H H -tensional hypersurfaces in 4-dimensional space forms Bouazza Kacimi, Ahmed Mohammed Cherif, Mustafa Özkan Abstract . In this paper, we investigate the classification of H H -tensional hypersurfaces M M in a 4 4 -dimensional space form N 4 ​ ( c ) N^{4}(c) of constant sectional curvature c c . Our results show that minimal hypersurfaces are the only H H -tensional hypersurfaces in 4 4 -dimensional space forms, thereby providing an affirmative partial answer to the Conjecture 3 proposed in [ 8 ] . 1991 Mathematics Subject Classification Primary 53C21; Secondary 53C50 1. Introduction Let \psi:(M,g)\rightarrow(N,h) be a smooth map between Riemannian manifolds of dimensions m and n , respectively. It is well known that \psi is harmonic (see [ 2 , 7 ] ) if its tension field \tau(\psi) vanishes; that is, \tau(\psi)=\operatorname{Tr}_{g}(\nabla d\psi)=0. A natural generalization of harmonic maps is the H -tensional map, introduced in [ 8 ] . A map \psi is said to be H -tensional if its tension field \tau(\psi) satisfies \Delta^{\psi}\tau(\psi)=0, where \Delta^{\psi} is the rough Laplacian. In [ 8 ] , several properties of H -tensional maps are investigated. In particular, for an isometric immersion \iota:M^{m}\rightarrow N^{n} with mean curvature vector H , we introduce the notion of H -tensional submanifolds by requiring that \Delta^{\iota}\tau(\iota)=\Delta^{\iota}(mH)=0 . We prove that an H -tensional submanifold M is minimal if it satisfies one of the following conditions: (1) M is a curve; (2) M has constant mean curvature; (3) M is compact; (4) M is a hypersurface with at most two distinct principal curvatures in a real space form; (5) M is a pseudo-umbilical submanifold of dimension m\neq 4 ; (6) M is a totally umbilical hypersurface in an Einstein space. Motivated by these nonexistence results for nonminimal H -tensional submanifolds, the following conjectures were proposed in [ 8 ] : Conjecture 1. The only H -tensional submanifolds of real space forms are the minimal ones. Conjecture 2. The only complete H -tensional submanifolds of real space forms are the minimal ones. These conjectures generalize, respectively, Chen’s biharmonic conjecture (see [ 4 ] ) and its global version (see [ 1 ] ). For recent developments related to these problems, we refer to [ 4 ] and the references therein. In this paper, by adapting a method developed by Defever in [ 6 ] and by Balmuş, Montaldo, and Oniciuc [ 3 ] , we prove that any H -tensional hypersurface with at most three distinct principal curvatures in a 4 -dimensional space form is minimal (Theorem 2.2 ). This result supports Conjecture 1. 2. H -tensional hypersurfaces in 4-dimensional space forms Let \iota:M^{m}\longrightarrow N^{m+1}(c) be the canonical inclusion of a hypersurface M in a manifold with constant sectional curvature c , N^{m+1}(c) . Then, we have the following characterization result obtained in [ 8 ] : Theorem 2.1 . A hypersurface \iota:M^{m}\longrightarrow N^{m+1}(c) in a space of constant sectional curvature c with mean curvature vector H=\alpha e_{m+1} is H -tensional if and only if \begin{cases}\Delta\alpha+\alpha|A|^{2}=0,\\ \frac{m}{4}\operatorname{grad}\alpha^{2}+A(\operatorname{grad}\alpha)=0,\end{cases} (2.1) where A denotes the Weingarten operator and \Delta\alpha is the Laplacian of \alpha . Theorem 2.2 . Let M^{3} be a hypersurface of the space form N^{4}(c) . Then, M is an H -tensional submanifold if and only if it is minimal. Proof. Suppose that the mean curvature function \alpha is not constant on M . Then, there exists an open subset U of M such that (\operatorname{grad}\alpha^{2})(p)\neq 0 for all p\in U . We may assume that \alpha>0 on U , and thus (\operatorname{grad}\alpha)(p)\neq 0 for all p\in U . If U has at most two distinct principal curvatures, then by Theorem 4.1 in [ 8 ] , we conclude that its mean curvature is constant, yielding a contradiction. Without loss of generality, we can suppose that all points in U have three distinct principal curvatures. On U , we set e_{4}=\frac{H}{|H|} and denote by \alpha=|H| the mean curvature function of U in N^{4}(c) , and let k_{i} ( i=1,2,3 ) be the principal curvatures with respect to e_{4} . Consequently, the hypothesis that M is a nonminimal H -tensional hypersurface with at most three distinct principal curvatures in N^{4}(c) and non-constant mean curvature implies the existence of an open connected subset U of M where (\operatorname{grad}\alpha)(p)\neq 0 and \alpha(p)>0 for all p\in U . We shall now contradict the condition (\operatorname{grad}\alpha)(p)\neq 0 . Since M is H -tensional in N^{4}(c) , the characterization ( 2.1 ) yields \begin{cases}\Delta\alpha=-|A|^{2}\alpha,\\ A(\operatorname{grad}\alpha)=-\frac{3}{2}\alpha\operatorname{grad}\alpha.\end{cases} By setting e_{1}=\frac{\operatorname{grad}\alpha}{|\operatorname{grad}\alpha|} on U , it follows that e_{1} is a principal direction with principal curvature k_{1}=-\frac{3}{2}\alpha . Recalling that 3\alpha=k_{1}+k_{2}+k_{3} , it follows that k_{2}+k_{3}=\frac{9}{2}\alpha. (2.2) Let \{e_{1},e_{2},e_{3}\} be an orthonormal frame field of principal directions and \{\omega^{a}\}_{a=1}^{3} be the dual frame field to \{e_{a}\}_{a=1}^{3} on U . Since e_{1} is parallel to \operatorname{grad}\alpha , it is clear that e_{i}(\alpha)=\langle e_{i},\operatorname{grad}\alpha\rangle=0\quad\text{for }i=2,3, which implies \operatorname{grad}\alpha=e_{1}(\alpha)e_{1}. (2.3) We write the covariant derivatives as \nabla e_{a}=\sum_{b=1}^{3}\omega^{b}_{a}\otimes e_{b},\quad\omega^{b}_{a}\in\mathcal{C}^{\infty}(T^{*}U). From the Codazzi equations for M , for mutually distinct indices a,b,d\in\{1,2,3\} , we obtain e_{a}(k_{b})=(k_{a}-k_{b})\,\omega^{b}_{a}(e_{b}), (2.4) and (k_{b}-k_{d})\,\omega^{d}_{b}(e_{a})=(k_{a}-k_{d})\,\omega^{d}_{a}(e_{b}). (2.5) Applying ( 2.4 ) first with a=1 and b=i , and then with a=i and b=j for i\neq j , we find \omega^{1}_{i}(e_{i})=\frac{e_{1}(k_{i})}{k_{i}-k_{1}},\quad\omega^{i}_{j}(e_{j})=\frac{e_{i}(k_{j})}{k_{j}-k_{i}}. For a=i and b=1 , since e_{i}(k_{1})=0 , ( 2.4 ) implies that \omega^{1}_{i}(e_{1})=0 , which allows us to write \omega^{1}_{a}(e_{1})=0\quad\text{for }a=1,2,3. Furthermore, since e_{i}(\alpha)=0 implies \langle[e_{i},e_{j}],e_{1}\rangle=0 , it follows that \omega^{j}_{1}(e_{i})=\omega^{i}_{1}(e_{j}) . Now, from ( 2.5 ), setting a=1 , b=i , and d=j with i\neq j , we obtain \omega^{1}_{2}(e_{3})=\omega^{2}_{3}(e_{1})=\omega^{3}_{1}(e_{2})=0. Consequently, the connection 1 -forms are determined by the following relations: \left\{ \begin{aligned} \omega^{1}_{2}(e_{1})&=0,&\quad&\omega^{1}_{2}(e_{2})=\frac{e_{1}(k_{2})}{k_{2}+\frac{3}{2}\alpha}=:\beta_{2},&\quad&\omega^{1}_{2}(e_{3})=0,\\ \omega^{1}_{3}(e_{1})&=0,&&\omega^{1}_{3}(e_{2})=0,&&\omega^{1}_{3}(e_{3})=\frac{e_{1}(k_{3})}{k_{3}+\frac{3}{2}\alpha}=:\beta_{3},\\ \omega^{2}_{3}(e_{1})&=0,&&\omega^{2}_{3}(e_{2})=\frac{e_{3}(k_{2})}{k_{3}-k_{2}}=:\rho_{2},&&\omega^{2}_{3}(e_{3})=\frac{e_{2}(k_{3})}{k_{3}-k_{2}}=:\rho_{3}.\end{aligned}\right. (2.6) By invoking ( 2.2 ), the squared norm of the second fundamental form is given by \displaystyle|A|^{2} \displaystyle=k_{1}^{2}+k_{2}^{2}+k_{3}^{2} \displaystyle=k_{1}^{2}+(k_{2}+k_{3})^{2}-2k_{2}k_{3} \displaystyle=\frac{45}{2}\alpha^{2}-2\kappa, (2.7) where \kappa=k_{2}k_{3} . Next, by ( 2.3 ), the Laplacian of \alpha is computed as follows: \displaystyle\Delta\alpha \displaystyle=\sum_{i=1}^{3}[(\nabla_{e_{i}}e_{i})\alpha-e_{i}(e_{i}(\alpha))] \displaystyle=\sum_{i=1}^{3}\left[\sum_{k=1}^{3}\omega_{i}^{k}(e_{i})(e_{k}(\alpha))-e_{i}(e_{i}(\alpha))\right] \displaystyle=e_{1}(\alpha)(\beta_{2}+\beta_{3})-e_{1}(e_{1}(\alpha)). (2.8) Now, substituting ( 2 ) and ( 2 ) into the equation \Delta\alpha=-|A|^{2}\alpha , we obtain e_{1}(e_{1}(\alpha))-e_{1}(\alpha)(\beta_{2}+\beta_{3})+\left(2\kappa-\frac{45}{2}\alpha^{2}\right)\alpha=0. (2.9) We shall now employ the Gauss equation: \displaystyle\langle\tilde{R}(X,Y)Z,W\rangle \displaystyle=\langle R(X,Y)Z,W\rangle+\langle B(X,Z),B(Y,W)\rangle \displaystyle\quad-\langle B(X,W),B(Y,Z)\rangle, (2.10) where \tilde{R} denotes the curvature tensor of N^{4}(c) . From the Gauss equation ( 2 ), we obtain: • By setting X=W=e_{1} and Y=Z=e_{i} ( i=2,3 ): \displaystyle e_{1}(\beta_{2}) \displaystyle=\beta_{2}^{2}+c-\tfrac{3}{2}\alpha k_{2}, (2.11) \displaystyle e_{1}(\beta_{3}) \displaystyle=\beta_{3}^{2}+c-\tfrac{3}{2}\alpha k_{3}. • By setting X=W=e_{2} and Y=Z=e_{3} : \kappa+c=e_{2}(\rho_{3})-e_{3}(\rho_{2})-\beta_{2}\beta_{3}-\rho_{2}^{2}-\rho_{3}^{2}, (2.12) where \kappa=k_{2}k_{3} is the intrinsic Gauss curvature of the e_{2}\wedge e_{3} plane. Following a similar argument as in [ 3 ] , one can show that \rho_{2} and \rho_{3} must vanish identically. Consequently, ( 2.12 ) reduces to \kappa+c=-\beta_{2}\beta_{3}. (2.13) By rewriting equations ( 2.11 ), we obtain \begin{cases}e_{1}(e_{1}(k_{2}))=\frac{21}{2}\,\beta_{2}\,e_{1}(\alpha)+2(\kappa+c)\left(k_{3}+\tfrac{3}{2}\alpha\right)+\left(c-\tfrac{3}{2}\alpha k_{2}\right)\left(k_{2}+\tfrac{3}{2}\alpha\right),\\ e_{1}(e_{1}(k_{3}))=\frac{21}{2}\,\beta_{3}\,e_{1}(\alpha)+2(\kappa+c)\left(k_{2}+\tfrac{3}{2}\alpha\right)+\left(c-\tfrac{3}{2}\alpha k_{3}\right)\left(k_{3}+\tfrac{3}{2}\alpha\right).\end{cases} Summing these two equations yields e_{1}(e_{1}(\alpha))=\tfrac{7}{3}e_{1}(\alpha)(\beta_{2}+\beta_{3})+\alpha(4\kappa+5c-9\alpha^{2}). (2.14) Henceforth, we define the following coefficients: \displaystyle A_{1} \displaystyle=-\tfrac{9}{2}\kappa-\tfrac{15}{4}c+\tfrac{189}{8}\alpha^{2}, \displaystyle A_{2} \displaystyle=-\tfrac{13}{2}\kappa-\tfrac{15}{4}c+\tfrac{369}{8}\alpha^{2}, \displaystyle A_{3} \displaystyle=\kappa+9\alpha^{2}, \displaystyle A_{4} \displaystyle=\tfrac{13}{2}\kappa+\tfrac{31}{4}c-108\alpha^{2}, \displaystyle A_{5} \displaystyle=\tfrac{13}{2}\kappa+\tfrac{15}{2}c-\tfrac{441}{4}\alpha^{2}. By combining ( 2.9 ) and ( 2.14 ), we obtain e_{1}(\alpha)(\beta_{2}+\beta_{3})=\alpha A_{1}. (2.15) Substituting ( 2.15 ) into ( 2.14 ) yields e_{1}(e_{1}(\alpha))=\alpha A_{2}. (2.16) In order to derive another relation between \alpha and \kappa , we first use ( 2.2 ) to obtain \displaystyle\beta_{2}k_{3}+\beta_{3}k_{2} \displaystyle=(\beta_{2}+\beta_{3})(k_{2}+k_{3})-(\beta_{2}k_{2}+\beta_{3}k_{3}) \displaystyle=\tfrac{9}{2}\alpha(\beta_{2}+\beta_{3})-(\beta_{2}k_{2}+\beta_{3}k_{3}). (2.17) Differentiating ( 2.2 ) along e_{1} and employing ( 2.6 ), we find \displaystyle\tfrac{9}{2}e_{1}(\alpha) \displaystyle=e_{1}(k_{2})+e_{1}(k_{3}) \displaystyle=\beta_{2}k_{2}+\beta_{3}k_{3}+\tfrac{3}{2}\alpha(\beta_{2}+\beta_{3}). Thus, we have \beta_{2}k_{2}+\beta_{3}k_{3}=\tfrac{9}{2}e_{1}(\alpha)-\tfrac{3}{2}\alpha(\beta_{2}+\beta_{3}). (2.18) Now, substituting ( 2.18 ) into ( 2 ), we find \beta_{2}k_{3}+\beta_{3}k_{2}=6\alpha(\beta_{2}+\beta_{3})-\tfrac{9}{2}e_{1}(\alpha). (2.19) From ( 2.11 ) and ( 2.13 ), we yield \displaystyle e_{1}(\kappa) \displaystyle=-e_{1}(\beta_{2}\beta_{3}) \displaystyle=-(\beta_{2}\beta_{3}+c)(\beta_{2}+\beta_{3})+\tfrac{3}{2}\alpha(\beta_{2}k_{3}+\beta_{3}k_{2}) \displaystyle=\kappa(\beta_{2}+\beta_{3})+\tfrac{3}{2}\alpha(\beta_{2}k_{3}+\beta_{3}k_{2}). (2.20) Substituting ( 2.19 ) into ( 2.20 ), we obtain e_{1}(\kappa)=A_{3}(\beta_{2}+\beta_{3})-\tfrac{27}{4}\alpha e_{1}(\alpha). (2.21) By differentiating ( 2.15 ) along e_{1} and employing ( 2.11 ), ( 2.15 ), ( 2.16 ), and ( 2.21 ), we get e_{1}(\alpha)A_{4}=\alpha(\beta_{2}+\beta_{3})A_{5}. (2.22) By multiplying ( 2.22 ) first by e_{1}(\alpha) and then by \beta_{2}+\beta_{3} , and subsequently employing ( 2.15 ), we obtain: \begin{cases}(e_{1}(\alpha))^{2}A_{4}=\alpha^{2}A_{1}A_{5},\\ A_{4}A_{1}=(\beta_{2}+\beta_{3})^{2}A_{5}.\end{cases} (2.23) Differentiating ( 2.22 ) along e_{1} and using ( 2.11 ), ( 2.15 ), ( 2.16 ), ( 2.21 ), and ( 2.23 ), we arrive at the following key polynomial relation: \displaystyle 0 \displaystyle=140608\,\kappa^{4}+\left(448864\,c-6157008\,\alpha^{2}\right)\kappa^{3} \displaystyle\quad+\left(375856\,c^{2}-11450088\,\alpha^{2}c+59355504\,\alpha^{4}\right)\kappa^{2} \displaystyle\quad+\left(-45240\,c^{3}-208800\,\alpha^{2}c^{2}-36099270\,\alpha^{4}c+485815806\,\alpha^{6}\right)\kappa \displaystyle\quad-111600\,c^{4}+5241780\,\alpha^{2}c^{3}-132575616\,\alpha^{4}c^{2} \displaystyle\quad+1642080519\,\alpha^{6}c-6863560515\,\alpha^{8}. (2.24) Let \gamma=\gamma(t) , t\in I , be an integral curve of e_{1} passing through p=\gamma(t_{0}) . Since e_{2}(\alpha)=e_{3}(\alpha)=0 and e_{2}(\kappa)=e_{3}(\kappa)=0 , and given that e_{1}(\alpha)\neq 0 , we may locally view t as a function of \alpha in a neighborhood of \alpha_{0}=\alpha(t_{0}) . Consequently, \kappa can be considered a function of \alpha , i.e., \kappa=\kappa(\alpha) . If A_{5}=0 or A_{4}=0 , then ( 2.24 ) implies that \alpha satisfies a polynomial equation of degree eight with constant coefficients. This forces \alpha to be constant, yielding a contradiction. Thus, assuming A_{4},A_{5}\neq 0 , ( 2.23 ) yields: \begin{cases}\left(\frac{d\alpha}{dt}\right)^{2}=\frac{\alpha^{2}A_{1}A_{5}}{A_{4}},\\ (\beta_{2}+\beta_{3})^{2}=\frac{A_{4}A_{1}}{A_{5}}.\end{cases} (2.25) By using ( 2.15 ), ( 2.21 ), and ( 2.25 ), the derivative of \kappa with respect to \alpha is computed as follows: \displaystyle\frac{d\kappa}{d\alpha} \displaystyle=\frac{d\kappa}{dt}\left(\frac{d\alpha}{dt}\right)^{-1} \displaystyle=\frac{A_{3}\frac{d\alpha}{dt}(\beta_{2}+\beta_{3})}{\left(\frac{d\alpha}{dt}\right)^{2}}-\frac{27}{4}\alpha \displaystyle=\frac{A_{3}A_{4}}{\alpha A_{5}}-\frac{27}{4}\alpha. (2.26) Next, differentiating the polynomial relation ( 2.24 ) with respect to \alpha and substituting the expression for \frac{d\kappa}{d\alpha} from ( 2 ), we obtain a new polynomial equation in \alpha and \kappa , which is of degree five in \kappa . By eliminating the \kappa^{5} term between this new equation and ( 2.24 ), we derive a polynomial in \alpha and \kappa of degree four in \kappa . Continuing this process to successively eliminate the highest powers of \kappa , we eventually arrive at a polynomial equation in \alpha with constant coefficients. Consequently, \alpha must be constant, yielding a contradiction to our initial assumption. Therefore, the hypersurface M is minimal. ∎ Conclusion In this work, we investigated the geometric properties of H -tensional hypersurfaces M in a 4 -dimensional space form N^{4}(c) . By analyzing the characterization ( 2.1 ) and the resulting Codazzi and Gauss equations, we established a rigorous link between the mean curvature \alpha and the principal curvatures of the manifold. The derivation of the key polynomial relation ( 2.24 ) and the subsequent elimination process showed that a non-constant mean curvature leads to an algebraic contradiction. Consequently, we have proved that such hypersurfaces must have constant mean curvature. Given the H -tensional condition, this constant must vanish, leading to the final classification: every H -tensional hypersurface in N^{4}(c) is minimal.