have xy = yx. Consequently, R∗ is an abelian group, and hence R is commutative, which is our ﬁnal contradiction. ✷

5 Let p be a prime number. For simplicity, let Γ(p) be the set of all ﬁnite rings A of the following types:

(a) If p> 2, then A is a direct sum of finite fields of order p or a direct sum of finite

fields of order p and Zpm or a direct sum of finite fields of order p and M2(GF (p)) or

M2(GF (p)).

(b) If p = 2, then A is a direct sum of finite fields of order 2 or a direct sum of finite

fields of order 2 with Z4 or a direct sum of finite fields of order 2 with M2(GF (2)) or

M2(GF (2)).

∗ ∗ Proposition 4. Let R = R0[R ]. If all Sylow p-subgroups of R are cyclic, then R ∈ Γ(p).

Proof. By Proposition 3, we may assume that R is a commutative ring. Let |R| = pβ. Then, we proceed by induction on β. First suppose that |R| = p2. If R is not a ﬁeld, then we have two possible isomorphism classes Zp2 and Zp L Zp. Clearly, R ∈ Γ in this case. Hence we may assume that β > 2. We consider two cases depending on the Jacobson radical: J(R)=0or J(R) =6 0.

Case 1. First, let J(R) = 0. Since R is a semi-simple ring, by Wedderburn Structure R ∼ k R R Theorem, = Li=1 i is a direct product of matrix algebras over division algebras. αi is a commutative ring, so Ri is a ﬁnite ﬁeld for each i. On the other hand, |Ri| = p for ∗ some integer αi(αi ≤ t), and hence we have |Ri | ≥ p − 1. Since the Sylow p−subgroup is

cyclic, |Ri| >p at most for one i, and so R ∈ Γ(p).

Case 2. Let J(R) =6 0, and I ⊆ J(R) be a minimal ideal of R. Similar to the proof of Proposition 3 case 2, we can prove that, char(I)= p, I2 =0 and 1+ I is an elementary abelian p−group. Since Sylow p−subgroups of R∗ are cyclic, we have |1+I| = |I| = p. So, ∗ I = {0, a, 2a, 3a, . . . , (p−1)a} for any non-zero element a belonging to I. Let P ∈ Sylp(R ) and P = hzi. Since zia =6 0 for all integers i and |I| = p, we have zp = 1, and it follows that |P | = p. Also, 1 + J(R) ≤ P , hence J(R)= I and 1+ J(R)= P .

Let {M1,...,Me} be the set of all maximal ideals of R. By the structure theorem of Artin-Wedderburn, there exists an isomorphism f from e R to R . For each x ∈ R, Li=1 Mi J(R) j p xa ja x j ann a |R| p there exists 1 ≤ ≤ such that = , and hence − ∈ R( ). Thus, |annR(a)| = ,

6 and so annR(a) is a maximal ideal of R. We may assume that M1 = annR(a).

annR(x)+J(R) |R| Let f((1 + M1, M2, M3,...,Mk)) = x + J(R). It is clear that | J(R) | = p|J(R)| , Rx R p J R ax a ann x Jac R and so | | = | annR(x) | = | ( )|. Since =6 0, we have 6∈ R( ). Also, ( ) is a minimal ideal of R and a 6∈ annR(x), so annR(x) ∩ J(R) = 0. Since |R| = |annR(x)|p, we deduce that R = annR(x) L Rx. By induction hypothesis, annR(x) and Rx belong ∗ ∗ to the set Γ(p). Evidently, R ∈ Γ(p), because gcd((|annR(x)) |,p)= gcd(|(Rx) |,p) = 1.

R Hence we may assume that e = 1. Then, M1 = J(R) and J(R) is a ﬁnite ﬁeld. It is easy to see that every Sylow q−subgroup of R∗ is cyclic for all q ||R∗|, and consequently by main result of [1], R ∈ Γ(p). ✷

Now, by Lemma 2, it remains to characterize the ring of order 2β in which all its Sylow ∗ 2-subgroups are cyclic and R =6 R0[R ].

β ∗ Proposition 5. Let R be a unitary ring of ﬁnite cardinality 2 and H = R0[R ] such that every Sylow 2-subgroup of R∗ is a cyclic group.

(a) If R is a commutative ring, then R ∈ Γ(2).

(b) If H is a commutative ring and R is a non-commutative ring, then either R ∼= T GF R ∼ T GF k GF ni k 2( (2)) or = 2( (2)) Li=1( (2 )) for some positive integer .

Proof. We proceed by induction |R|. Let I ⊆ J(R) be a minimal ideal of R. Since Sylow 2−subgroups of R∗ are cyclic, we have |1+ I| = |I| = 2. Therefore, I = {0, a} for ∗ any non-zero element a belongs to I. Let P ∈ Sylp(R ), and let P = hzi. Then, since zia =6 0 for all integers i and |I| = 2, we have z2 = 1 and it follows that |P | = 2. But since |R| 1+J(R) ≤ P , we deduce that J(R)= I and 1+J(R)= P . It is clear that |annR(a)| = 2 and by Proposition of [5], every unitary non-commutative ring of order 8 is isomorphic R to T2(GF (2)). So we may assume that |R| > 8. Also, J( I ) = 0. Thus by Wedderburn R ∼ k R Structure Theorem, I = Li=1 i is a direct product of matrix algebras over division ∗ algebras, and since R0[R ] is commutative, Ri is a ﬁnite ﬁeld for each i =1, 2,...,k.

Let {M1,...,Mk} be the set of all maximal ideals of R. If k = 1, then J(R) = R M1 = annR(a), because I is commutative. Since [R : annR(a)] = 2, we have R =

7 ∗ R0[(1 + J(R))] = R0[R ], which is a contradiction. So k > 1. We may assume that

M1 = annR(a).

Let f be an ismorphism from R R ... R to R . Let M1 L M2 L L Mk J(R) f((1 + M1, M2, M3,...,Mk)) = x + J(R). Since I is the unique minimal ideal of R annR(x)+J(R) |R| and a 6∈ annR(x), we have annR(x) ∩ J(R) = 0. It is clear that | J(R) | = 2|J(R)| . Rx R J R ax a ann x Consequently, | | = | annR(x) | = 2| ( )|. Since =6 0, we have 6∈ R( ). Also, I = J(R). So I ⊆ Rx, and hence R = annR(x) L Rx. Clearly, x is the identity element of Rx. Hence (x + a)2 = x2 = x, and so x + a is an

element of Rx of order 2. Let y ∈ annR(x) be the identity element of annR(x) and let 2 u ∈ annR(x) such that u = y. Since o(u +(x + a)) = o(y +(x + a)) = 2 and the Sylow

2−subgroup is cyclic, we have u + x = y + x + a, and hence u − y = a ∈ annR(x), a ∗ contradiction. Therefore, |(annR(x)) | is an odd number. Since J(annR(x)) = 0, we have

annR(x) ∈ Γ(2). ∼ If R is a commutative ring, then R = Z4 L annR(x) ∈ Γ(2). If R is not a commutative ring, then Rx is not a commutative ring, because otherwise x(ax)=(ax)x = ax2 = x, and so (xa − 1)x = 0, consequently, x = 0, which is a contradiction. So by induction Rx ∼ T GF R ∼ T GF m GF ni hypothesis, = 2( (2)) or = 2( (2)) Li=1( (2 )) for some positive integer m. ✷

Theorem 6. Let R be a unitary ring of ﬁnite cardinality pβ, where p is a prime. If all Sylow p-subgroups of R∗ are cyclic, then, R is isomorphic to one of the following six types:

k c ni ni Zpα , M2(GF (p)), T2(GF (2)), Zpα M(GF (p )), M2(GF (p)) M(GF (p )), i=1 i=1 s ni T2(GF (2)) M(GF (2 )), ni,k,c,s,α ∈ N. i=1 Moreover, if p =2, then α ≤ 2.

Proof. The proof is clear by Propositions 3 and 5. ✷

Every element of order two in any group is called an involution. The following lemma helps us to classify all ﬁnite unitary rings with a single involution.

8 β ∗ ∗ Lemma 7. Let R be a unitary ring of ﬁnite cardinality 2 and H = R0[R ] such that R has only one involution. Then, all Sylow 2−subgroups of R∗ are cyclic.

∗ Proof. We proceed by induction on |R|. Let P ∈ Syl2(R ). The 2-groups with unique involution were determined by Burnside; they must be cyclic or generalized quaternion groups. Hence, if R is commutative, then Sylow 2-subgroup is cyclic. So suppose that R is a non-commutative ring. Then,

2n 4 n 2 −1 −1 P = Q4n = h{u,l : u = l =1,u = l ,l ul = u }i.

The case |R| =23, is clear by Proposition of [5]. Assume that |R| > 23.

First, let J(R) = 0. Since R is a semi-simple ring, by Wedderburn Structure Theorem, R ∼ k R = Li=1 i is a direct product of matrix algebras over division algebras. On the other hand, R is not commutative. So, we may assume that n1 > 1, and in this case, the number of involutions of R is more than one, which is a contradiction. Hence J(R) =6 0. Since 1+ J(R) is a 2-group, we have |J(R)| ≤ 4n. Now, let I ⊆ J(R) be a minimal ideal of R. It is easy to prove that 1 + I is an elementary abelian 2−group, and since R∗ has only one involution, we have |1+ I| = |I| = 2. Hence, I = {0, a} for any non-zero element a |R| ∗ R belonging to I. Clearly, |annR(a)| = 2 and 1+ a ∈ Z(R ). If |P | ≥ 16, then I has only P +I ∼ P +I one involution, because I = Q2(n−1), and hence by induction hypothesis, I is cyclic, which is a contradiction. So |P | ≤ 8.

Let {M1,...,Mk} be the set of all maximal ideals of R. If k = 1, then J(R)= M1. If R ∗ | J(R) | > 2, then by Lemma 1.1 of [10] we have R = R0[R ], and so a ∈ Z(R). It follows ann a R ann a Jac R R that R( ) is a two sided ideal of , and so R( ) = ( ). But | annR(a) | = 2, R which is a contradiction. So J(R) is a ﬁnite ﬁeld. Let w be a generator for the cyclic R ∗ R group ( J(R) ) . Since w − 1 ∈ annR(a), we have | J(R) | = 2, and so J(R) = annR(a). Hence |R| ≤ 16. If |J(R)| ≤ 4, then |R| ≤ 8, which is a contradiction. So |J(R)| = 8, R R and then |J( I )| =4. If I is not commutative, then by Proposition of [5], every unitary

non-commutative ring of order 8 is isomorphic to T2(GF (2)). But |J(T2(GF (2))| = 2, R G ∼ which is a contradiction. Therefore I is commutative. Since J(R) = GF (2), we have u + l,u3 + l ∈ J(R) \ Z(R), and consequetly u + l + 1 and u3 + l + 1 have order 4. We have P = hui∪hli∪huli. It follows from P is not an abelian group that, u+l+1,u3+l+1 ∈huli.

9 Thus, u + l +1= ul or u3 + l +1= ul. If u + l +1= ul, then u(l − 1) = l +1, and so lu(l − 1) = u(l − 1)l. Hence, we have lu = ul, which is a contradiction. If u3 + l +1= ul, then (u − 1)l = u3 + 1, and hence u(u − 1)l =(u − 1)lu.Thus u2l = ulu, and so lu = ul, which is a contradiction. So k > 1.

R R R R We may assume that M = annR(a). Let f : ... ∼= and 1 M1 L M2 L L Mk J(R) let f(1 + M1, M2, M3,...,Mk) = x + J(R). Then ax =6 0 imply that a 6∈ annR(x).

Since I is the unique minimal ideal of R, we conclude that annR(x) ∩ J(R) = 0. But annR(x)+J(R) |R| RxR R J R ax RxR | J(R) | = 2|J(R)| and consequently, | | = | annR(x) | = 2| ( )|. So ∈

implies that I ⊆ RxR. Since annR(x) ∩ RxR is a two sided ideal of R, we have annR(x) ∩

RxR = 0 or I ⊆ annR(x) ∩ RxR. The later case is impossible, so annR(x) ∩ RxR = 0 and then R = annR(x) L RxR. Hence annR(x) and RxR are unitary rings. First assume ∗ that |(annR(x)) | is not an odd number. If h is the identity element of RxR, then 2 2 (h + a) = h = h, and so h + a is an element of RxR of order 2. Let y ∈ annR(x) be the 2 identity element of annR(x) and let u ∈ annR(x) such that u = y. Since there is only one involution and o(u +(h + a)) = o(y +(h + a)) = 2, we see that u + h = y + h + a. ∗ Consequently, u − y = a ∈ annR(x), which is a contradiction. Therefore, |(annR(x)) | is an odd number, and then by induction hypothesis, Sylow 2-subgroupof(RxR)∗ is cyclic, which is a contradiction. ✷

Now, we can prove our main theorem.

α1 αk Proof of Theorem 1. Let |R| = p1 ...pk be the canonical decomposition of |R| into its prime divisors pi. Then R = R1 L R2 L ... L Rk, where each Ri is an ideal of αi m ∗ order pi . Hence we may assume that |R| = p . Let P ∈ Sylp(R ). If p > 2, then P is cyclic and the proof is clear by Theorem 6. If p = 2, then P is cyclic or P is generalized quaternion groups. The latter case is impossible by Lemma 7. Hence P is cyclic, and so we have the result by Theorem 6.

10 References

[1] M. Amiri, M. Ariannejad, Finite unitary rings in which all sylow subgroups of the group of units are cyclic, Bull. Aust. Math. Soc. Vol 99, Issue 3, (2019) 403-412.

[2] J. B. Bateman and D. B. Coleman, Group Algebras with Nilpotent Unit Groups, Proc. Math. Soc.1 9(1968),448-449.

[3] P.B.Bhattacharya and S. K. Jain, Rings with Solvable Adjoint Groups, Proc. Amer. Math. Soc. (1970), 563-565

[4] P.B.Bhattacharya and S. K. Jain, A Note on the Adjoint Group of a Ring, Archiv der Math. Basel, (1070), 366-368

[5] D.B. Erickson, Orders for ﬁnite noncommutative rings, The American Mathematical Monthly, 73 (1966) 376-377.

[6] D. Dolzan, Nilpotency of the group of units of a ﬁnite ring, Bull. Aust. Math. Soc, 79 (2009) 177-182.

[7] K.E. Eldridge, Orders for ﬁnite noncommutative rings with unity, The American Mathematical Monthly, 75, 5 (1968) 512-514.

[8] B. Farb, R. Keith Dennis, Noncommutative algebra, Graduate texts in mathematics, Springer Verlag, New York, 144, 1993.

[9] B. Fine, Classification of Finite Rings of Order p2, Fairfield University, Fairfield, CT 06430.

[10] G. Groza, Artinian rings having a nilpotent group of units, J. Algebra, 121 (1989) 253-262.

[11] T.Y. Lam, A ﬁrst course in noncommutative rings, Gratuate texts in mathematics, Springer Verlag, New York, 2nd ed., 2001.

[12] D.J.S. Robinson, A course in the theory of groups, Springer Verlag, New York, 1980.

11 [13] J. F. Waters, On the Adjoint Group of a Radical Ring, J. London Math .Soc. (1968),725-729.

[14] J.H.M. Wedderburn, A theorem on ﬁnite algebra, Trans. Amer. Math. Soc., 6(1905)349-352. 1996.