1 Introduction

Apart from their physical importance (see, for example, [25, 34]), the interest in the study of spacelike submanifolds immersed in a Lorentzian space is motivated by their nice Bernstein-type properties. For instance, it was proved by Calabi [10] (for \(n\le 4\)) and by Cheng and Yau [14] (for all n) that the only complete maximal spacelike hypersurfaces of the Lorentz-Minkowski space \(\mathbb {L}^{n+1}\) are the spacelike hyperplanes. In [29], Nishikawa proved that a complete maximal spacelike hypersurface (that is, with mean curvature identically zero) in the de Sitter space \(\mathbb {S}_{1}^{n+1}\) must be totally geodesic. In [18], Goddard conjectured that the complete spacelike hypersurfaces of \(\mathbb {S}_{1}^{n+1}\) with constant mean curvature H must be totally umbilical. Ramanathan [32] proved Goddard’s conjecture in \(\mathbb {S}_{1}^{3}\) for \(0\le H\le 1\). Moreover, for \(H>1\), he showed that the conjecture is false, as can be seen from an example due to Dajczer and Nomizu in [16]. Independently, Akutagawa [2] proved that Goddard’s conjecture is true when either \(n=2\) and \(H^{2}\le 1\) or \(n\ge 3\) and \(H^{2}<\frac{4(n-1)}{n^{2}}\). He also constructed complete spacelike rotation surfaces in \(\mathbb {S}_{1}^{3}\) having constant mean curvature \(H>1\) and which are not totally umbilical. Next, Montiel [26] showed that Goddard’s conjecture is true provided that \(M^{n}\) is compact. Furthermore, he exhibited examples of complete spacelike hypersurfaces in \(\mathbb {S}_{1}^{n+1}\) with constant mean curvature H satisfying \(H^{2}\ge \frac{4(n-1)}{n^2}\) and being non totally umbilical, the so-called hyperbolic cylinders.

In higher codimension, Cheng [12] extended Akutagawa’s result for complete spacelike submanifolds with parallel mean curvature vector (that is, the mean curvature vector field is parallel as a section of the normal bundle) in the de Sitter space \(\mathbb {S}_{p}^{n+p}\) of index p. Afterwards, Aiyama [1] studied compact spacelike submanifolds in \(\mathbb {S}_{p}^{n+p}\) with parallel mean curvature vector and proved that if the normal connection of \(M^{n}\) is flat, then \(M^{n}\) is totally umbilical. Furthermore, she proved that a compact spacelike submanifold in \(\mathbb {S}_{p}^{n+p}\) with parallel mean curvature vector and nonnegative sectional curvature must be totally umbilical. Meanwhile, Alías and Romero [4] developed some integral formulas for compact spacelike submanifolds in \(\mathbb {S}_{p}^{n+p}\) which have a very clear geometric meaning and, as application, they obtained a Bernstein type result for complete maximal submanifolds in \(\mathbb {S}_{q}^{n+p}\), extending a previous result due to Ishihara [19]. Moreover, they extended Ramanathan’s result [32] showing that the only compact spacelike surfaces in \(\mathbb {S}_{p}^{2+p}\) with parallel mean curvature vector are the totally umbilical ones and, in particular, they also reproved Cheng’s result [14] establishing that every complete spacelike surface in \(\mathbb {S}_{p}^{2+p}\) with parallel mean curvature vector such that \(H^2<1\) is totally umbilical. Next, Li [22] showed that Montiel’s result [26] still holds for higher codimensional spacelike submanifolds in \(\mathbb {S}_{p}^{n+p}\). More recently, Araújo and Barbosa [6], assuming appropriated controls on the second fundamental form and on the scalar curvature, extended the techniques developed in [23, 33, 35] and proved that a compact spacelike submanifold in \(\mathbb {S}_{p}^{n+p}\) with nonzero mean curvature and parallel mean curvature vector must be isometric to a sphere.

When the ambient spacetime is the anti-de Sitter space \(\mathbb H_1^{n+1}\), Choi et al. [15] used the generalized maximum principle of Omori [30] and Yau [36] in order to obtain a Myers type theorem [28] concerning complete maximal spacelike hypersurfaces. More precisely, they showed that if the height function with respect to a timelike vector of such a hypersurface obeys a certain boundedness, then it must be totally geodesic. Extending a technique due to Yau [37], the first author jointly with Camargo [7] obtained another rigidity results to complete maximal spacelike hypersurfaces in \(\mathbb H_1^{n+1}\), imposing suitable conditions on both the norm of the second fundamental form and a certain height function naturally attached to the hypersurface. Afterwards, working with a suitable warped product model for an open subset of \(\mathbb H_1^{n+1}\), the same authors jointly with Caminha and Parente [8] extended the main result of [7] showing that if \(M^n\) is a complete spacelike hypersurface with constant mean curvature and bounded scalar curvature in \(\mathbb H_1^{n+1}\), such that the gradient of its height function with respect to a timelike vector has integrable norm, then \(M^n\) must be totally umbilical. More recently, the first author jointly Aquino [5] obtained another characterizations theorems concerning complete constant mean curvature spacelike hypersurfaces of \(\mathbb H_1^{n+1}\), under suitable constraints on the behavior of the Gauss mapping. In higher codimension, Ishihara [19] proved that a n-dimensional complete maximal spacelike submanifold immersed in the anti-de Sitter space \(\mathbb H_p^{n+p}\) of index p must have the squared norm of the second fundamental form bounded from above by np. Moreover, the only ones that attain this estimate are the hyperbolic cylinders. Later on, Cheng [13] obtained a refinement of Ishihara’s result [19] for the case of complete maximal spacelike surfaces immersed in \(\mathbb H_p^{2+p}\).

Motivated by the works above described, our purpose in this paper is to study the geometry of complete spacelike submanifolds immersed in the anti-de Sitter space \(\mathbb {H}^{n+p}_{q}\) of index q. In this setting, we extend the technique due to Alías and Romero in [4] and, under appropriated constraints on the Ricci curvature and second fundamental form, we show that a complete maximal spacelike submanifold \(M^n\) of \(\mathbb {H}^{n+p}_{q}\) must be totally geodesic (see Theorem 1 and Corollaries 1 and 2). Furthermore, we establish sufficient conditions to guarantee that a complete spacelike submanifold with nonzero parallel mean curvature vector \(\mathbf{H}\) in \(\mathbb {H}^{n+p}_{p}\) must be pseudo-umbilical, which means that \(\mathbf{H}\) is an umbilical direction (see Theorem 2 and Corollary 3). Our approach is based on a generalized form of a maximum principle at the infinity of Yau [37] (see Lemma 1 and Remark 1).

2 Preliminaries

Let \(\mathbb {R}^{n+p+1}_{q+1}\) be the \((n+p+1)\)-dimensional semi-Euclidean space endowed with metric tensor \(\langle ,\,\rangle \) of index q, with \(1\le q\le p\), given by

$$\begin{aligned} \langle v,w\rangle =\sum _{i=1}^{n+p-q}\!\!\!\!\!v_{i}w_{i}-\!\!\!\!\!\!\sum _{j=n+p-q+1}^{n+p+1}\!\!\!\!\!\!v_{j}w_{j}, \end{aligned}$$

and let \(\mathbb {H}^{n+p}_{q}\) be the \((n+p)\)-dimensional unitary anti-de Sitter space of index q, that is,

$$\begin{aligned} \mathbb {H}^{n+p}_{q}=\{x\in \mathbb {R}^{n+p+1}_{q+1}\,;\,\langle x,x\rangle =-1\}, \end{aligned}$$

which has constant sectional curvature equal to \(-1\).

Along this work, we will consider \(x:M^{n}\rightarrow \mathbb {H}^{n+p}_{q}\subset \mathbb {R}^{n+p+1}_{q+1}\) a spacelike submanifold isometrically immersed in \(\mathbb {H}^{n+p}_{q}\). We recall that a submanifold immersed is said to be spacelike if its induced metric is positive definite. In this setting, we will denote by \(\nabla ^{\circ }\), \(\overline{\nabla }\) and \(\nabla \) the Levi-Civita connections of \(\mathbb {R}^{n+p+1}_{q+1}\), \(\mathbb {H}^{n+p}_{q}\) and \(M^{n}\), respectively, and \(\nabla ^{\perp }\) will stand for the normal connection of \(M^{n}\) in \(\mathbb {H}^{n+p}_{q}\).

We will denote by \(\alpha \) the second fundamental form of \(M^{n}\) in \(\mathbb {H}^{n+p}_{q}\) and by \(A_{\xi }\) the shape operator associated to a fixed vector field \(\xi \) normal to \(M^{n}\) in \(\mathbb {H}^{n+p}_{q}\). We note that, for each \(\xi \in \mathfrak {X}^{\perp }(M)\), \(A_{\xi }\) is a symmetric endomorphism of the tangent space \(T_{x}M\) at \(x\in M^{n}\). Moreover, \(A_{\xi }\) and \(\alpha \) are related by

$$\begin{aligned} \langle A_{\xi }X,Y\rangle =\langle \alpha (X,Y),\xi \rangle , \end{aligned}$$
(2.1)

for all tangent vector fields \(X,Y\in \mathfrak {X}(M)\).

We also recall that the Gauss and Weingarten formulas of \(M^{n}\) in \(\mathbb {H}^{n+p}_{q}\) are given by

$$\begin{aligned} \nabla ^{\circ }_{X}Y=\overline{\nabla }_{X}Y+\langle X,Y\rangle x=\nabla _{X}Y+\alpha (X,Y)+\langle X,Y\rangle x, \end{aligned}$$
(2.2)

and

$$\begin{aligned} \nabla ^{\circ }_{X}\xi =\overline{\nabla }_{X}\xi =-A_{\xi }X+\nabla ^{\perp }_{X}\xi , \end{aligned}$$
(2.3)

for all tangent vector fields \(X,Y\in \mathfrak {X}(M)\) and normal vector field \(\xi \in \mathfrak {X}^{\perp }(M)\).

As in [31], the curvature tensor R of the spacelike submanifold \(M^{n}\) is given by

$$\begin{aligned} R(X,Y)Z=\nabla _{[X,Y]}Z-[\nabla _{X},\nabla _{Y}]Z, \end{aligned}$$

where [, ] denotes the Lie bracket and \(X,Y,Z\in \mathfrak {X}(M)\).

A well known fact is that the curvature tensor R of \(M^{n}\) can be described in terms of its second fundamental form \(\alpha \) and the curvature tensor \(\overline{R}\) of the ambient spacetime \(\mathbb {H}^{n+p}_{q}\) by the so-called Gauss equation, which is given by

$$\begin{aligned} \langle R(X,Y)Z,W\rangle= & {} \langle Y,Z\rangle \langle X,W\rangle -\langle X,Z\rangle \langle Y,W\rangle \nonumber \\&+\,\langle \alpha (X,W),\alpha (Y,Z)\rangle -\langle \alpha (X,Z),\alpha (Y,W)\rangle , \end{aligned}$$
(2.4)

for all tangent vector fields \(X,Y,Z,W\in \mathfrak {X}(M)\). Moreover, Codazzi equation asserts that

$$\begin{aligned} (\nabla _XA_\xi )Y=(\nabla _YA_\xi )X, \end{aligned}$$
(2.5)

for all \(X,Y\in \mathfrak {X}(M)\) and \(\xi \in \mathfrak {X}^\perp (M)\).

We will define the mean curvature vector of \(M^{n}\) in \(\mathbb {H}^{n+p}_{q}\) by

$$\begin{aligned} \mathbf{H}=\frac{1}{n}\mathrm{tr}(\alpha ). \end{aligned}$$

We recall that \(M^{n}\) is called maximal when \(\mathbf{H}\equiv 0\) and we say that \(M^{n}\) has parallel mean curvature vector when \(\nabla ^{\perp }_{X}{} \mathbf{H}\equiv 0\), for every \(X\in \mathfrak {X}(M)\). In this last case, when \(q=p\) and \(\mathbf{H}\ne 0\), we have that \(\langle \mathbf{H},\mathbf{H}\rangle \) is a negative constant along \(M^{n}\). Moreover, \(M^{n}\) is called totally geodesic when its second fundamental form \(\alpha \) vanishes identically and it is called totally umbilical when

$$\begin{aligned} \alpha (X,Y)=\langle X,Y\rangle \mathbf{H}, \end{aligned}$$
(2.6)

for all tangent vector fields \(X,Y\in \mathfrak {X}(M)\).

We close this section describing the main analytical tool which is used along the proofs of our results in the next sections. In [37] Yau, generalizing a previous result due to Gaffney [17], established the following version of Stokes’ Theorem on an n-dimensional, complete noncompact Riemannian manifold \(M^n\): if \(\omega \in \Omega ^{n-1}(M)\) is an integrable \((n-1)\)-differential form on \(M^n\), then there exists a sequence \(B_i\) of domains on \(M^n\) such that \(B_i\subset B_{i+1}\), \(M^n=\bigcup _{i\ge 1}B_i\) and

$$\begin{aligned} \lim _{i\rightarrow +\infty }\displaystyle \int _{B_i}d\omega =0. \end{aligned}$$

Suppose that \(M^n\) is oriented by the volume element dM. If \(\omega =\iota _XdM\) is the contraction of dM in the direction of a smooth vector field X on \(M^n\), then Caminha obtained a suitable consequence of Yau’s result, which can be regarded as an extension of Hopf’s maximum principle for complete Riemannian manifolds (cf. Proposition 2.1 of [9]). In what follows, \(\mathcal {L}^1(M)\) and \(\mathrm{div}\) denote the space of Lebesgue integrable functions and the divergence on \(M^{n}\), respectively.

Lemma 1

Let X be a smooth vector field on the n-dimensional complete noncompact oriented Riemannian manifold \(M^{n}\), such that \(\mathrm{div}X\) does not change sign on \(M^{n}\). If \(|X|\in \mathcal {L}^{1}(M)\), then \(\mathrm{div}X=0\).

Remark 1

Lemma 1 can also be seen as a consequence of the version of Stokes’ Theorem given by Karp in [20]. In fact, using Theorem in [20], condition \(|X|\in \mathcal L^1(M)\) can be weakened to the following technical condition:

$$\begin{aligned} \liminf _{r\rightarrow +\infty }\frac{1}{r}\int _{B(2r)\setminus B(r)}|X|dM=0, \end{aligned}$$

where B(r) denotes the geodesic ball of radius r center at some fixed origin \(o\in M^n\). See also Corollary 1 and Remark in [20] for some another geometric conditions guaranteing this fact.

Remark 2

Reasoning in a similar way of that in the beginning of Section 4 of [5] (see also Section 4 in [3]), it is not difficult to verify that there exist no n-dimensional compact (without boundary) spacelike submanifolds immersed in \(\mathbb {H}^{n+p}_{p}\). Motivated by this fact, along this paper we will deal with complete spacelike submanifolds.

3 Complete maximal submanifolds immersed in \(\mathbb {H}^{n+p}_{q}\)

Let \(a\in \mathbb {R}^{n+p+1}_{q+1}\) be a fixed arbitrary vector and put

$$\begin{aligned} a=a^{\top }+a^{N}-\langle a,x\rangle x, \end{aligned}$$
(3.1)

where \(a^{\top }\in \mathfrak {X}(M)\) and \(a^{N}\in \mathfrak {X}^{\perp }(M)\) denote, respectively, the tangential and normal components of a with respect to \(M^{n}\hookrightarrow \mathbb {H}^{n+p}_{q}\). By taking covariant derivative in (3.1) and using (2.2) and (2.3), we get for all tangent vector field \(X\in \mathfrak {X}(M)\) that

$$\begin{aligned} \nabla _{X}a^{\top }=A_{a^{N}}X+\langle a,x\rangle X \end{aligned}$$
(3.2)

and

$$\begin{aligned} \nabla _{X}^{\perp }a^{N}=-\alpha (a^{\top },X). \end{aligned}$$
(3.3)

Hence, from (2.1) and (3.2) we obtain

$$\begin{aligned} \mathrm{div}(a^{\top })=\mathrm{tr}(A_{a^{N}})+n\langle a,x\rangle =n\langle a,\mathbf{H}\rangle +n\langle a,x\rangle . \end{aligned}$$
(3.4)

Moreover, we also have that

$$\begin{aligned} \mathrm{tr}(\nabla _{a^{\top }}A_{\xi })= & {} \sum _{i}\langle \nabla _{a^{\top }}A_{\xi }e_{i},e_{i}\rangle -\sum _{i}\langle \nabla _{a^{\top }}e_{i},A_{\xi }e_{i}\rangle \\&+\,n\langle \nabla ^{\perp }_{a^{\top }}{} \mathbf{H},\xi \rangle -\sum _{i}a^{\top }\langle A_{\xi }e_{i},e_{i}\rangle .\nonumber \end{aligned}$$

So, considering a local orthonormal frame \(\{e_{1},\ldots ,e_{n}\}\) on \(M^{n}\) such that \(A_{\xi }e_{i}=\lambda ^{\xi }_{i}e_{i}\), with a straightforward computation we can verify that

$$\begin{aligned} \mathrm{tr}(\nabla _{a^{\top }}A_{\xi })=n\langle \nabla ^{\perp }_{a^{\top }}{} \mathbf{H},\xi \rangle . \end{aligned}$$
(3.5)

From Codazzi Eq. (2.5) jointly with (3.2) and (3.5) we obtain, for all \(\xi \in \mathfrak {X}^{\perp }(M)\),

$$\begin{aligned} \mathrm{div}(A_{\xi }a^{\top })= & {} n\langle \nabla _{a^{\top }}^{\perp }{} \mathbf{H},\xi \rangle +\mathrm{tr}(A_{a^{N}}\circ A_{\xi })+\langle a,x\rangle \mathrm{tr}(A_{\xi })\nonumber \\&+\sum _{i}\langle \alpha (a^{\top },e_{i}),\nabla _{e_{i}}^{\perp }\xi \rangle . \end{aligned}$$
(3.6)

On the other hand, taking the trace in Gauss Eq. (2.4), we have

$$\begin{aligned} \mathrm{Ric}(X,Y)=-(n-1)\langle X,Y\rangle +n\langle \alpha (X,Y),\mathbf{H}\rangle -\sum _{i}\langle \alpha (X,e_{i}),\alpha (Y,e_{i})\rangle , \end{aligned}$$
(3.7)

where \(\mathrm{Ric}\) denotes the Ricci curvature of \(M^{n}\). Considering \(X=Y=a^{\top }\) in (3.7), we obtain

$$\begin{aligned} \mathrm{Ric}(a^{\top },a^{\top })= & {} -(n-1)\big |a^{\top }\big |^{2}+n\langle \alpha (a^{\top },a^{\top }),\mathbf{H}\rangle \nonumber \\&\!\!\!-\sum _{i}\langle \alpha (a^{\top },e_{i}),\alpha (a^{\top },e_{i})\rangle . \end{aligned}$$
(3.8)

Furthermore, from (3.3) and (3.6) we get

$$\begin{aligned} \mathrm{div}(A_{a^{N}}a^{\top })= & {} n\langle \nabla _{a^{\top }}^{\perp }{} \mathbf{H},a^{N}\rangle +\mathrm{tr}(A_{a^{N}}^{2})+\langle a,x\rangle \mathrm{tr}(A_{a^{N}})\nonumber \\&-\sum _{i}\langle \alpha (a^{\top },e_{i}),\alpha (a^{\top },e_{i})\rangle . \end{aligned}$$
(3.9)

Hence, from (3.8) and (3.9) we conclude that

$$\begin{aligned} \mathrm{div}(A_{a^{N}}a^{\top })= & {} n\langle \nabla _{a^{\top }}^{\perp }{} \mathbf{H},a^{N}\rangle +\mathrm{tr}(A_{a^{N}}^{2})+\langle a,x\rangle \mathrm{tr}(A_{a^{N}})\nonumber \\&+\,\mathrm{Ric}(a^{\top },a^{\top })+(n-1)\big |a^{\top }\big |^{2}-n\langle \alpha (a^{\top },a^{\top }),\mathbf{H}\rangle . \end{aligned}$$
(3.10)

Based on the previous computations, we obtain the following Bernstein type result concerning maximal submanifolds immersed in \(\mathbb {H}^{n+p}_{q}\)

Theorem 1

Let \(M^{n}\) be a complete maximal spacelike submanifold immersed in \(\mathbb {H}^{n+p}_{q}\), with \(1\le q\le p\). Suppose that \(Ric\ge -(n-1)\) on \(M^{n}\). If there exist p vectors \(a_1,\ldots ,a_p\in \mathbb {R}^{n+p+1}_{q+1}\) such that \(a_1^N,\ldots ,a_p^N\) are linearly independent, with \(A_{a_i^{N}}\) bounded on \(M^{n}\) and \(|a_i^{\top }|\in \mathcal {L}^{1}(M)\) for each \(1\le i\le p\), then \(M^{n}\) is totally geodesic.

Proof

Let us consider \(a=a_i\) for some \(i\in \{1,\dots ,p\}\). Provided that \(\mathbf{H}=0\), from (2.1) we see that Eq. (3.10) can be rewritten as follows

$$\begin{aligned} \mathrm{div}(A_{a^{N}}a^{\top })=\mathrm{tr}(A_{a^{N}}^{2})+\mathrm{Ric}(a^{\top },a^{\top })+(n-1)\big |a^{\top }\big |^{2}. \end{aligned}$$
(3.11)

Thus, since \(\mathrm{Ric}(a^{\top },a^{\top })\ge -(n-1)\big |a^{\top }\big |^{2}\), from (3.11) we obtain that

$$\begin{aligned} \mathrm{div}(A_{a^{N}}a^{\top })\ge 0. \end{aligned}$$
(3.12)

Moreover, whereas \(A_{a^{N}}\) is bounded on \(M^{n}\), \(|A_{a^{N}}|\le C_{1}\), for some constant \(C_{1}>0\). Thus, as we are assuming that \(|a^{\top }|\in \mathcal {L}^{1}(M)\), we have

$$\begin{aligned} |A_{a^{N}}a^{\top }|\le |A_{a^{N}}||a^{\top }|\le C_{1}|a^{\top }|\in \mathcal {L}^{1}(M). \end{aligned}$$
(3.13)

Hence, taking into account (3.12) and (3.13), we can apply Lemma 1 to guarantee that \(\mathrm{div}(A_{a^{N}}a^{\top })=0\). Consequently, returning to Eq. (3.11), we conclude that \(A_{a^{N}}\equiv 0\). Therefore, since \(\alpha (X,Y)=\sum _{i=1}^p\langle A_{a_i^{N}}X,Y\rangle a_i^{N}\), we have that \(M^{n}\) must be totally geodesic. \(\square \)

Remark 3

Despite our assumption on the \(a_{i}^{N}\) in Theorem 1 to be a technical hypothesis, it is motivated by the fact that it occurs in a natural way in the context of spacelike hypersurfaces (see Corollary 2). In this sense, it is a mild hypothesis.

In the case that \(p=q\), being \(M^{n}\) a maximal submanifold of \(\mathbb {H}^{n+p}_{p}\), a classical result due to Ishihara [19] assures us that \(|A|^{2}\le np\) (see also Cheng [13] for the case \(n=2\)). Moreover, each maximal submanifold in \(\mathbb {H}^{n+p}_{p}\) meets the condition \(\mathrm{Ric}\ge -(n-1)\). Thus, as a consequence of Theorem 1 we obtain

Corollary 1

Let \(M^{n}\) be a complete maximal spacelike submanifold immersed in \(\mathbb {H}^{n+p}_{p}\). If there exist p vectors \(a_1,\ldots ,a_p\in \mathbb {R}^{n+p+1}_{p+1}\) such that \(a_1^N,\ldots ,a_p^N\) are linearly independent, with \(|a_i^{\top }|\in \mathcal {L}^{1}(M)\) for each \(1\le i\le p\), then \(M^{n}\) is totally geodesic.

Taking into account that the warped product model \(-(-\frac{\pi }{2},\frac{\pi }{2})~\times _{\cos t}~\mathbb H^n\), which is considered in [7] in order to prove their results, models just only an open subset of \(\mathbb {H}^{n+1}_{1}\) (cf. Example 4.3 of [27]), from Corollary 1 we obtain the following improvement of Theorem 1.2 of [7]

Corollary 2

Let \(M^{n}\) be a complete maximal spacelike hypersurface immersed in \(\mathbb {H}^{n+1}_{1}\). If there exists a vector \(a\in \mathbb {R}^{n+2}_{2}\) such that \(a^N\) does not vanish on \(M^{n}\) and \(|a^{\top }|\in \mathcal {L}^{1}(M)\), then \(M^{n}\) is a totally geodesic hyperbolic space.

4 Submanifolds with parallel mean curvature vector in \(\mathbb {H}^{n+p}_{p}\)

In this section, we study the rigidity of a complete spacelike submanifold \(M^n\) of \(\mathbb {H}^{n+p}_{p}\) with nonzero parallel mean curvature vector H. For this, fixed a nonzero vector \(a\in \mathbb {R}^{n+p+1}_{p+1}\), we observe that Eq. (3.4) gives us

$$\begin{aligned} \mathrm{div}\left( \langle a,\mathbf{H}\rangle a^{\top }\right)= & {} \dfrac{1}{n}\mathrm{tr}(A_{a^{N}})^{2}+\langle a,x\rangle \mathrm{tr}(A_{a^{N}})\nonumber \\&-\langle \alpha (a^{\top },a^{\top }),\mathbf{H}\rangle +\langle a,\nabla ^{\perp }_{a^{\top }}{} \mathbf{H}\rangle . \end{aligned}$$
(4.1)

Thus, from (3.10) jointly with (4.1) we obtain

$$\begin{aligned} \mathrm{div}\left[ (A_{a^{N}}-\langle a,\mathbf{H}\rangle I)a^{\top }\right]= & {} (n-1)\langle \nabla _{a^{\top }}^{\perp }{} \mathbf{H},a^{N}\rangle +\mathrm{tr}(A_{a^{N}}^{2})\nonumber \\&\!\!-\dfrac{1}{n}\mathrm{tr}(A_{a^{N}})^{2}+T(a^{\top },a^{\top }), \end{aligned}$$
(4.2)

where I denotes the identity operator in the algebra of smooth vector fields on \(M^n\) and, following the terminology established in [4], T stands for a covariant tensor on \(M^n\) which is given by

$$\begin{aligned} T(X,X)=\mathrm{Ric}(X,X)+(n-1)\big |X\big |^{2}-(n-1)\langle \alpha (X,X),\mathbf{H}\rangle . \end{aligned}$$
(4.3)

According to [4, 11], a spacelike submanifold \(M^{n}\) of \(\mathbb {H}^{n+p}_{p}\) with nonzero mean curvature vector \(\mathbf{H}\) is said pseudo-umbilical if \(\mathbf{H}\) is an umbilical direction. From (2.6) we see that a totally umbilical spacelike submanifold is always pseudo-umbilical. Conversely, we get

Proposition 1

Let \(M^{n}\) be a complete pseudo-umbilical spacelike submanifold with nonzero parallel mean curvature vector \(\mathbf{H}\) in \(\mathbb {H}^{n+p}_{p}\). If there exist p vectors \(a_1,\ldots ,a_p\in \mathbb {R}^{n+p+1}_{p+1}\) such that \(a_1^N,\ldots ,a_p^N\) are linearly independent, with \(\langle a_i,\mathbf{H}\rangle \) and \(A_{a_i^{N}}\) bounded on \(M^{n}\) and \(|a_i^{\top }|\in \mathcal {L}^{1}(M)\) for each \(1\le i\le p\), then \(M^{n}\) is totally umbilical.

Proof

Let us consider \(a=a_i\) for some \(i\in \{1,\dots ,p\}\). We have that

$$\begin{aligned} \big |(A_{a^{N}}-\langle a,\mathbf{H}\rangle I)a^{\top }\big |\le \left( |A_{a^{N}}|+|\langle a,\mathbf{H}\rangle |\right) |a^{\top }|\le C_{2}|a^{\top }|\in \mathcal {L}^{1}(M). \end{aligned}$$
(4.4)

Since we are assuming that \(M^{n}\) is pseudo-umbilical of \(\mathbb {H}^{n+p}_{p}\), Lemma 4.1 of [4] assures that

$$\begin{aligned} \mathrm{Ric}(X,X)\ge -(n-1)|X|^{2}+(n-1)\langle \alpha (X,X),\mathbf{H}\rangle , \end{aligned}$$
(4.5)

for all \(X\in \mathfrak {X}(M)\). Thus, from (4.3) and  (4.5) we get that \(T(a^{\top },a^{\top })\ge 0\). Moreover, we observe that the function \(u=\mathrm{tr}(A_{a^{N}}^{2})-\dfrac{1}{n}\mathrm{tr}(A_{a^{N}})^{2}\) is always nonnegative with \(u=0\) if, and only if, \(a^{N}\) is a umbilical direction. From (4.2), we obtain

$$\begin{aligned} \mathrm{div}\left[ (A_{a^{N}}-\langle a,\mathbf{H}\rangle I)a^{\top }\right] \ge 0. \end{aligned}$$
(4.6)

Thus, from (4.4) and (4.6), Lemma 1 assure us

$$\begin{aligned} \mathrm{tr}(A_{a^{N}}^{2})-\dfrac{1}{n}\mathrm{tr}(A_{a^{N}})^{2}+T(a^{\top },a^{\top })=0. \end{aligned}$$

Then, \(\mathrm{tr}(A_{a^{N}}^{2})-\dfrac{1}{n}\mathrm{tr}(A_{a^{N}})^{2}=0\) and, hence, \(a^{N}\) is a umbilical direction of \(M^n\). Therefore, since we are supposing the existence of such vectors \(a_1,\ldots ,a_p\in \mathbb {R}^{n+p+1}_{p+1}\) whose normal projections \(a_1^N,\ldots ,a_p^N\) with respect to \(M^n\) are linearly independent, we conclude that (2.6) holds, that is, \(M^{n}\) must be totally umbilical. \(\square \)

Proceeding, we establish sufficient conditions to guarantee that a spacelike submanifold immersed in \(\mathbb {H}^{n+p}_{p}\) with nonzero parallel mean curvature vector must be pseudo-umbilical.

Theorem 2

Let \(M^{n}\) be a complete spacelike submanifold immersed in \(\mathbb {H}^{n+p}_{p}\) with nonzero parallel mean curvature vector \(\mathbf{H}\) and bounded normalized scalar curvature R. If there exists a nonzero vector \(a\in \mathbb {R}^{n+p+1}_{p+1}\) such that \(a^{N}\) is timelike, i.e., \(\langle a^{N},a^{N}\rangle <0\), collinear to \(\mathbf{H}\) and \(|a^{\top }|\in \mathcal {L}^{1}(M)\), then \(M^{n}\) is pseudo-umbilical.

Proof

Initially, taking a local orthonormal frame \(\{e_{1},\ldots ,e_{n}\}\) on \(M^n\), from (3.7) we get that the squared norm of second form fundamental \(\alpha \) of \(M^{n}\) satisfies

$$\begin{aligned} |\alpha |^{2}=\sum _{i,j}|\alpha (e_{i},e_{j})|^{2}=n^{2}\langle \mathbf{H},\mathbf{H}\rangle +n(n-1)(R+1). \end{aligned}$$
(4.7)

Now, let us consider a nonzero vector \(a\in \mathbb {R}^{n+p+1}_{p+1}\) such that \(a^{N}\) is timelike, collinear to \(\mathbf{H}\) and with \(|a^{\top }|\in \mathcal {L}^{1}(M)\). Since \(M^{n}\) has bounded normalized scalar curvature and nonzero parallel mean curvature vector \(\mathbf{H}\), from (4.7) we conclude that \(|\alpha |^{2}\) is bounded on \(M^{n}\). So, taking \(\xi =\mathbf{H}\) in (3.6) we get

$$\begin{aligned} \mathrm{div}(A_{\mathbf{H}}a^{\top })=\mathrm{tr}(A_{a^{N}}\circ A_{\mathbf{H}})+\langle a,x\rangle \mathrm{tr}(A_{\mathbf{H}}), \end{aligned}$$
(4.8)

where \(A_{\mathbf{H}}\) denotes the Weingarten operator associated to \(\mathbf{H}\).

On the other hand, from (3.4) we have

$$\begin{aligned} \langle a,x\rangle =\dfrac{1}{n}\mathrm{div}(a^{\top })-\langle a,\mathbf{H}\rangle . \end{aligned}$$
(4.9)

Consequently, from (4.8) and (4.9)

$$\begin{aligned} \mathrm{div}(A_{\mathbf{H}}a^{\top })=\mathrm{tr}(A_{a^{N}}\circ A_{\mathbf{H}})+\mathrm{tr}(A_{\mathbf{H}})\dfrac{1}{n}\mathrm{div}(a^{\top })-\dfrac{1}{n}\mathrm{tr}({A_{a^{N}}})\mathrm{tr}(A_{\mathbf{H}}). \end{aligned}$$
(4.10)

Since

$$\begin{aligned} \mathrm{div}\left( \mathrm{tr}(A_{\mathbf{H}})a^{\top }\right) =\mathrm{tr}(A_{\mathbf{H}})\mathrm{div}(a^{\top }), \end{aligned}$$
(4.11)

from (4.10) and (4.11) we obtain

$$\begin{aligned} \mathrm{div}\,V=\mathrm{tr}(A_{a^{N}}\circ A_{\mathbf{H}})-\dfrac{1}{n}\mathrm{tr}({A_{a^{N}}})\mathrm{tr}(A_{\mathbf{H}}), \end{aligned}$$
(4.12)

where V is a tangent vector field on \(M^n\) given by

$$\begin{aligned} V=\left( A_{\mathbf{H}}-\dfrac{1}{n}\mathrm{tr}(A_{\mathbf{H}})I\right) a^{\top }. \end{aligned}$$

We note that, since we are supposing \(a^N\) timelike and collinear to \(\mathbf{H}\), there exists on \(M^n\) a smooth function \(\lambda \) having strict sign such that \(a^{N}=\lambda \mathbf{H}\). Thus, from (2.3) and (4.12) we get

$$\begin{aligned} \mathrm{div}\,V=\lambda \left( \mathrm{tr}(A^{2}_{\mathbf{H}})-\dfrac{1}{n}\mathrm{tr}(A_{\mathbf{H}})^{2}\right) . \end{aligned}$$
(4.13)

Consequently, from (4.13) we conclude that \(\mathrm{div}\,V\) does not change sign on \(M^n\). Moreover, we also have that

$$\begin{aligned} |V|\le \left( |A_{\mathbf{H}}|+|\langle \mathbf{H},\mathbf{H}\rangle |\right) |a^{\top }|\in \mathcal {L}^{1}(M). \end{aligned}$$

Hence, we can apply once more Lemma 1 to assure that \(\mathrm{div}\,V=0\) on \(M^n\).

Therefore, returning to (4.13) we obtain that

$$\begin{aligned} \lambda \left( \mathrm{tr}(A^{2}_{\mathbf{H}})-\dfrac{1}{n}\mathrm{tr}(A_{\mathbf{H}})^{2}\right) =0, \end{aligned}$$

which implies that \(\mathbf{H}\) is an umbilical direction. \(\square \)

We observe that, in the case \(p=1\), the notion of pseudo-umbilical coincides with that of totally umbilical. Moreover, we note that the hypothesis that \(a^{N}\) is timelike amounts to the support function \(f_a=\langle a,\nu \rangle \) having strict sign on the spacelike hypersurface \(M^n\hookrightarrow \mathbb {H}^{n+1}_{1}\), where \(\nu \) stands for the Gauss mapping of \(M^n\). Consequently, taking into account the classification of the totally umbilical hypersurfaces of \(\mathbb {H}^{n+1}_{1}\) (see, for instance, Example 1 of [24]) and that Theorem 1 of [21] assures us that a complete constant mean curvature spacelike hypersurface of \(\mathbb {H}^{n+1}_{1}\) must have bounded second fundamental form (or, equivalently, bounded normalized scalar curvature), from Theorem 2 we obtain the following

Corollary 3

Let \(M^{n}\) be a complete spacelike hypersurface immersed in \(\mathbb {H}^{n+1}_{1}\) with nonzero constant mean curvature. If there exists a nonzero vector \(a\in \mathbb {R}^{n+2}_{2}\) such that the support function \(f_a\) has strict sign on \(M^n\) and \(|a^{\top }|\in \mathcal {L}^{1}(M)\), then \(M^{n}\) is a totally umbilical hyperbolic space.