1 Introduction

The graphs discussed here are finite, undirected and simple. Let G be a graph. The vertex set and the edge set of G are denoted by V(G) and E(G), respectively. Denote by \(d_G(v)\) the degree of a vertex v in G. Let i(G) and \(\omega (G)\) denote the number of isolated vertices and connected components in G, respectively. For any \(X\subseteq V(G)\), G[X] is the subgraph of G induced on X, and \(G-X\) denotes the subgraph \(G[V(G)\setminus X]\). For any \(E'\subseteq E(G)\), \(G-E'\) denotes the subgraph derived from G by removing \(E'\). A vertex subset X of G is called an independent set if G[X] has no edges. The path and the complete graph with n vertices are denoted by \(P_n\) and \(K_n\), respectively. We use \(G_1\vee G_2\) to denote the join of two disjoint graphs \(G_1\) and \(G_2\), and \(G_1\cup G_2\) to denote the union of two disjoint graphs \(G_1\) and \(G_2\).

Yang, Ma and Liu [14] introduced the notion of isolated toughness, which is defined by

$$\begin{aligned} I(G)=\min \{\frac{|X|}{i(G-X)}:X\subseteq V(G), i(G-X)\ge 2\} \end{aligned}$$

if G is not a complete graph; otherwise, \(I(G)=+\infty \).

A spanning subgraph F of G is called a path-factor if each component of F is a path. A \(P_{\ge k}\)-factor of G means a path-factor such that each component is a path with at least k vertices, where \(k\ge 2\) is an integer.

Johnson et al. [7] studied the existence of path-factors in graphs. Egawa and Furuya [3] posed some sufficient conditions for a graph to admit a path-factor. Asratian and Casselgren [2] derived a sufficient condition for the existence of path-factors in graphs. Ando et al. [1] verified that a claw-free graph with minimum degree at least k admits a \(P_{\ge k+1}\)-factor. Kano, Lee and Suzuki [10] proved that every connected cubic bipartite graph with at least 8 vertices admits a \(P_{\ge 8}\)-factor. Johansson [6] gave an El-Zahár type condition ensuring path-factors in graphs. Kano, Lu and Yu [11] claimed that a graph G with \(i(G-X)\le \frac{2}{3}|X|\) for any \(X\subseteq V(G)\) admits a \(P_{\ge 3}\)-factor. Zhou [17,18,19], Zhou, Bian and Pan [20], Zhou, Sun and Liu [23], Zhou, Wu and Bian [24], Zhou, Wu and Xu [25],and Gao, Wang and Chen [5] presented some sufficient conditions for graphs admitting \(P_{\ge 3}\)-factors with given properties. Some other results on graph factors see [13, 16, 21]. Las Vergnas [12] showed a criterion for a graph with a \(P_{\ge 2}\)-factor.

Theorem 1

( [12]). A graph G has a \(P_{\ge 2}\)-factor if and only if

$$\begin{aligned} i(G-X)\le 2|X| \end{aligned}$$

for any \(X\subseteq V(G)\).

A graph R is called a factor-critical graph if for any \(v\in V(R)\), \(R-v\) has a 1-factor. A graph H is called a sun if \(H=K_1\), \(H=K_2\) or H is the corona of a factor-critical graph R with at least three vertices, namely, H is derived from R by adding a new vertex \(u=u(v)\) together with a new edge vu for each \(v\in V(R)\). A sun with at least six vertices is called a big sun. The number of sun components of G is denoted by sun(G). Kaneko [8] presented a characterization of a graph with a \(P_{\ge 3}\)-factor, and Kano, Katona and Király [9] gave a shorter proof.

Theorem 2

( [8, 9]). A graph G has a \(P_{\ge 3}\)-factor if and only if

$$\begin{aligned} sun(G-X)\le 2|X| \end{aligned}$$

for any \(X\subseteq V(G)\).

A graph G is called a \(P_{\ge k}\)-factor covered graph if for each \(e\in E(G)\), G has a \(P_{\ge k}\)-factor covering e, which was first defined by Zhang and Zhou [15]. Furthermore, they put forward two characterizations for a graph to a \(P_{\ge 2}\)-factor covered graph and \(P_{\ge 3}\)-factor covered graph, which are stated as follows.

Theorem 3

( [15]). A connected graph G is a \(P_{\ge 2}\)-factor covered graph if and only if

$$\begin{aligned} i(G-X)\le 2|X|-\varepsilon _1(X) \end{aligned}$$

for any \(X\subseteq V(G)\), where \(\varepsilon _1(X)\) is defined by

$$\begin{aligned} \varepsilon _1(X)=\left\{ \begin{array}{ll} 2,&{}if \ X \ is \ not \ an \ independent \ set;\\ 1,&{}if \ X \ is \ a \ nonempty \ independent \ set, \ and \ G-X \ admits\\ &{}a \ nontrivial \ component;\\ 0,&{}otherwise.\\ \end{array} \right. \end{aligned}$$

Theorem 4

( [15]). A connected graph G is a \(P_{\ge 3}\)-factor covered graph if and only if

$$\begin{aligned} sun(G-X)\le 2|X|-\varepsilon _2(X) \end{aligned}$$

for any subset X of V(G), where \(\varepsilon _2(X)\) is defined by

$$\begin{aligned} \varepsilon _2(X)=\left\{ \begin{array}{ll} 2,&{}if \ X \ is \ not \ an \ independent \ set;\\ 1,&{}if \ X \ is \ a \ nonempty \ independent \ set, \ and \ G-X \ admits\\ &{}a \ non-sun \ component;\\ 0,&{}otherwise.\\ \end{array} \right. \end{aligned}$$

A graph G is called a \(P_{\ge k}\)-factor uniform graph if for any two different edges \(e_1,e_2\in E(G)\), G has a \(P_{\ge k}\)-factor covering \(e_1\) and avoiding \(e_2\). In other word, a graph G is called a \(P_{\ge k}\)-factor uniform graph if for any \(e\in E(G)\), the graph \(G-e\) is a \(P_{\ge k}\)-factor covered graph. Gao and Wang [4] derived a result on the existence of \(P_{\ge 3}\)-factor uniform graphs. Zhou, Sun and Liu [22] posed two isolated toughness conditions for a graph to be a \(P_{\ge 2}\)-factor uniform graph and \(P_{\ge 3}\)-factor uniform graph, which are stated as follows.

Theorem 5

( [22]). A 3-edge-connected graph G is a \(P_{\ge 2}\)-factor uniform graph if its isolated toughness \(I(G)>1\).

Theorem 6

( [22]). A 3-edge-connected graph G is a \(P_{\ge 3}\)-factor uniform graph if its isolated toughness \(I(G)>2\).

In Theorems 5 and 6, the conditions on I(G) are sharp. However, it is natural to expect that we can weaken the condition on I(G) if we replace the assumption that G is 3-edge-connected by a stronger assumption. Along this line, we derive the following results.

Theorem 7

Let r be a nonnegative integer. An \((r+3)\)-edge-connected graph G is a \(P_{\ge 2}\)-factor uniform graph if its isolated toughness \(I(G)>\frac{r+3}{2r+3}\).

Theorem 8

Let r be a nonnegative integer. An \((r+3)\)-edge-connected graph G is a \(P_{\ge 3}\)-factor uniform graph if its isolated toughness \(I(G)>\frac{3r+6}{2r+3}\).

2 The proof of Theorem 7

Proof of Theorem 7

Theorem 7 holds obviously for a complete graph. Hence, we may assume that G is not a complete graph.

We proceed by contradiction. Assume that there exists an edge \(e=uv\) in G such that \(G'=G-e\) is not a \(P_{\ge 2}\)-factor covered graph. It follows from Theorem 3 that

$$\begin{aligned} i(G'-X)\ge 2|X|-\varepsilon _1(X)+1 \end{aligned}$$
(1)

for some vertex subset X of \(G'\).

The following proof will be divided into three cases by the value of |X|.

Case 1.\(0\le |X|\le r+1\).

According to (1) and \(\varepsilon _1(X)\le |X|\), we obtain

$$\begin{aligned} i(G'-X)\ge 2|X|-\varepsilon _1(X)+1\ge |X|+1, \end{aligned}$$

which implies that \(G'-X\) has at least one isolated vertex w, and so \(d_{G'-X}(w)=0\). Thus, we have

$$\begin{aligned} d_G(w)\le d_{G'}(w)+1\le d_{G'-X}(w)+|X|+1=|X|+1\le r+2, \end{aligned}$$

which contradicts that G is \((r+3)\)-edge-connected.

Case 2.\(|X|=r+2\).

In terms of (1) and \(\varepsilon _1(X)\le 2\), we admit

$$\begin{aligned} i(G'-X)\ge 2|X|-\varepsilon _1(X)+1\ge 2|X|-1. \end{aligned}$$
(2)

Note that \(i(G-X)\ge i(G'-X)-2\). Combining this with (2), we get

$$\begin{aligned} i(G-X)\ge i(G'-X)-2\ge 2|X|-3=2(r+2)-3=2r+1, \end{aligned}$$

which implies that there exists \(w\in V(G-X)\) with \(d_{G-X}(w)=0\). Since G is \((r+3)\)-edge-connected, we have

$$\begin{aligned} r+3\le d_G(w)\le d_{G-X}(w)+|X|=|X|=r+2, \end{aligned}$$

which is a contradiction.

Case 3.\(|X|\ge r+3\).

Note that \(i(G-X)\ge i(G'-X)-2\). It follows from (1) and \(\varepsilon _1(X)\le 2\) that

$$\begin{aligned} i(G-X)\ge i(G'-X)-2\ge 2|X|-\varepsilon _1(X)+1-2\ge 2|X|-3\ge 2r+3. \end{aligned}$$
(3)

In light of (3) and the definition of I(G), we derive

$$\begin{aligned} I(G)\le & {} \frac{|X|}{i(G-X)}\le \frac{|X|}{2|X|-3}=\frac{1}{2}+\frac{3}{4|X|-6}\\\le & {} \frac{1}{2}+\frac{3}{4(r+3)-6}=\frac{1}{2}+\frac{3}{4r+6}\\= & {} \frac{r+3}{2r+3}, \end{aligned}$$

which contradicts \(I(G)>\frac{r+3}{2r+3}\). The proof of Theorem 7 is complete. \(\square \)

Remark 1

We now show that \(I(G)>\frac{r+3}{2r+3}\) in Theorem 7 cannot be weakened to \(I(G)\ge \frac{r+3}{2r+3}\).

We construct a graph \(G=K_{r+3}\vee ((2r+3)K_1)\cup K_2\), where r is a nonnegative integer. Then \(I(G)=\frac{|V(K_{r+3})|}{i(G-V(K_{r+3}))} =\frac{r+3}{2r+3}\) and G is \((r+3)\)-edge-connected. Write \(G'=G-e\) for \(e\in E(K_2)\). Let \(X=V(K_{r+3})\). Then \(|X|=r+3\) and \(\varepsilon _1(X)=2\). Thus, we admit

$$\begin{aligned} i(G'-X)=2r+5>2r+4=2|X|-\varepsilon _1(X). \end{aligned}$$

In light of Theorem 3, \(G'\) is not a \(P_{\ge 2}\)-factor covered graph, and so G is not a \(P_{\ge 2}\)-factor uniform graph.

Remark 2

The condition that G is \((r+3)\)-edge-connected in Theorem 7 cannot be replaced by G being \((r+2)\)-edge-connected.

To show this, we construct a graph \(G=K_{r+2}\vee ((2r+1)K_1\cup K_2)\), where \(r\ge 1\) is an integer. Then G is \((r+2)\)-edge-connected and \(I(G)=\frac{|V(K_{r+2})|}{i(G-V(K_{r+2}))}=\frac{r+2}{2r+1}>\frac{r+3}{2r+3}\). Let \(G'=G-e\) for \(e\in E(K_2)\) and \(X=V(K_{r+2})\). Then \(\varepsilon _1(X)=2\), and so

$$\begin{aligned} i(G'-X)=2r+3>2r+2=2|X|-\varepsilon _1(X). \end{aligned}$$

According to Theorem 3, \(G'\) is not a \(P_{\ge 2}\)-factor covered graph, and so G is not a \(P_{\ge 2}\)-factor uniform graph.

3 The proof of Theorem 8

Proof of Theorem 8

Theorem 8 is true for a complete graph. Therefore, we may assume that G is not a complete graph.

Theorem 8 holds for \(r=0\) by Theorem 6. In what follows, we may assume \(r\ge 1\). We proceed by contradiction. Assume that there exists an edge \(e=uv\) in G such that \(G'=G-e\) is not a \(P_{\ge 3}\)-factor covered graph. Then by Theorem 4 we obtain

$$\begin{aligned} sun(G'-X)\ge 2|X|-\varepsilon _2(X)+1 \end{aligned}$$
(4)

for some vertex subset X of \(G'\).

Claim 1. \(|X|\ge r+2\).

Proof

If \(0\le |X|\le r\), then it follows from (4) and \(\varepsilon _2(X)\le |X|\) that

$$\begin{aligned} sun(G'-X)\ge 2|X|-\varepsilon _2(X)+1\ge |X|+1\ge 1, \end{aligned}$$

which implies that there exists \(w\in V(G'-X)\) such that \(d_{G'-X}(w)\le 1\). Thus, we derive

$$\begin{aligned} d_G(w)\le d_{G'}(w)+1\le d_{G'-X}(w)+|X|+1\le |X|+2\le r+2, \end{aligned}$$

which contradicts that G is \((r+3)\)-edge-connected.

If \(|X|=r+1\), the by (4), \(\varepsilon _2(X)\le 2\) and \(r\ge 1\), we have

$$\begin{aligned} sun(G-X)\ge sun(G'-X)-2\ge 2|X|-\varepsilon _2(X)+1-2\ge 2|X|-3=2r-1\ge 1, \end{aligned}$$

which implies that there exists \(w\in V(G-X)\) with \(d_{G-X}(w)\le 1\), and so

$$\begin{aligned} d_G(w)\le d_{G-X}(w)+|X|\le |X|+1=r+2, \end{aligned}$$

which contradicts that G is \((r+3)\)-edge-connected. Hence, \(|X|\ge r+2\). This completes the proof of Claim 1. \(\square \)

Assume that \(G'-X\) admits a isolated vertices, b \(K_2\)’s and c big sun components \(H_1,H_2,\cdots ,H_c\), where \(|V(H_i)|\ge 6\) for \(1\le i\le c\). In terms of (4), \(\varepsilon _2(X)\le 2\) and Claim 1, we infer

$$\begin{aligned} sun(G'-X)=a+b+c\ge 2|X|-\varepsilon _2(X)+1\ge 2|X|-1=2r+3. \end{aligned}$$
(5)

Write \(R_i\) for the factor-critical graph of \(H_i\), \(i=1,2,\cdots ,c\). Set \(D_i=V(R_i)\) and \(D=\bigcup \limits _{i=1}^{c}{D_i}\). Obviously, \(i(H_i-D_i)=|D_i|=\frac{|V(H_i)|}{2}\ge 3\) and \(|D|\ge 3c\). Select one vertex from each \(K_2\) component of \(G'-X\), and denote the set of such vertices by Q. We denote by W the union of all non-sun components of \(G'-X\).

Note that \(i(G'-X)-2\le i(G-X)\le i(G'-X)\). The following proof will be divided into three cases.

Case 1. \(i(G-X)=i(G'-X)\).

Clearly, \(u,v\notin V(aK_1)\) (otherwise, \(i(G-X)<i(G'-X)\), a contradiction).

Subcase 1.1. \(u\in V(W)\).

In this subcase, we get \(i(G-X-Q-D-u)\ge a+b+|D|\), and so

$$\begin{aligned} I(G)\le \frac{|X\cup Q\cup D\cup \{u\}|}{i(G-X-Q-D-u)}\le \frac{|X|+b+|D|+1}{a+b+|D|}. \end{aligned}$$

Combining this with \(I(G)>\frac{3r+6}{2r+3}\), we obtain

$$\begin{aligned} \frac{3r+6}{2r+3}<I(G)\le \frac{|X|+b+|D|+1}{a+b+|D|}, \end{aligned}$$

which implies

$$\begin{aligned} 0>(3r+6)a+(r+3)b+(r+3)|D|-(2r+3)|X|-(2r+3). \end{aligned}$$
(6)

It follows from (5), (6), \(|D|\ge 3c\) and Claim 1 that

$$\begin{aligned} 0> & {} (3r+6)a+(r+3)b+(r+3)|D|-(2r+3)|X|-(2r+3)\\\ge & {} (r+3)a+(r+3)b+(r+3)c-(2r+3)|X|-(2r+3)\\= & {} (r+3)(a+b+c)-(2r+3)|X|-(2r+3)\\\ge & {} (r+3)(2|X|-1)-(2r+3)|X|-(2r+3)\\= & {} 3|X|-(3r+6)=3(r+2)-(3r+6)=0, \end{aligned}$$

a contradiction.

Subcase 1.2. \(u\notin V(W)\).

In this subcase, \(u\in V(bK_2)\), or \(u\in D_i\) (\(1\le i\le c\)), or \(u\in V(H_i)\setminus D_i\) (\(1\le i\le c\)).

Claim 2. \(I(G)\le \frac{|X|+b+|D|}{a+b+|D|}\).

Proof

If \(u\in V(bK_2)\), then we choose such set Q with \(u\in Q\). Thus, we get \(i(G-X-Q-D)=a+b+|D|\), and so

$$\begin{aligned} I(G)\le \frac{|X\cup Q\cup D|}{i(G-X-Q-D)}=\frac{|X|+b+|D|}{a+b+|D|}. \end{aligned}$$

If \(u\in D_i\) (\(1\le i\le c\)), then we admit \(i(G-X-Q-D)=a+b+|D|\), and so

$$\begin{aligned} I(G)\le \frac{|X\cup Q\cup D|}{i(G-X-Q-D)}=\frac{|X|+b+|D|}{a+b+|D|}. \end{aligned}$$

If \(u\in V(H_i)\setminus D_i\) (\(1\le i\le c\)), then there exists \(w\in D_i\) with \(uw\in E(H_i)\). Thus, we derive \(i(G-X-Q-u-(D\setminus \{w\}))=a+b+|D|\), and so

$$\begin{aligned} I(G)\le \frac{|X\cup Q\cup \{u\}\cup (D\setminus \{w\})|}{i(G-X-Q-u-(D\setminus \{w\}))}=\frac{|X|+b+|D|}{a+b+|D|}. \end{aligned}$$

We finish the proof of Claim 2.

According to Claim 2 and \(I(G)>\frac{3r+6}{2r+3}\), we acquire

$$\begin{aligned} \frac{3r+6}{2r+3}<I(G)\le \frac{|X|+b+|D|}{a+b+|D|}, \end{aligned}$$

namely,

$$\begin{aligned} 0>(3r+6)a+(r+3)b+(r+3)|D|-(2r+3)|X|. \end{aligned}$$
(7)

It follows from (5), (7), \(|D|\ge 3c\) and Claim 1 that

$$\begin{aligned} 0> & {} (3r+6)a+(r+3)b+(r+3)|D|-(2r+3)|X|\\\ge & {} (r+3)(a+b+c)-(2r+3)|X|\\\ge & {} (r+3)(2|X|-1)-(2r+3)|X|\\= & {} 3|X|-(r+3)=3(r+2)-(r+3)=2r+3>0, \end{aligned}$$

which is a contradiction.

Case 2. \(i(G-X)=i(G'-X)-1\).

In this case, \(u\in V(aK_1)\) and \(v\notin V(aK_1)\), or \(u\notin V(aK_1)\) and \(v\in V(aK_1)\). Without loss of generality, let \(u\in V(aK_1)\) and \(v\notin V(aK_1)\). Obviously, \(a\ge 1\).

Claim 3. \(I(G)\le \frac{|X|+b+|D|+1}{a+b+|D|}\).

Proof

The proof is similar to that of Claim 2 by discussing \(v\in V(bK_2)\), \(v\in D_i\), \(v\in V(H_i)\setminus D_i\) or \(v\in V(W)\). Claim 3 is verified. \(\square \)

According to Claim 3 and \(I(G)>\frac{3r+6}{2r+3}\), we have

$$\begin{aligned} \frac{3r+6}{2r+3}<I(G)\le \frac{|X|+b+|D|+1}{a+b+|D|}, \end{aligned}$$

that is,

$$\begin{aligned} (2r+3)|X|>(3r+6)a+(r+3)b+(r+3)|D|-(2r+3). \end{aligned}$$
(8)

In terms of (5), (8), \(a\ge 1\) and \(|D|\ge 3c\), we deduce

$$\begin{aligned} (2r+3)|X|> & {} (3r+6)a+(r+3)b+(r+3)|D|-(2r+3)\\\ge & {} (3r+6)a+(r+3)b+(r+3)c-(2r+3)\\= & {} (r+3)(a+b+c)+(2r+3)(a-1)\\\ge & {} (r+3)(2|X|-1), \end{aligned}$$

which implies \(|X|<\frac{r+3}{3}\), which contradicts Claim 1.

Case 3. \(i(G-X)=i(G'-X)-2\).

In this case, \(u,v\in V(aK_1)\), and so \(a\ge 2\). Thus, we possess \(i(G-X-Q-v-D)=a+b+|D|-1\), and so

$$\begin{aligned} I(G)\le \frac{|X\cup Q\cup \{v\}\cup D|}{i(G-X-Q-v-D)}=\frac{|X|+b+|D|+1}{a+b+|D|-1}. \end{aligned}$$
(9)

In light of (9) and \(I(G)>\frac{3r+6}{2r+3}\), we derive

$$\begin{aligned} \frac{3r+6}{2r+3}<I(G)\le \frac{|X|+b+|D|+1}{a+b+|D|-1}, \end{aligned}$$

which implies

$$\begin{aligned} (2r+3)|X|>(3r+6)a+(r+3)b+(r+3)|D|-5r-9. \end{aligned}$$
(10)

It follows from (5), (10), \(a\ge 2\) and \(|D|\ge 3c\) that

$$\begin{aligned} (2r+3)|X|> & {} (3r+6)a+(r+3)b+(r+3)|D|-5r-9\\\ge & {} (3r+6)a+(r+3)b+(r+3)c-5r-9\\= & {} (r+3)(a+b+c)+(2r+3)a-5r-9\\\ge & {} (r+3)(2|X|-1)+2(2r+3)-5r-9\\= & {} 2(r+3)|X|-2r-6, \end{aligned}$$

which implies \(|X|<\frac{2r+6}{3}\), which contradicts Claim 1 by \(r\ge 1\). Theorem 8 is verified. \(\Box \)

Remark 3

In what follows, we claim that \(I(G)>\frac{3r+6}{2r+3}\) in Theorem 8 cannot be replaced by \(I(G)>\frac{3r+5}{2r+3}\).

To explain this, we construct a graph \(G=K_{r+3}\vee ((2r+4)K_2)\), where r is a nonnegative integer. We select one vertex from each \(K_2\) component of \(G-V(K_{r+3})\), and denote the set of such vertices by Y. Then \(\frac{3r+6}{2r+3}>I(G)=\frac{|V(K_{r+3})\cup Y|}{i(G-(V(K_{r+3})\cup Y))} =\frac{3r+7}{2r+4}>\frac{3r+5}{2r+3}\) and G is \((r+4)\)-edge-connected. Let \(G'=G-e\) for \(e\in E((2r+4)K_2)\). Let \(X=V(K_{r+3})\). Then \(|X|=r+3\) and \(\varepsilon _2(X)=2\). Thus, we obtain

$$\begin{aligned} sun(G'-X)=2r+5>2r+4=2|X|-\varepsilon _2(X). \end{aligned}$$

In view of Theorem 4, \(G'\) is not a \(P_{\ge 3}\)-factor covered graph, and so G is not a \(P_{\ge 3}\)-factor uniform graph.

Remark 4

We now claim that 3-edge-connected in Theorem 8 is best possible in some sense.

To show this, we construct a graph \(G=K_{r+1}\vee ((2r)K_2)\), where r is an integer with \(1\le r\le 2\). We select one vertex from each \(K_2\) component of \(G-V(K_{r+1})\), and denote the set of such vertices by Y. Then G is 2-edge-connected and \(I(G)=\frac{|V(K_{r+1})\cup Y|}{i(G-(V(K_{r+1})\cup Y))}=\frac{3r+1}{2r}>\frac{3r+6}{2r+3}\). Write \(G'=G-e\) for \(e\in E((2r)K_2)\) and \(X=V(K_{r+1})\). Then \(|X|=r+1\) and \(\varepsilon _2(X)=2\). Hence, we derive

$$\begin{aligned} sun(G'-X)=2r+1>2r=2|X|-\varepsilon _2(X). \end{aligned}$$

By virtue of Theorem 4, \(G'\) is not a \(P_{\ge 3}\)-factor covered graph, and so G is not a \(P_{\ge 3}\)-factor uniform graph.