1 Introduction

Every group considered in this paper is finite. Most of the notation is standard and can be found in [3, 10]. Let \(|\,G|\) denote the order of a group G, \(|\,G|_{p}\) denote the p-part of \(|\,G|\) and \(\pi (G)\) denote the set of all prime divisors of \(|\,G|\). Let \(A\rtimes B\) denote the semidirect product of groups A and B, where B is an operator group of A. Let \(\mathcal F\) be a class of groups and I / K be a chief factor of a group G. I / K is called Frattini provided that \(I/K \le \Phi (G/K)\). Moreover, I / K is called \(\mathcal F\)-central if \(I/K \rtimes (G/C_G(I/K))\in \mathcal F\). Otherwise, I / K is called \(\mathcal F\)-eccentric. The symbol \(Z_\mathcal{F}(G)\)(\(Z_{p\mathcal F}(G))\) denotes the \(\mathcal F\)-hypercentre(\(p\mathcal F\)-hypercentre) of a group G, which is the product of all such normal subgroups H of G whose G-chief factors(whose G-chief factors of order divisible by p)are \(\mathcal F\)-central. In addition, \(\mathcal {U}\) and \(p\mathcal {U}\) denote the class of all supersolvable groups and all p-supersolvable groups, respectively.

Suppose that A is a subgroup of G, \(K\le H\le G\). (1) if \(AH=AK\), then A covers the pair (KH); (2) if \(A\cap H=A\cap K\), then A avoids (KH). In 1939, Ore [11] introduced the notion of quasinormal subgroups. Furthermore, if E is a quasinormal subgroup of G, then for every maximal pair of G, that is, a pair (KH), where K is a maximal subgroup of H, E either covers or avoids (KH). In 1992, Doerk and Hawkes [3] gave the definition of CAP-subgroups, that is, a subgroup A of G is called a CAP-subgroup if A either covers or avoids each pair (KH), where H / K is a chief factor of G. Based on the definitions and observations above, Guo and Skiba introduced new concepts as follows:

Definition 1.1

[6, Definition 1.1] Let \(\Sigma =\{G_0\le G_1\le \ldots \le G_n\}\) be some subgroup series of G and A be a subgroup of G. Then A is \(\Sigma \)-embedded in G if A either covers or avoids every maximal pair (KH) such that \(G_{i-1}\le K<H\le G_i\), for some i.

In [6], quasinormal subgroups, CAP-subgroups, and partial CAP-subgroups [1] (or a semi cover-avoiding subgroups [8]) are \(\Sigma \)-embedded subgroups.

Definition 1.2

[6, Definition 2.7] Let A be a subgroup of G. Then A is m-embedded in G if G has a subnormal subgroup T and a \(\{1\le G\}\)-embedded subgroup C in G such that \(G=AT\) and \(A\cap T\le C\le A\).

In [6, Example 2.8], every c-normal subgroup of G is also m-embedded in G.

On the other hand, In 2014, Xu and Zhang [14] investigated p-nilpotency of a group by using m-embedded property of primary subgroups. In 2015, Tang and Miao [13] obtained some results about p-supersolvability of finite groups by using m-embedded primary subgroups. They proved the following theorem:

Theorem 1.3

[13, Theorem 1.3] Let G be a p-solvable group and P be a Sylow p-subgroup of G where p is an odd prime divisor of |G|. If every maximal subgroup of P is m-embedded in G,  then G is p-supersolvable.

It is clear that the p-solvability in [13, Theorem 1.3] is essential. Naturally, the question is that

What is the structure of a group if the p-solvability is removed in [13, Theorem 1.3]?

Along the clue, we obtained the following result:

Theorem 1.4

Let E be a normal subgroup of G and P be a Sylow p-subgroup of E where p is an odd prime divisor of |E|. If every maximal subgroup of P is m-embedded in G,  then every G-chief factor A / B below E satisfies one of the following conditions : 

$$\begin{aligned} (1) \ A/B\le \Phi (G/B); \quad \quad \quad (2) \ A/B\text { is a } p{'}-\text { group}; \quad \quad \quad (3) \ |A/B|_{p}=p. \end{aligned}$$

With the further consideration, we will study the p-supersolvability and generalized hypercentre of a finite group by using m-embedded subgroups, where p is a prime. Some theorems in [9, 12,13,14] are corollaries of our results.

2 Preliminaries

For the sake of convenience, we first list here some known results which will be useful in the sequel.

Lemma 2.1

[6, Lemma 2.13] Let K and H be subgroups of G. Suppose that K is m-embedded in G and H is normal in G. Then

  1. (1)

    If \(H\le K\) \(,\) then K / H is m-embedded in G / H.

  2. (2)

    If \(K\le E\le G\) \(,\) then K is m-embedded in E.

  3. (3)

    If \((|H|, |K|)=1\) \(,\) then KH / H is m-embedded in G / H.

Lemma 2.2

[6, Lemma 2.14] Let P be a normal non-identity p-subgroup of G with \(|P|=p^{n}\) and \(P\cap \Phi (G)=1\). Suppose that there is an integer k such that \(1\le k<n\) and the subgroups of P of order \(p^{k}\) are m-embedded in \(G\) \(,\) then some maximal subgroup of P is normal in G.

Lemma 2.3

[6, Lemma 2.5] Every \(\{1\le G\}\)-embedded subgroup of G is subnormal in G.

Lemma 2.4

[5, Lemma 2.8] Let G be a p-supersolvable group. If \(O_{p{'}}(G)=1,\) then G is supersolvable.

3 Main Results

Theorem 3.1

Let p be an odd prime divisor of |G| and P be a normal p-subgroup of G. If every minimal subgroup of P is \(\{1\le G\}\)-embedded in G,  then \(P\le Z_\mathcal{U}(G)\).

Proof

Assume that the assertion is false and choose (GP) to be a counterexample of minimal order.

  1. (1)

    G has a unique normal subgroup N such that P / N is a chief factor of G, \(N\le Z_{\mathcal {U}}(G)\) and \(|P/N|>p\).

    Let P / N be a chief factor of G. Then, clearly, (GN) satisfies the hypothesis of the theorem. The choice of (GP) implies that \(N\le Z_{\mathcal {U}}(G)\). If \(|P/N|=p\), then \(P/N\le Z_{\mathcal {U}}(G/N)\) and so \(P\le Z_{\mathcal {U}}(G)\), a contradiction. Hence \(|P/N|>p\). Assume that P / L is a chief factor of G with \(P/N\ne P/L\). With the same discussion as above, we have that \(L\le Z_{\mathcal {U}}(G)\). Then \(P/N=NL/N\le NZ_{\mathcal {U}}(G)/N\le Z_{\mathcal {U}}(G/N)\). It follows from \(N\le Z_{\mathcal {U}}(G)\) that \(P\le Z_{\mathcal {U}}(G)\), a contradiction.

  2. (2)

    The exponent of P is p.

    Let C be a Thompson critical subgroup of P. If \(\Omega (C)<P\), then \(\Omega (C)\le N\le Z_{\mathcal {U}}(G)\) by (1), so \(P\le Z_{\mathcal {U}}(G)\) by [7, Lemma 4.4], which is impossible. Hence \(P=C=\Omega (C)\). Then by [7, Lemma 4.3], the exponent of P is p.

  3. (3)

    P is a minimal normal subgroup of G.

    If not, then \(N\ne 1\). Let H / N be a minimal subgroup of P / N. Then there exists an element \(x\in H\backslash N\), \(H=\langle x\rangle N\) and \(|\langle x\rangle |=p\) by (2). By hypothesis and [6, Lemma 2.3], H / N is \(\{1\le G/N\}\)-embedded in G / N. Then \(P/N\le Z_{\mathcal {U}}(G/N)\) by the choice of (GP). Hence \(|P/N|=p\), a contradiction. Hence \(N=1\) and (3) holds.

  4. (4)

    \(P\le \Phi (G)\).

    If not, then \(P\nleq \Phi (G)\). By (1), we may choose a minimal subgroup H of P such that \(G=HM=PM\) and \(P\cap M=1\). Since \(|P:P\cap M|=p\), \(|P|=p\) and \(P\le Z_\mathcal{U}(G)\), a contradiction.

  5. (5)

    The final contradiction.

    By [6, Lemma 2.3], every minimal subgroup of P is \(\{1\le M\}\)-embedded in M. Then \(P\le Z_\mathcal{U}(M)\) by the choice of (GP), for every maximal subgroup M of G. We assert that \(C_{G}(P)\le \Phi (G)\). If not, then \(C_{G}(P)\nleq \Phi (G)\) and \(G=C_{G}(P)M_{1}\) for some maximal subgroup \(M_{1}\) of G. Next, we choose a minimal normal subgroup N of \(M_{1}\) contained in P. Then \(|N|=p\). Further, \(N^{G}=N^{C_{G}(P)M_{1}}=N^{M_{1}}=N\) and \(N\unlhd G\). Then \(|N|=|P|=p\), a contradiction. Set \(Z=\cap (C_{M}(X/Y))\), where X / Y is an M-chief factor below P for every maximal subgroup M of G. Then M / Z is an abelian group of exponent dividing \(p-1\) and \(O^{p}(Z)\le C_{G}(P)\le \Phi (G)\). Hence \(M/\Phi (G)\) is a strictly p-closed group and \(M/\Phi (G)\) is supersolvable by [15, Theorem 1.9]. Then \(G/\Phi (G)\) is minimal non-supersolvable and G is solvable by [15, Theorem 2.3]. Further, we have \(F(G)\le C_{G}(P)\le \Phi (G)<F(G)\), a contradiction.

The final contradiction completes our proof.\(\square \)

Theorem 3.2

Let p be an odd prime divisor of |G| and P be a Sylow p-subgroup of G. If every minimal subgroup of P is m-embedded in G,  then G is p-supersolvable.

Proof

Assume that the assertion is false and choose G to be a counterexample of minimal order. Furthermore, we have that

  1. (1)

    \(O_{p{'}}(G)=1\).

    Assume that \(T=O_{p{'}}(G)\ne 1\). By Lemma 2.1(3), G / T satisfies the conditions of the theorem, and the minimal choice of G implies that G / T is p-supersolvable. Hence G is p-supersolvable, a contradiction.

  2. (2)

    \(P\cap E\unlhd E\), where E is a proper normal subgroup of G.

    Assume that E is a proper normal subgroup of G. By (1), \(P\cap E\ne 1\). By Lemma 2.1 and the choice of G, E is p-supersolvable. Hence E is supersolvable by (1) and Lemma 2.4. Then \(P\cap E\unlhd E\).

  3. (3)

    There exists a minimal subgroup H of P such that H has a normal complement in G. Otherwise, all minimal subgroups of P are \(\{1\le G\}\)-embedded in G. Then all minimal subgroups of P are contained in \(O_{p}(G)\) by Lemma 2.3. Further, \(O_{p}(G)\le Z_\mathcal{U}(G)\). Hence G is p-supersolvable by [2, Theorem 6], a contradiction.

  4. (4)

    The final contradiction.

    By (3) and hypothesis, \(G=HM\), \(M\unlhd G\), \(H\cap M=1\). Then \(P\cap M\unlhd M\) by (2). If every minimal subgroup of \(P\cap M\) is \(\{1\le G\}\)-embedded in G, then \(P\cap M\le Z_\mathcal{U}(G)\) by Theorem 3.1 and \(1\unlhd P\cap M\unlhd M\unlhd G\) a normal subgroup series of G such that every G-chief factor either cyclic of order p or \(p{'}\)-group. Hence G is p-supersolvable, a contradiction.

Now we assume that there exists a minimal subgroup \(H_{1}\) of \(P\cap M\) such that \(H_{1}\) has a normal complement in G. Further, \(G=H_{1}M_{1}\), \(M_{1}\unlhd G\), \(H_{1}\cap M_{1}=1\). Similar to the previous discussion, we consider \(P\cap M \cap M_{1}\). Now, we set \(G_{0}=G\), \(G_{1}=M\), \(G_{2}=M\cap M_{1}\). Repeat above discussion, we have \(1=G_{s+1}\unlhd G_{s}\unlhd \cdots \unlhd G_{3}\unlhd G_{2}\unlhd G_{1}\unlhd G_{0}=G\) a normal subgroup series of G such that every G-chief factor either cyclic of order p or \(p{'}\)-group. Then G is p-supersolvable, a contradiction.

The final contradiction completes our proof.\(\square \)

Theorem 3.3

Let E be a normal subgroup of G and P be a Sylow p-subgroup of E where p is an odd prime divisor of |E|. If every minimal subgroup of P is m-embedded in G,  then \(E\le Z_{p\mathcal U}(G)\).

Proof

Assume that the assertion is false and choose (GE) to be a counterexample of minimal order. Furthermore, we have that

  1. (1)

    \(O_{p{'}}(E)=1\).

    If \(K=O_{p{'}}(E)\ne 1\), then we consider G / K. (G / KE / K) satisfies the hypothesis of the theorem by Lemma 2.1(3). The minimal choice of (GE) implies that \(E/K\le Z_{p\mathcal U}(G/K)\), and so \(E\le Z_{p\mathcal U}(G)\), a contradiction.

  2. (2)

    \(P\unlhd G\).

    By Theorem 3.2, E is p-supersolvable. By (1) and Lemma 2.4, E is supersolvable. Then \(P\unlhd G\).

  3. (3)

    There exists a minimal subgroup H of P such that H has a normal complement in G.

    Otherwise, all minimal subgroups of P are \(\{1\le G\}\)-embedded in G. By Theorem 3.1, \(P\le Z_\mathcal{U}(G)\). Hence \(E\le Z_{p\mathcal U}(G)\), a contradiction.

  4. (4)

    The final contradiction.

    By (3) and hypothesis, \(G=HM\), \(M\unlhd \unlhd G\), \(H\cap M=1\). Then \(M\unlhd G\), \(|P:P\cap M|=p\) and \(P\cap M\unlhd G\). By the choose of (GE), we have that \(P\cap M\le Z_{p\mathcal U}(G)\). Hence \(P\le Z_{p\mathcal U}(G)\) and \(E \le Z_{p\mathcal U}(G)\), a contradiction.

The final contradiction completes our proof.\(\square \)

Corollary 3.4

Let E be a normal subgroup of G such that G / E is p-supersolvable and P be a Sylow p-subgroup of E where p is an odd prime divisor of |E|. If every minimal subgroup of P is m-embedded in G,  then G is p-supersolvable.

By Theorem 3.3, it is easy to prove the following corollaries:

Corollary 3.5

[9, Theorem 3.8] Let p be an odd prime number dividing the order of a group G and \(\mathcal F\) a saturated formation containing the class \(p\mathcal U\) of all p-supersolvable groups. Also let N be a normal subgroup of G such that \(G/N \in \mathcal F\). If P is a Sylow p-subgroup of N and every minimal subgroup of P is c-normal in G,  then \(G \in \mathcal F\).

Theorem 3.6

Let G be a group and P be a Sylow p-subgroup of G where p is an odd prime divisor of |G|. If every maximal subgroup of P is m-embedded in G,  then every chief factor A / B of G satisfies one of the following conditions : 

$$\begin{aligned} (1) \ A/B\le \Phi (G/B); \quad \quad \quad (2) \ A/B \ \text { is a } p{'}\text {-group} ; \quad \quad \quad (3) \ |A/B|_{p}=p. \end{aligned}$$

Proof

Assume that the theorem is false and let G be a counterexample of minimal order.

  1. (1)

    \(O_{p{'}}(G)=1\).

    Assume that \(O_{p{'}}(G)\ne 1\). The hypothesis also holds for \(G/O_{p{'}}(G)\) by Lemma 2.1, and for G, a contradiction.

  2. (2)

    If \(O_{p}(G)\ne 1\), then \(O_{p}(G)\cap \Phi (G)=1\).

    Assume that \(O_{p}(G)\cap \Phi (G)\ne 1\). We may choose a minimal normal subgroup L of G such that \(L\le O_{p}(G)\cap \Phi (G)\). By induction, G / L holds and so G holds, a contradiction.

  3. (3)

    \(O_{p}(G)=1\).

Assume that \(O_{p}(G)\ne 1\). By (2) and [4, Theorem 1.8.17], \(O_{p}(G)=L_1\times L_2\times \cdots \times L_t\) where \(L_i\) are the minimal normal subgroups of G, \(i=1,2,\ldots ,t\). For every \(L\in \{L_i\}\) and we consider G / L. Clearly, G / L holds by the choice of G. Further, we assert that \(O_{p}(G)\) is a minimal normal subgroup of G. Otherwise, there exists two different minimal normal subgroups \(L_1\) and \(L_2\) such that \(G/L_{j}\) satisfies the hypothesis of theorem and so every \(G/L_{j}\)-chief factor holds by the choice of G where \(j=1, 2\). If \(L_{1}L_{2}/L_{2}\le \Phi (G/L_{2})\), then \(L_{1}L_{2}\le \Phi (G)L_{2}\) by [3, A. Lemma 9.11]. Since \(L_{1}L_{2}\le O_{p}(G)\), \(L_{1}L_{2}\le O_{p}(G)\cap \Phi (G)L_{2}=(O_{p}(G)\cap \Phi (G))L_{2}=L_{2}\) by (2), a contradiction. Hence \(L_{1}\cong L_{1}L_{2}/L_{2}\) satisfies the condition (2) or (3), then every G-chief factor holds, a contradiction.

Hence \(O_{p}(G)\) is a minimal normal subgroup of G and \(\Phi (G)=1\). Then there exists a maximal subgroup M of G such that \(G=O_{p}(G)M\). We assert that \(O_{p}(G)<P\). If not, by Lemma 2.2, \(|\,P|=p\) and so G holds, a contradiction. Hence we may choose a maximal subgroup \(P_{1}\) of P such that \(M_{p}\le P_{1}\) and \(O_{p}(G)\nleq P_{1}\). By hypothesis, \(P_{1}\) is m-embedded in G, there exists a subnormal subgroup T in G and a \(\{1\le G\}\)-embedded subgroup C in G such that \(G=P_{1}T\) and \(P_{1}\cap T\le C \le P_{1}\). We assert that \(C=1\). Otherwise, \(C\ne 1\). If \(C<O_{p}(G)\) by Lemma 2.3, then we obtain C neither covers nor avoids maximal pair (MG) since \(O_{p}(G)\cap M=1\), a contradiction. Hence we may assume that \(C=O_{p}(G)\) by Lemma 2.3, that is, \(O_{p}(G)\le P_1\), a contradiction. Then we have \(|\,T_{p}|=p\).

If \(O_{p}(G)\cap T\ne 1\), then \(O_{p}(G)\cap T=T_{p}\unlhd T\) and so T is p-solvable. Furthermore, \(T_{p{'}}\) is a Hall \({p{'}}\)-subgroup of G. By [4, Theorem 1.8.19] and (1), \(C_{T}(T_{p})=T_{p}\). Hence \(N_{T}(T_{p})/C_{T}(T_{p})=T/C_{T}(T_{p})\hookrightarrow C_{p-1}\). By Schur–Zassenhaus Theorem, \(T_{p{'}}\) is cyclic and T is supersolvable. We assert that p is the largest prime divisor of |G|. Otherwise, if \(q\ne p\) is the largest prime divisor of |G|, then \(Q\unlhd G\), which contradicts (1). Then we assume that \(p_{1}<p_{2}<\cdots <p_{n}=p\), where \(\pi (G)=\{p_{1}, p_{2}, \ldots , p_{n}=p\}\). Since \(T_{p{'}}\) is cyclic, \(G_{p{'}}\) is cyclic and \(G_{p_{1}}\) is cyclic where \(G_{p_{1}}\) a Sylow \(p_{1}\)-subgroup of G. By Burnside Theorem, G is \(p_{1}\)-nilpotent and \(G_{p_{1}{'}}\unlhd G\). Next, we consider \(G_{p_{1}{'}}\). Similar to the previous discussion, \(G_{\{p_{1}, p_{2}\}{'}}\unlhd G_{p_{1}{'}}\). Repeat above discussion, we get a normal subgroup series of G: \(1\unlhd P \unlhd \ldots \unlhd G_{\{p_{1}, p_{2}\}{'}}\unlhd G_{p_{1}{'}}\unlhd G\). Hence G has supersolvable type Sylow tower and \(P\unlhd G\), a contradiction.

If \(O_{p}(G)\cap T=1\), then \(O_{p}(G)\le N_{G}(T)\) by [3, Lemma A.14.3] and \(O_{p}(G)T=O_{p}(G)\times T\). Since \(O_{p}(G)\cap Z(P)\ne 1\), we may pick a minimal subgroup H of \(O_{p}(G)\cap Z(P)\) and \(H\unlhd G\). Clearly, G / H satisfies the hypothesis of Theorem and so G / H holds by the choice of G. Then every chief factor of G satisfies one of the three conditions in the conclusion of Theorem, a contradiction.

  1. (4)

    The final contradiction.

Let \(P_{2}\) be a maximal subgroup of P. By hypothesis and (3), \(P_{2}\) is m-embedded in G, we may choose a subnormal subgroup \(K_{2}\) of G such that \(G=P_{2}K_{2}\) and \(P_{2}\cap K_{2}=1\). Hence there exists a maximal normal subgroup K such that \(|G:K|=p\). Clearly, \(K_{p}\) is m-embedded in G, where \(K_{p}\) is a Sylow p-subgroup of K, we may choose a subnormal subgroup \(K_{3}\) of G such that \(G=K_{p}K_{3}=KK_{3}\) and \(K_{p}\cap K_{3}=1\). Since \(|\,K_{p}K_{3}|=|\,KK_{3}|\), we have \(|\,K\cap K_{3}|=\frac{|\,K|}{|\,K_{p}|}\). If \(K\cap K_{3}\ne 1\), then \(K\cap K_{3}\) is a \(p{'}\)-group, which contradicts (1). Hence \(K\cap K_{3}=1\) and K is a normal p-subgroup, which contradicts (3).

The final contradiction completes our proof.\(\square \)

From Theorem 3.6, it is easy to prove the following corollaries:

Corollary 3.7

[14, Theorem 3.1] Let G be a group and P be a Sylow p-subgroup of G where p is an odd prime divisor of |G|. If every maximal subgroup \(P_{1}\) of P is m-embedded in G and \(N_{G}(P_{1})\) is p-nilpotent,  then G is p-nilpotent.

Proof

Clearly, G is not a non-abelian simple group and \(O_{p{'}}(G)=1\). Then we may pick a minimal normal subgroup L of G. Further, L satisfies one of the three conditions in Theorem 3.6. We only need to consider the condition that \(|L|_{p}=p\). Then we consider the group \(N_{G}(L_{p})\) where \(L_{p}\) is a Sylow p-subgroup of L. Next, we prove that \(|L|=p\). If \(N_{G}(L_{p})<G\), then \(N_{G}(L_{p})\) is p-nilpotent since \(P\le N_{G}(L_{p})\). Further, \(N_{L}(L_{p})=C_{L}(L_{p})\) and so L is p-nilpotent by Burnside Theorem. Then \(|L|=p\). If \(N_{G}(L_{p})=G\), then \(|L|=p\).

Since G / L is p-nilpotent, G is p-supersolvable and so G is supersolvable by Lemma 2.4. Hence \(P\unlhd G\), p is the largest prime divisor of |G|. Since \(G=L\rtimes M\), \(P\cap M\) is a maximal subgroup of P and \(P\cap M \unlhd G\). Then \(G=N_{G}(P\cap M)\) is p-nilpotent.\(\square \)

Corollary 3.8

[13, Theorem 1.2] Let G be a group and P be a Sylow p-subgroup of G where p is an odd prime divisor of |G|. If every maximal subgroup of P is m-embedded in G and \(N_{G}(P)\) is p-nilpotent,  then G is p-nilpotent.

Proof

See the proof of Corollary 3.7.\(\square \)

Corollary 3.9

[9, Theorem 3.1] Let p be an odd prime dividing the order of a group G and P a Sylow p-subgroup of G. If \(N_{G}(P)\) is p-nilpotent and every maximal subgroup of P is c-normal in G,  then G is p-nilpotent.

Corollary 3.10

[13, Theorem 1.3] Let G be a p-solvable group and P be a Sylow p-subgroup of G where p is an odd prime divisor of |G|. If every maximal subgroup of P is m-embedded in G,  then G is p-supersolvable.

Proof

Clearly, G is not a non-abelian simple group. Then we may pick a minimal normal subgroup L of G. Further, L satisfies one of the three conditions in Theorem 3.6. Since G / L satisfies the hypothesis of Theorem 3.6, G / L is p-supersolvable by induction. Then G is p-supersolvable.\(\square \)

Theorem 3.11

Let E be a normal subgroup of G and P be a Sylow p-subgroup of E where p is an odd prime divisor of |E|. If every maximal subgroup of P is m-embedded in G,  then every G-chief factor A / B below E satisfies one of the following conditions : 

$$\begin{aligned} (1) \ A/B\le \Phi (G/B); \quad \quad \quad (2) \ A/B \text { is a } p{'}\text {-}\,\text {group}; \quad \quad \quad (3) \ |A/B|_{p}=p. \end{aligned}$$

Proof

Assume that the theorem is false and let (GE) be a counterexample with |G||E| minimal.

  1. (1)

    \(O_{p{'}}(E)=1\).

    Assume that \(O_{p{'}}(E)\ne 1\). The hypothesis also holds for \((G/O_{p{'}}(E), E/O_{p{'}}(E))\) by Lemma 2.1, and for (GE). Then every G-chief factor below E holds, a contradiction.

  2. (2)

    If \(O_{p}(E)\ne 1\), then \(O_{p}(E)\cap \Phi (G)=1\).

    Assume that \(O_{p}(E)\cap \Phi (G)\ne 1\). We may choose a minimal normal subgroup L of G such that \(L\le O_{p}(E)\cap \Phi (G)\). By induction, (G / LE / L) satisfies the hypothesis of theorem and so every G / L-chief factor below E / L holds. Then every G-chief factor below E holds, a contradiction.

  3. (3)

    \(O_{p}(E)=1\).

Assume that \(O_{p}(E)\ne 1\). By (2) and [4, Theorem 1.8.17], \(O_{p}(E)=L_1\times L_2\times \cdots \times L_t\) where \(L_i\) are the minimal normal subgroups of G, \(i=1,2,\ldots ,t\). We assert that \(O_{p}(E)\) is a minimal normal subgroup of G. Otherwise, there exists two minimal normal subgroups \(L_1\) and \(L_2\), then we consider \((G/L_1, E/L_1)\) and \((G/L_2, E/L_2)\). Clearly, \((G/L_{j}, E/L_{j})\) satisfies the hypothesis of theorem and so every \(G/L_{j}\)-chief factor below \(E/L_{j}\) holds by the choice of (GE) where \(j=1, 2\). If \(L_{1}L_{2}/L_{2}\le \Phi (G/L_{2})\), then \(L_{1}L_{2}\le \Phi (G)L_{2}\) by [3, A. Lemma 9.11]. Since \(L_{1}L_{2}\le O_{p}(E)\), \(L_{1}L_{2}\le O_{p}(E)\cap \Phi (G)L_{2}=(O_{p}(E)\cap \Phi (G))L_{2}=L_{2}\) by (2), a contradiction. Hence \(L_{1}\cong L_{1}L_{2}/L_{2}\) satisfies the condition (2) or (3), then every G-chief factor holds, a contradiction. Clearly, \(O_{p}(E)\nleq \Phi (G)=1\). Then there exists a maximal subgroup M of G such that \(G=O_{p}(E)M\). Then \(P=O_{p}(E)(P\cap M)\). We assert that \(O_{p}(E)<P\). If not, by Lemma 2.2, \(|\,P|=p\) and so (GE) holds, a contradiction. Hence we may choose a maximal subgroup \(P_{1}\) of P such that \(P\cap M\le P_{1}\) and \(O_{p}(E)\nleq P_{1}\). By hypothesis, \(P_{1}\) is m-embedded in G, there exists a subnormal subgroup T in G and a \(\{1\le G\}\)-embedded subgroup C in G such that \(G=P_{1}T\) and \(P_{1}\cap T\le C\le P_{1}\). We assert that \(C=1\). Otherwise, \(C\ne 1\). If \(C<O_{p}(E)\) by Lemma 2.3, then we obtain C neither covers nor avoids maximal pair (MG) since \(O_{p}(E)\cap M=1\), a contradiction. Hence we may assume that \(C=O_{p}(E)\) by Lemma 2.3, i.e., \(O_{p}(E)\le P_1\) and so \(P\le O_{p}(E)(P\cap M)\le P_1<P\), a contradiction. Then we have \(|\,T_{p}|=p\).

If \(O_{p}(E)\cap T\ne 1\), then \(O_{p}(E)\cap T=T_{p}\unlhd T\) and so T is p-solvable. Furthermore, \(T_{p{'}}\) is a Hall \({p{'}}\)-subgroup of G. By [4, Theorem 1.8.19] and (1), \(C_{T}(T_{p})=T_{p}\). Hence \(N_{T}(T_{p})/C_{T}(T_{p})=T/C_{T}(T_{p})\hookrightarrow C_{p-1}\). By Schur–Zassenhaus Theorem, \(T_{p{'}}\) is cyclic and T is supersolvable. Then \(E\cap T\) is supersolvable. We assert that p is the largest prime divisor of |E|. Otherwise, if \(q\ne p\) is the largest prime divisor of |E|, then \(Q\le O_{q}(E)=1\) where Q is a Sylow q-subgroup of \(E\cap T\), which contradicts (1). Then we assume that \(p_{1}<p_{2}<\cdots <p_{n}=p\), where \(\pi (E)=\{p_{1}, p_{2}, \ldots , p_{n}=p\}\). Since \(T_{p{'}}\) is cyclic, \(E_{p{'}}=E\cap T_{p{'}}\) is cyclic and \(E_{p_{1}}\) is cyclic where \(E_{p_{1}}\) a Sylow \(p_{1}\)-subgroup of E. By Burnside Theorem, E is \(p_{1}\)-nilpotent and \(E_{p_{1}{'}}\unlhd E\). Next, we consider \(E_{p_{1}{'}}\). Similar to the previous discussion, \(E_{\{p_{1}, p_{2}\}{'}}\unlhd E_{p_{1}{'}}\). Repeat above discussion, we get a normal subgroup series of E: \(1\unlhd P \unlhd \cdots \unlhd E_{\{p_{1}, p_{2}\}{'}}\unlhd E_{p_{1}{'}}\unlhd E\). Hence E has supersolvable type Sylow tower and \(P\unlhd E\), a contradiction.

If \(O_{p}(E)\cap T=1\), then \(O_{p}(E)\le N_{G}(T)\) by [3, Lemma A.14.3] and \(O_{p}(E)T=O_{p}(E)\times T\). Since \(O_{p}(E)\cap Z(G_{p})\ne 1\) where \(G_{p}\) is a Sylow p-subgroup of G, we may pick a minimal subgroup H of \(O_{p}(E)\cap Z(G_{p})\) and \(H\unlhd G\). Clearly, (G / HE / H) satisfies the hypothesis of Theorem and so (G / HE / H) holds by the choice of (GE). Then every G-chief factor below E holds, a contradiction.

  1. (4)

    The final contradiction.

Let \(P_{2}\) be a maximal subgroup of P. By hypothesis and (3), \(P_{2}\) is m-embedded in E, we may choose a subnormal subgroup \(K_{2}\) of E such that \(E=P_{2}K_{2}\) and \(P_{2}\cap K_{2}=1\). Hence there exists a maximal normal subgroup K such that \(|E:K|=p\). Clearly, \(K_{p}\) is m-embedded in E, where \(K_{p}\) is a Sylow p-subgroup of K, we may choose a subnormal subgroup \(K_{3}\) of E such that \(E=K_{p}K_{3}=KK_{3}\) and \(K_{p}\cap K_{3}=1\). Since \(|\,K_{p}K_{3}|=|\,KK_{3}|\), we have \(|\,K\cap K_{3}|=\frac{|\,K|}{|\,K_{p}|}\). If \(K\cap K_{3}\ne 1\), then \(K\cap K_{3}\) is a \(p{'}\)-group, which contradicts (1). Hence \(K\cap K_{3}=1\) and K is a normal p-subgroup, which contradicts (3).

The final contradiction completes our proof.\(\square \)

Corollary 3.12

Let E be a normal subgroup of G and P be a Sylow p-subgroup of E where p is an odd prime divisor of |E|. Suppose that \(G/E=\overline{G}\) and every chief factor \(\overline{A}/\overline{B}\) of \(\overline{G}\) satisfies one of the following conditions : 

$$\begin{aligned} (1) \,\overline{A}/\overline{B}\le \Phi (\overline{G}/\overline{B}); \quad \quad \quad (2)\, \overline{A}/\overline{B} \text { is a} p{'}\text {-}\,\text {group}; \quad \quad \quad (3)\, |\overline{A}/\overline{B}|_{p}=p. \end{aligned}$$

If every maximal subgroup of P is m-embedded in G,  then every chief factor A / B of G satisfies one of the following conditions : 

$$\begin{aligned} (1)\, A/B\le \Phi (G/B); \quad \quad \quad (2)\, A/B \text { is a} p{'}\text {-}\,\text {group}; \quad \quad \quad (3)\, |A/B|_{p}=p. \end{aligned}$$

Corollary 3.13

[12, Theorem 3.1] Let p be a prime, G be a p-solvable group and let H be a normal subgroup of G such that \(G/H\in \) \(p\mathcal U\) \(,\) \(p\mathcal U\) is the class of all p-supersolvable groups. If the maximal subgroups of the Sylow p-subgroups of H are c-normal in G,  then \(G \in \) \(p\mathcal U\).