6.1 Introduction

Throughout the chapter, p denotes a prime number. For group G, we denote by G, Z(G), cl(G), d(G), Φ(G), and Aut(G), respectively, the commutator subgroup, the center, the nilpotency class, the rank, the Frattini subgroup, and the automorphism group of G. An automorphism σ of group G is called central if σ commutes with every automorphism in Inn(G), the group of inner automorphisms of G, (equivalently, if g−1σ(g) lies in the center Z(G) of G, for all g in G.)

The central automorphisms of G fix the commutator subgroup of G elementwise and form a normal subgroup of the full automorphism group Aut(G); we denote this subgroup by Autz(G) in this paper. For groups G having Aut(G) abelian, it is necessarily the case that Autz(G) = Aut(G). The non-abelian groups G with Aut(G) abelian are called as Miller groups (see [19]). However, several people constructed various groups G for which Aut(G) is non-abelian and Autz(G) = Aut(G) (see [7, 11, 15, 18]). In 2001, Curran and McCaughan [6] considered the case where the central automorphisms are just the inner automorphisms of G, that is, Autz(G) = Inn(G); one can also see [4, 23]. Continuing in this direction, in 2004, Curran [8], for group G, derived the equality \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}},\); the same is derived in [1, 12, 22]. Let \({Aut_z^{z}}(G)\) be the set of all central automorphisms of a group G which fixes the center Z(G) of G elementwise. In 2007, Attar [2] characterized finite \( \operatorname {\mathrm {p-groups}}\) for which \({Aut_z^{z}}(G)={Inn}(G)\) holds. In 2009, Yadav [25] characterized \( \operatorname {\mathrm {p-groups}}\) of nilpotency class 2 for which \({Aut_z}(G)={Aut_z^{z}}(G)\) (for the same equality, also see [14]).

An \( \operatorname {\mathrm {automorphism}}\) ϕ of a group G is called class preserving if ϕ(x) is conjugate to x for all x ∈ G. The set Autc(G) of all class-preserving automorphisms of G forms a normal subgroup of Aut(G) and contains Inn(G). In 2013, Yadav [26] characterized finite \( \operatorname {\mathrm {p-groups}}\), and Kalra and Gumber [16] characterized all finite \( \operatorname {\mathrm {p-groups}}\) of order ≤ p6 (for any prime p) and ≤ p5 (for odd prime p) for which the set of all central automorphisms is equal to the set of all class-preserving automorphisms, that is, Autz(G) = Autc(G); the same equality is derived in [10].

An automorphism σ of a group G is called IA − automorphism if it induces the identity automorphism on the abelian quotient GG. Let IAz(G) be the group of those IA automorphisms which fix the center of G elementwise. In 2014, Rai [21] characterized finite \( \operatorname {\mathrm {p-groups}}\) in which Autz(G) = IAz(G) if and only if γ2(G) = Z(G). In 2016, Kalra and Gumber [17], characterized finite non-abelian \( \operatorname {\mathrm {p-groups}}\) G for which Autz(G) = IAz(G) if and only if G = Z(G).

Hegarty [13] defined the notions of absolute center and autocommutator of a group G (analogous to Z(G) and G as follows:

$$\displaystyle \begin{aligned} L(G)=\{ g\in G|\operatorname{\mathrm{\alpha}}(g)=g\,\forall \operatorname{\mathrm{\alpha}}\in {Aut}(G) \} \end{aligned}$$
$$\displaystyle \begin{aligned} G^*=\langle g^{-1}\operatorname{\mathrm{\alpha}}(g)| g\in G,\, \operatorname{\mathrm{\alpha}}\in{Aut}(G)\rangle\end{aligned}$$

These are clearly characteristic subgroups of G. Also, Z(G) ⊃ L(G) and G⊂ G. Hegarty [13] also defined absolute central \( \operatorname {\mathrm {automorphism}}\) of G as follows: an \( \operatorname {\mathrm {automorphism}}\) γ of a group G is called an absolute central \( \operatorname {\mathrm {automorphism}}\) if it induces identity \( \operatorname {\mathrm {automorphism}}\) on GL(G). The set Autl(G) of all absolute central automorphisms of G forms a normal subgroup of Aut(G); it is also a subgroup of Autz(G). Let \({Aut_{l}^z}(G)\) denote the group of absolute central automorphisms of G which fix Z(G) elementwise.

In 2020, Singh and Gumber [24] gave a necessary and sufficient condition on finite \( \operatorname {\mathrm {p-group}}\) G for which Autz(G) = Autl(G) and also for which \({Aut_z}(G)={Aut_{l}^z}(G)\).

6.2 Equalities of Central Automorphisms

6.2.1 Equalities with Group of All Automorphisms

Definition 6.1

Following Earnley, a non-abelian group with abelian automorphism group is called Miller group.

If Aut(G) is abelian, then it is clear that all the automorphisms are central, i.e., Autz(G) = Aut(G). The obvious examples of groups with abelian automorphism group are the cyclic groups. There are non-abelian groups with abelian automorphism groups; these are called Miller groups (seeEarnley [9]). Several researchers constructed various examples of groups for which Autz(G) = Aut(G), even if Aut(G) is non-abelian. Curran, in 1982, found first such example. He constructed a group of order 27 for which Autz(G) = Aut(G) and Aut(G) is non-abelian.

Theorem 6.1 ([7], Proposition, p. 394)

There exists a non-abelian group G of order 27 which has a non-abelian automorphism group of order 212 in which every automorphism is central, that is, Autz(G) = Aut(G).

Example of such group is given below:

Let M be the Miller group of order 26, and let

$$\displaystyle \begin{aligned} G=M\times Z_2=\langle a,b,c,d\,|\,a^8=b^4=c^2=d^2=1,\, a^b=a^5,\,b^c=b^{-1},\,\end{aligned}$$
$$\displaystyle \begin{aligned}{}[a,c]=[a,d]=[b,d]=[c,d]=1 \rangle\end{aligned}$$

This result of Curran leads the motivation to \( \operatorname {\mathrm {p-groups}}\) for p an odd prime in which Autz(G) = Aut(G) and Aut(G) is non-abelian. In 1984, Malone proved the following result:

Theorem 6.2 ([18], Proposition, p. 36)

For each odd prime p, there exists a non-abelian p-group with a non-abelian automorphism group in which each automorphism is central, that is, Autz(G) = Aut(G).

For each odd prime p, we consider the group

$$\displaystyle \begin{aligned} F=\langle a_1,\,a_2,\,a_3,\,a_4,\,|\,(a_i,\,a_j,\,a_k)=1\, and \,a_i^{p^2}=1 \,\,for\end{aligned}$$
$$\displaystyle \begin{aligned} 1\leq i,j,k\leq 4;\, (a_1,a_2)=a_{1}^{p};\,(a_1,a_3)=a_{3}^{p};\,(a_1,a_4)=a_{4}^{p}\end{aligned}$$
$$\displaystyle \begin{aligned} (a_2,a_3)=a_{2}^{p};\,(a_2,a_4)=1\,;(a_3,a_4)=a_{3}^{p}\rangle\end{aligned}$$

Aut(F) is abelian group. We set B = 〈 b| bp = 1 〉. Group G = F × B is non-abelian group which has Aut(G) non-abelian in which each automorphism is central.

Curran in [7] and Malone in [18] derived the examples of groups with direct factors for which Autz(G) = Aut(G) and Aut(G) is non-abelian. The question was left if there is a group G with no direct factors for which Autz(G) = Aut(G) and Aut(G) is non-abelian. Continuing in this direction, in 1986, Glasby produced an infinite family of 2-groups having no direct factors and which have a non-abelian automorphism group in which all automorphisms are central.

Definition 6.2

Define Gn to be the group generated by x1, …, xn \(x_{i}^{2^{i}}=1 \,\,\,\, (1 \leq i \leq n)\) \([ x_{i},x_{i+1}]=x_{i+1}^{2^{i}},\,\,\,\,(1 \leq i < n) \). [xi, xj] = 1, (1 < i + 1 < j ≤ n)

Theorem 6.3 ([11], Theorem, p. 234)

For n ≥ 3, Gn has no direct factors, and Aut(Gn) is non-abelian of order 2p(n), where p(n) = (n − 1)(2n2 − n = 6∕6)(n ≥ 4), in which every \( \operatorname {\mathrm {automorphism}}\) is central.

In 2012, Jain and Yadav [15] constructed the following family of groups Gn with no direct factor, for which Autz(Gn) = Aut(Gn).

Definition 6.3

Let n be a natural number greater than 2 and p an odd prime. Define Gn to be the group generated by x1, x2, x3, x4

$$\displaystyle \begin{aligned} x_{1}^{p^{n}}=x_{2}^{p^{3}}=x_{3}^{P^{2}}=x_{4}^{p^{2}}=1,\end{aligned}$$
$$\displaystyle \begin{aligned}{}[x_{1},x_{2}]=x_{2}^{p^{2}},\,\,\,\,[x_{1},x_{3}]=x_{3}^{p}\end{aligned}$$
$$\displaystyle \begin{aligned}{}[x_{1},x_{4}]=x_{4}^{p},\,\,\,\,[x_{2},x_{3}]=x_{1}^{p^{n-1}}\end{aligned}$$
$$\displaystyle \begin{aligned}{}[x_{2},x_{4}]=x_{2}^{p^{2}},\,\,\,\, [x_3,x_4]=x_4^p .\end{aligned}$$

This group G is a regular p-group of nilpotency class 2 having order pn+7 and exponent pn. Further, Z(G) =  Φ(G) and therefore G is purely non-abelian.

Theorem 6.4 ([15], Theorem A, p. 228)

Let m = n + 7 and p be an odd prime, where n is a positive integer greater than or equal to 3. Then there exists a group G of order pm, exponent pn, and with no nontrivial abelian direct factor such that Autz(G) = Aut(G) is non-abelian.

6.2.2 Equalities with Group of Inn(G) and \( \operatorname {\mathrm {Z(Inn(G))}}\)

In 2001, Curran and McCaughan [6] characterized finite \( \operatorname {\mathrm {p-groups}}\) in which central automorphisms are precisely the inner automorphisms.

Theorem 6.5 ([6], Theorem, p. 2081)

If G is a finite p-group, then Autz(G) = Inn(G) if and only if G = Z(G) and Z(G) is cyclic.

Definition 6.4

A group G, whose only element of finite order is the identity, is called torsion-free group.

Definition 6.5

A non-abelian group G is purely non-abelian if G has no nontrivial abelian direct factor.

In 2016, Azhdari characterized all finitely generated groups G for which the equality Autz(G) = Inn(G) holds. He proved the following:

Theorem 6.6 ([4], Theorem 2, p. 4134)

Let G be a finitely generated group. Then Autz(G) = Inn(G) if and only if one of the following assertion holds:

  • G is purely non-abelian and Z(G) = G is cyclic.

  • GC2 × N where N is purely non-abelian with |Z(N)| odd and Z(N) = N is cyclic (or Z(G) = C2 × G is cyclic).

  • G is torsion-free with Z(G) = G is cyclic and det(MG) = 1 where MG is skew-symmetric matrix corresponding to G.

In 2018, Sharma et al. [23] verified the equality Autz(G) = Inn(G) for the finite p-groups of order up to p7 as follows:

Theorem 6.7 ([23], Theorem 2.1, p. 3)

There is no \( \operatorname {\mathrm {p-group}}\) G of order up to p6 satisfying Autz(G) = Inn(G).

Theorem 6.8 ([23], Theorem 2.2, p. 3)

A \( \operatorname {\mathrm {p-group}}\) G of order p7 satisfies Autz(G) = Inn(G) if and only if \(Z(G)\cong C_p^2\), |G| = p4 and cl(G) = 4.

In 2004, Curran [8] considered the case where the central automorphism group is as small as possible. Clearly, \( \operatorname {\mathrm {Z(Inn(G))}} \leq {Aut_z}(G)\), for any group G. When G is arbitrary, Autz(G) and \( \operatorname {\mathrm {Z(Inn(G))}}\) may coincide because both these subgroups of Aut(G) can be trivial. However, the situation becomes interesting if G is a p-group, since both subgroups are nontrivial.

Theorem 6.9 ([8], Theorem 1.1, p. 223)

Let G be a finite non-abelian p-group. If \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\), then Z(G) ≤ G, and furthermore, \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if \( \operatorname {\mathrm {Hom}}(G/G,Z(G)) \approx Z(G/Z(G))\).

In 2013, Sharma and Gumber [22] characterized p-groups of order ≤ p5(for any prime p) and of order p6(for p odd), for which \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\).

Theorem 6.10 ([22], Theorem 3.2, p. 3)

Let G be p-group of order p5 and cl(G) = 3. Then \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if d(G) = 2 and Z(G)≅Cp.

Theorem 6.11 ([22], Theorem 3.3, p. 3)

Let G be a p-group of order p6, for an odd prime p, and cl(G) = 3 0r4. Then \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if d(G) = 2 and Z(G)≅Cp.

Continuing the study of Curran [8] of minimum order of Autz(G), Gumber and Kalra [12] obtained the following:

Let \(G/G^{\prime } \cong C_{p^{r_1}}\times \dots C_{p^{r_{n}}}\) (r1 ≥⋯ ≥ rn ≥ 1) and \(Z_{2}(G)/Z(G) \cong C_{p^{s_1}}\times \dots C_{p^{s_{m}}}\) (s1 ≥⋯ ≥ sm ≥ 1).

Theorem 6.12 ([12], Theorem 2.1, p. 1803)

Let G be a finite p-group with \(Z(G)\cong C_{p^{b_1}}\). Then \({Aut_z} (G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if either GGZ2(G)∕Z(G) or d(G) = d(Z2(G)∕Z(G)), si = b1 for 1 ≤ i  c, and si = ri for c + 1 ≤ i  n, where c, 1 ≤ c  n is the largest such that rc ≥ b1.

Definition 6.6

The coclass of a finite \( \operatorname {\mathrm {p-group}}\) G of order pn is n − c, where c is the class of the group.

Corollary 6.1 ([12], Corollary 2.2, p. 1804)

Let G be a finite p-group of coclass 2. Then \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if Z(G)≅Cp and d(G) = d(Z2(G)∕Z(G)) = 2.

Corollary 6.2 ([12], Corollary 2.3, p. 1804)

Let G be a finite p-group of coclass 3. Then \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if Z(G)≅Cp and d(G) = d(Z2(G)∕Z(G)) = 2, 3 or \(Z(G)\cong C_{p^2}\) and Z2(G)∕Z(G)≅GG.

Corollary 6.3 ([12], Corollary 2.4, p. 1804)

Let G be a finite p-group of coclass 4. Then \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if one of the following conditions holds:

  1. (a)

    Z(G)≅Cp and d(G) = d(Z2(G)∕Z(G)) = 2, 3, 4.

  2. (b)

    \(Z(G)\cong C_{p^2}\) and either Z2(G)∕Z(G)≅GG or \(Z_2(G)/Z(G)\cong C_{p^2}\times C_p\) and \(G/G^{\prime } \cong C_{p^3}\times C_p\) or \(Z_2(G)/Z(G)\cong C_{p^2}\times C_p\) and \(G/G^{\prime } \cong C_{p^4}\times C_p\).

  3. (c)

    \(Z(G)\cong C_{p^3}\) and Z2(G)∕Z(G)≅GG.

Gumber and Kalra also generalized the results of Sharma and Gumber [22] as follows:

Theorem 6.13 ([12], Theorem 3.1, p. 1804)

Let G be p-group of order = p5 and cl(G) = 3. Then \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if Z(G)≅Cp and d(G) = d(Z2(G)∕Z(G)) = 2.

Theorem 6.14 ([12], Theorem 3.2, p. 1805)

Let G be a finite \( \operatorname {\mathrm {p-group}}\) such that cl(G) = 3 or 4. Then, \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if Z(G)≅Cp and d(G) = d(Z2(G)∕Z(G)) = 2.

Also, Gumber and Kalra obtained the result for |G| = p7 as in [22]; it was up to p6.

Theorem 6.15 ([12], Theorem 3.3, p. 1805)

Let G be a p-group of order p 7 . Then \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if one of the following holds:

  • cl(G) = 3, Z(G) ≃ Cp and rank(G) = rank(Z2(G)∕Z(G)) = 2, 3, 4.

  • cl(G) = 4 and either Z(G) ≃ Cp and rank(G) = rank(Z2(G)∕Z(G)) = 2, 3 or Z(G) is cyclic group of order p2 and Z2(G)∕Z(G) ≃ GG.

  • cl(G) = 5, Z(G) ≃ Cp and rank(G) = rank(Z2(G)∕Z(G)) = 2.

Let G be a non-abelian p-group G. Let \(G/G^{\prime }\cong C_{p^{c_{1}}}\times C_{p^{c_{2}}}\times \dots \times C_{p^{c_{r}}}\) (c1 ≥⋯ ≥ cr ≥ 1) and \(Z_{2}G/Z(G)\cong C_{p^{d_{1}}}\times C_{p^{d_{2}}}\times \dots \times C_{p^{d_{s}}}\) (d1 ≥ d2 ≥…ds ≥ 1), where \(C_{p^{a_i}}\) is a cyclic group of order \(p^{a_i}.\)

In 2020, Attar [1] characterized the finite p-groups in some special cases, including p-groups G with CG(Z( Φ(G)) ≠  Φ(G), \( \operatorname {\mathrm {p-groups}}\) with an abelian maximal subgroup, metacyclic \( \operatorname {\mathrm {p-groups}}\) with p ≥ 2, \( \operatorname {\mathrm {p-groups}}\) of order pn and exponent pn−2, and Camina \( \operatorname {\mathrm {p-groups}}\), for which Autz(G) is of minimal order, as follows:

Theorem 6.16 ([1], Theorem 3.1, p. 4)

Let G be a finite \( \operatorname {\mathrm {p-group}}\) such that CG(Z( Φ(G)) ≠  Φ(G). Then \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if Z(G) is cyclic and one of the following is true:

  • GGZ2(G)∕Z(G).

  • r = s, di = h for 1 ≤ i  t, di = ci for t + 1 ≤ i  r, where ph = exp(Z(G)) and t is the largest integer between 1 and s such that ct > h.

Corollary 6.4 ([1], Corollary 3.2, p. 5)

Let G be a non-abelian finite p-group with an abelian maximal subgroup. Then \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if G = Z(G) and Z(G) is cyclic.

Theorem 6.17 ([1], Theorem 3.3, p. 6)

Let G be a non-abelian metacyclic finite p-group with p > 2. Then \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if Z(G) ≤ G.

Corollary 6.5 ([1], Corollary 3.4, p. 6)

The finite non-abelian \( \operatorname {\mathrm {p-groups}}\) G of order p n and exponent p n−1 for which \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) are of the following isomorphism types:

  1. (1)

    \(M(p^3)=\langle \operatorname {\mathrm {\alpha }}, \operatorname {\mathrm {\beta }} | \operatorname {\mathrm {\alpha }}^{p^2}= \operatorname {\mathrm {\beta }}^p=1,\, \operatorname {\mathrm {\beta }}^{-1} \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }}= \operatorname {\mathrm {\alpha }}^{1+p}\rangle (p>2)\).

  2. (2)

    \(D_8=\langle \operatorname {\mathrm {\alpha }}, \operatorname {\mathrm {\beta }} | \operatorname {\mathrm {\alpha }}^4= \operatorname {\mathrm {\beta }}^2=1,\, \operatorname {\mathrm {\beta }}^{-1} \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }}= \operatorname {\mathrm {\alpha }}^{-1}\rangle \).

  3. (3)

    \(Q_8=\langle \operatorname {\mathrm {\alpha }}, \operatorname {\mathrm {\beta }} | \operatorname {\mathrm {\alpha }}^4=1,\, \operatorname {\mathrm {\beta }}^2= \operatorname {\mathrm {\alpha }}^2,\, \operatorname {\mathrm {\beta }}^{-1} \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }}= \operatorname {\mathrm {\alpha }}^{-1}\rangle \).

Corollary 6.6 ([1], Corollary 3.5, p. 7)

Let p be an odd prime. Then finite non-abelian \( \operatorname {\mathrm {p-groups}}\) of order p n and exponent p n−2 for which \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) are one of the following isomorphism types:

  1. (1)

    \(G=\langle \operatorname {\mathrm {\alpha }},\, \operatorname {\mathrm {\beta }},\, {\gamma }| \operatorname {\mathrm {\alpha }}^p= \operatorname {\mathrm {\beta }}^p={\gamma }^p=1,\, \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }}= \operatorname {\mathrm {\beta }} \operatorname {\mathrm {\alpha }},\, {\gamma }^{-1} \operatorname {\mathrm {\alpha }}{\gamma }= \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }},\, \operatorname {\mathrm {\beta }}{\gamma }={\gamma } \operatorname {\mathrm {\beta }} \rangle \).

  2. (2)

    \(G=\langle \operatorname {\mathrm {\alpha }},\, \operatorname {\mathrm {\beta }} | \operatorname {\mathrm {\alpha }}^{p^3}= \operatorname {\mathrm {\beta }}^{p^2}=1,\, \operatorname {\mathrm {\beta }}^{-1} \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }}= \operatorname {\mathrm {\alpha }}^{1+p}\rangle \).

  3. (3)

    \(G=\langle \operatorname {\mathrm {\alpha }},\, \operatorname {\mathrm {\beta }} | \operatorname {\mathrm {\alpha }}^{p^4}= \operatorname {\mathrm {\beta }}^{p^2}=1,\, \operatorname {\mathrm {\beta }}^{-1} \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }}= \operatorname {\mathrm {\alpha }}^{1+{p^2}}\rangle \).

Corollary 6.7 ([1], Corollary 3.6, p. 8)

The finite non-abelian 2-groups G of order 2n and exponent 2n−2 for which \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) are one of the following:

  1. (1)

    \(G=\langle \operatorname {\mathrm {\alpha }},\, \operatorname {\mathrm {\beta }},\, {\gamma }| \operatorname {\mathrm {\alpha }}^8= \operatorname {\mathrm {\beta }}^2={\gamma }^2=1,\, \operatorname {\mathrm {\beta }}^{-1} \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }}= \operatorname {\mathrm {\alpha }}^5,\, {\gamma }^{-1} \operatorname {\mathrm {\alpha }}{\gamma }= \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }},\, \operatorname {\mathrm {\beta }}{\gamma }={\gamma } \operatorname {\mathrm {\beta }} \rangle \).

  2. (2)

    \(G=\langle \operatorname {\mathrm {\alpha }},\, \operatorname {\mathrm {\beta }},\, {\gamma }| \operatorname {\mathrm {\alpha }}^{2^{n-2}}=1,\, \operatorname {\mathrm {\beta }}^2=1,\,{\gamma }^2= \operatorname {\mathrm {\beta }},\, \operatorname {\mathrm {\beta }}^{-1} \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }}= \operatorname {\mathrm {\alpha }}^{1+{2^{n-3}}},\, {\gamma }^{-1} \operatorname {\mathrm {\alpha }}{\gamma }= \operatorname {\mathrm {\alpha }}^{-1} \operatorname {\mathrm {\beta }}, \rangle \).

  3. (3)

    \(G=\langle \operatorname {\mathrm {\alpha }}, \, \operatorname {\mathrm {\beta }} | \operatorname {\mathrm {\alpha }}^{16}= \operatorname {\mathrm {\beta }}^4=1,\, \operatorname {\mathrm {\beta }}^{-1} \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }}= \operatorname {\mathrm {\alpha }}^5\rangle \).

  4. (4)

    \(G=\langle \operatorname {\mathrm {\alpha }}, \, \operatorname {\mathrm {\beta }} | \operatorname {\mathrm {\alpha }}^{2^{n-2}}=1,\, \operatorname {\mathrm {\beta }}^4=1,\, \operatorname {\mathrm {\beta }}^{-1} \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }}= \operatorname {\mathrm {\alpha }}^{-1+2^{n-4}}\rangle \), where n ≥ 6.

  5. (5)

    \(G=\langle \operatorname {\mathrm {\alpha }},\, \operatorname {\mathrm {\beta }},\, {\gamma }| \operatorname {\mathrm {\alpha }}^{2^{n-2}}=1,\, \operatorname {\mathrm {\beta }}^2=1,\,{\gamma }^2=1,\, \operatorname {\mathrm {\beta }}^{-1} \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }}= \operatorname {\mathrm {\alpha }}^{1+{2^{n-3}}},\, {\gamma }^{-1} \operatorname {\mathrm {\alpha }}{\gamma }= \operatorname {\mathrm {\alpha }}^{-1+2^{n-4}} \operatorname {\mathrm {\beta }}, \, \operatorname {\mathrm {\beta }}{\gamma }={\gamma } \operatorname {\mathrm {\beta }} \rangle \), where n ≥ 6.

  6. (6)

    \(G=\langle \operatorname {\mathrm {\alpha }},\, \operatorname {\mathrm {\beta }},\, {\gamma }| \operatorname {\mathrm {\alpha }}^{2^{n-2}}=1,\, \operatorname {\mathrm {\beta }}^2=1,\,{\gamma }^2= \operatorname {\mathrm {\alpha }}^{2^{n-3}},\, \operatorname {\mathrm {\beta }}^{-1} \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }}= \operatorname {\mathrm {\alpha }}^{1+{2^{n-3}}},\, {\gamma }^{-1} \operatorname {\mathrm {\alpha }}{\gamma }= \operatorname {\mathrm {\alpha }}^{-1+2^{n-4}} \operatorname {\mathrm {\beta }}, \, \operatorname {\mathrm {\beta }}{\gamma }={\gamma } \operatorname {\mathrm {\beta }} \rangle \), where n ≥ 6.

  7. (7)

    \(G=\langle \operatorname {\mathrm {\alpha }}, \operatorname {\mathrm {\beta }}, {\gamma }| \operatorname {\mathrm {\alpha }}^8=1,\, \operatorname {\mathrm {\beta }}^2=1,\, {\gamma }^2= \operatorname {\mathrm {\alpha }}^4,\, \operatorname {\mathrm {\beta }}^{-1} \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }}= \operatorname {\mathrm {\alpha }}^5,\, {\gamma }^{-1} \operatorname {\mathrm {\alpha }}{\gamma }= \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }},\, \operatorname {\mathrm {\beta }}{\gamma }={\gamma } \operatorname {\mathrm {\beta }} \rangle \).

A pair (G, N) is called Camina pair if 1 < N < G is normal subgroup of G and for every element g ∈ GN, the element g is conjugate to all gN.

Theorem 6.18 ([1], Theorem 3.7, p. 12)

Let G be a non-abelian finite \( \operatorname {\mathrm {p-group}}\) such that (G, Z(G)) is a Camina pair. Then \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if Z(G)≅Cp and GGZ2(G)∕Z(G).

Theorem 6.19 ([1], Corollary 3.8, p. 12)

Let G be a finite non-abelian Camina \( \operatorname {\mathrm {p-group}}\). Then \({Aut_z}(G)= \operatorname {\mathrm {Z(Inn(G))}}\) if and only if G = Z(G) and Z(G) is cyclic.

6.2.3 Equalities with Class-Preserving Automorphisms

For a finite \( \operatorname {\mathrm {p-group}}\) G, the subgroup Ωm(G) is defined as \(\langle x\in G|\,\, x^{p^m}=1\rangle \), and m(G) is defined as \(\langle x^{p^m}|\,x\in G \rangle \). For a finite \( \operatorname {\mathrm {p-group}}\) G with cl(G) = 2 , GZ(G) is abelian. Consider the following cyclic decomposition of GZ(G) : 

$$\displaystyle \begin{aligned} G/Z(G)\cong C_{p^{e_1}}\times\ldots\times C_{p^{e_k}}\,\,\, (e_1\geq e_2\geq\dots\geq e_k\geq1).\end{aligned}$$

In 2013, Yadav (see [26]) and Kalra and Gumber (see [16]) characterized \( \operatorname {\mathrm {p-groups}}\) of class 2 with Autz(G) = Autc(G) as follows:

Theorem 6.20 ([26], Theorem A, p. 2)

Let G be a finite \( \operatorname {\mathrm {p-group}}\) of class 2. Then Autz(G) = Autc(G) if and only if G = Z(G) and \(|{Aut_{c}(G)}|=\Pi _{i=1}^{d}|\Omega _{m_i}(G^{\prime })|\)

Theorem 6.21 ([26], Theorem B, p. 2)

Let G be a finite \( \operatorname {\mathrm {p-group}}\) and cl(G) = 2 with Autz(G) = Autc(G) and then rank of G is even.

Theorem 6.22 ([16], Theorem 3.1, p. 3)

Let G be a finite \( \operatorname {\mathrm {p-group}}\). Then Autz(G) = Autc(G) if and only if Autc(G)≅Hom(GZ(G), G) and G = Z(G).

Theorem 6.23 ([16], Theorem 3.3, p. 4)

Let G be a finite non-abelian \( \operatorname {\mathrm {p-group}}\) such that the center of the group is elementary abelian. Then Autz(G) = Autc(G) if and only if G is a Camina \( \operatorname {\mathrm {p-group}}\) and cl(G) = 2.

Theorem 6.24 ([16], Theorem 3.4, p. 4)

Let G be a finite non-abelian \( \operatorname {\mathrm {p-group}}\) such that Z(G) is cyclic. Then Autz(G) = Autc(G) if and only if Z(G) = G.

Definition 6.7

A finite \( \operatorname {\mathrm {p-group}}\) G of class 2 is said to have property (∗) if for some π \(\mho _{m_{i}^{\pi }}(\Omega _{n_i}(Z(G))\leq [x,G]\) for all x ∈ GZ(G) and i ∈{1, …, k}.

In 2015, Ghoraishi found a necessary and sufficient condition for a finite \( \operatorname {\mathrm {p-group}}\) G to satisfy Autz(G) = Autc(G), as follows:

Theorem 6.25

Let G be a finite \( \operatorname {\mathrm {p-group}}\). Then Autz(G) = Autc(G) if and only if Z(G) = G and G has property (∗).

6.2.4 Equalities with Absolute Central and IA Automorphisms

Definition 6.8

A finite non-Abelian group G is said to be purely non-Abelian if it has no nontrivial Abelian direct factor.

Let \(C_{{Aut}(G)}({Aut_{l}}(G))= \{ \operatorname {\mathrm {\alpha }}\in {Aut}(G)\,| \operatorname {\mathrm {\alpha }} \operatorname {\mathrm {\beta }}= \operatorname {\mathrm {\beta }} \operatorname {\mathrm {\alpha }},\,\forall \, \operatorname {\mathrm {\beta }}\in {Aut_{l}}(G)\}\) denote the centralizer of Autl(G) in Aut(G). In [20], Moghaddam and Safa defined \(E(G)=[G,C_{{Aut}(G)}({Aut_{l}}(G))]=\langle g^{-1} \operatorname {\mathrm {\alpha }}(g)\,|g\in G,\, \operatorname {\mathrm {\alpha }}\in C_{Aut(G)}({Aut_{l}}(G))\rangle .\) One can easily see that E(G) is a characteristic subgroup of G containing the derived group G = [G, Inn(G)], and each absolute central \( \operatorname {\mathrm {automorphism}}\) of G fixes E(G) elementwise [20, Theorem C].

Let

$$\displaystyle \begin{aligned} {G/E(G) \cong C_{p^{e_{1}}}\times C_{p^{e_{2}}}\times \dots \times C_{p^{e_{k}}}}, \,\,\,(e_{1}\geq \dots e_{k}\geq 1)\end{aligned}$$
$$\displaystyle \begin{aligned} {G/G^{\prime} \cong C_{p^{f_{1}}}\times C_{p^{f_{2}}}\times \dots \times C_{p^{f_{l}}}},\,\,\,(f_{1}\geq \dots f_{l}\geq 1)\end{aligned}$$
$$\displaystyle \begin{aligned} L{(G )\cong C_{p^{g_{1}}}\times C_{p^{g_{2}}}\times \dots \times C_{p^{g_{m}}}},\,\,\,(g_{1}\geq \dots g_{m}\geq 1)\end{aligned}$$
$$\displaystyle \begin{aligned} Z(G)\cong C_{p^{h_{1}}}\times C_{p^{h_{2}}}\times \dots\times C_{p^{h_{n}}}\,\,\,(h_{1}\geq \dots h_{n}\geq 1).\end{aligned}$$

Since GE(G) is a quotient group of GG, it follows that k ≤ l and ei ≤ fi for all 1 ≤ i ≤ k.

In the same year, M. Singh and D. Gumber [24] obtained the equalities of Autz(G) with Autl(G), the group of absolute central automorphisms, and \({Aut_{l}^z}(G)\), the group of absolute central automorphisms that fix the center elementwise, as follows:

Theorem 6.26 ([24], Theorem 1, p. 864)

Let G be a finite non-Abelian \( \operatorname {\mathrm {p-group}}\). Then \({Aut_z}(G)={Aut_{l}^z}(G)\) if and only if either L(G) = Z(G) or Z(G) ≤ Φ(G), G = E(G), m = n, and e1 ≤ gt, where t is the largest integer between 1 and m such that gt < ht.

Theorem 6.27 ([24], Theorem 2, p. 865)

Let G be a finite non-abelian \( \operatorname {\mathrm {p-group}}\) such that L(G) < Z(G). Then \({Aut_z}(G)={Aut_{l}^z}(G)\) if and only if Z(G) ≤ Φ(G), G = E(G)Z(G), m = n, e1 ≤ gt, where t is the largest integer between 1 and m such that gt < ht.

In 2014, Rai [21] characterized finite \( \operatorname {\mathrm {p-groups}}\) for which Autz(G) = IAz(G), where IAz(G) denote the group of those IA automorphisms which fix the center elementwise, as follows:

Theorem 6.28 ([21], Theorem B(1), p. 170 )

Let G be a finite p-group. Then Autz(G) = IAz(G) if and only if G = Z(G).

Let X and Y  be the two finite abelian \( \operatorname {\mathrm {p-groups}}\), and let \(X\cong C_{p^{a_{1}}}\times C_{p^{a_{2}}}\times \dots \times C_{p^{a_{i}}}\) and \(Y\cong C_{p^{b_{1}}}\times C_{p^{b_{2}}}\times \dots \times C_{p^{b_{j}}}\) be the cyclic decomposition of X and Y , where at ≥ at+1 and bs ≥ bs+1 are positive integers. If either X is proper subgroup or proper quotient group of Y  and d(X) = d(Y ), then there certainly exists r, 1 ≤ r ≤ i such that ar < br, ak = bk for r + 1 < k < i. For this unique fixed r, let var(X, Y ) = pr. In other words, var(X, Y ) denotes the order of the last cyclic factor of X whose order is less than that of corresponding cyclic factor of Y.

In 2016, Kalra and Gumber obtained Autz(G) = IAz(G) for finite non-abelian \( \operatorname {\mathrm {p-groups}}\) as follows:

Theorem 6.29 ([17], Theorem 2.12, p. 5)

Let G be a finite non-abelian \( \operatorname {\mathrm {p-group}}\). Then Autz(G) = IAz(G) if and only if either G = Z(G) or G < Z(G), d(G) = d(Z(G)) and exp(GG) ≤ var(G, Z(G)).

6.2.5 Equalities with Central Automorphisms Fixing the Center Elementwise

In 2007, Attar [2] characterized groups in which the central automorphisms fixing the center elementwise are precisely inner automorphisms, as follows:

Theorem 6.30 ([2], Theorem, p. 297)

If G is a \( \operatorname {\mathrm {p-group}}\) of finite order, then \({Aut_z^{z}}(G)={Inn}(G)\) if and only if G is abelian or nilpotency class of G is 2 and Z(G) is cyclic.

Let G be a finite p-group of class 2. Then GZ(G) and G have equal exponent pC(say). Let

$$\displaystyle \begin{aligned} G/Z(G)\cong C_{p^{c_{1}}}\times C_{p^{c_{2}}}\times \dots\times C_{p^{c_{m}}}\,\,\,(c_1\geq \dots \geq c_m\geq 1)\end{aligned}$$

where \( C_{p^{c_{i}}}\) is a cyclic group of order \(p^{c_{i}}\), 1 ≤ i ≤ r. Let k be the largest integer between 1 and r such that c1 = c2 = ck = e. Note that k ≥ 2. “Let M be the subgroup of G containing Z(G) such that

$$\displaystyle \begin{aligned} \bar{M}=M/Z(G)= C_{p^{c_{1}}}\times C_{p^{c_{2}}}\times \dots\times C_{p^{c_{k}}}.\text{''}\end{aligned}$$

Let

$$\displaystyle \begin{aligned} G/G^{\prime}\cong C_{p^{d_{1}}}\times C_{p^{d_{2}}}\times \dots\times C_{p^{d_{n}}}\,\,\,d_{1}\geq d_{2}\geq \dots d_{s}\geq 1\end{aligned}$$

be a cyclic decomposition of GG such that \(\bar {M}\) is isomorphic to a subgroup of

$$\displaystyle \begin{aligned} \bar{N}=N/G^{\prime}:= C_{p^{d_{1}}}\times C_{p^{d_{2}}}\times \dots\times C_{p^{d_{k}}}.\end{aligned}$$

In 2009, using the above terminology, Yadav proved the following:

Theorem 6.31 ([25], Theorem, p. 4326)

Let G be a finite p-group of class 2. Then \({Aut_z}(G)={Aut_z^{z}}(G)\) if and only if m = n, \(G/Z(G)/\bar {M}\cong (G/G^{\prime })/\bar {N}\), and exp(Z(G)) = exp(G).

In 2011, Azhdari and Akhavan-Malayeri [5] generalized the result of Attar in [2] for the finitely generated groups of nilpotency class 2. They got the following:

Theorem 6.32 ([5], Theorm 0.1, p. 1284)

Let G be a finitely generated of cl(G) = 2. Then \({Aut_z^{z}}(G)={Inn}(G)\) if and only if Z(G)≅Cp or \(Z(G) \cong C_{n}\times \mathbb {Z}^{s} \) where exp(GZ(G))∕n and s is torsion-free rank of Z(G).

Theorem 6.33 ([5], Corollary 0.2)

Let G be a finitely generated group of class 2, which is not torsion-free. Then \({Aut_z^{z}}(G)={Inn}(G)\) if and only if cl(G) = 2 and Z(G) is cyclic or \(Z(G) \cong C_{n}\times \mathbb {Z}^{s} \) with exp(GZ(G)) divides n and s is torsion-free rank of Z(G).

Theorem 6.34 ([5], Corollary 0.3)

Let G be a finitely generated of cl(G) = 2. G is torsion-free, and \({Aut_z^{z}}(G)={Inn}(G)\) if and only if Z(G) is infinite cyclic.

In the same year, Jafari also found a necessary and sufficient condition on a finite \( \operatorname {\mathrm {p-group}}\) G such that \({Aut_z}(G)={Aut_z^{z}}(G)\), as follows:

Theorem 6.35

Let G be a finite \( \operatorname {\mathrm {p-group}}\) . Then \({Aut_z}(G)={Aut_z^{z}}(G)\) if and only if \(Z(G)G{'}\subseteq G^{p^n}G{'}\) , where \(\exp (Z(G))=p^n.\)

Let G be a non-abelian finite p-group. Let

$$\displaystyle \begin{aligned} G/G^{\prime}= C_{p^{c_{1}}}\times C_{p^{c_{2}}}\times \dots\times C_{p^{c_{r}}}\,\,\, (c_{1}\geq \dots c_{r}\geq 1).\end{aligned}$$
$$\displaystyle \begin{aligned} G/G^{\prime} Z(G)\cong C_{p^{d_{1}}}\times C_{p^{d_{2}}}\times \dots\times C_{p^{d_{s}}}\,\,\,(d_{1}\geq \dots d_{s}\geq 1).\end{aligned}$$

and \(Z(G)\cong C_{p^{e_{1}}}\times C_{p^{e_{2}}}\times \dots \times C_{p^{e_{t}}}\,\,\,(e_{1}\geq \dots e_{t}\geq 1).\)

since GGZ(G) is a quotient of GG.

In 2012, Attar [3] gave a necessary and sufficient condition on finite \( \operatorname {\mathrm {p-group}}\) G such that Autz(G) to be \({Aut_z^{z}}(G)\), as follows:

Theorem 6.36 ([3], Theorem A, p. 1097)

Let G be a non-abelian finite \( \operatorname {\mathrm {p-group}}\). Then \({Aut_z}(G)={Aut_z^{z}}(G)\) if and only if Z(G) ≤ G or Z(G) ≤ Φ(G), r = s, and c1 ≤ bm where m is the largest integer between 1 and r such that am > bm.

Theorem 6.37 ([3], Corollary 2.1, p. 1098)

Let G be a non-abelian finite \( \operatorname {\mathrm {p-group}}\) such that exponent of Z(G) is p. Then \({Aut_z}(G)={Aut_z^{z}}(G)\) if and only if Z(G) ≤ Φ(G).