betaFeedback

WorksMathematics

Geometric Rigidity of Manifolds Admitting mm-Modified Conformal Vector Fields

Abstract

We establish certain integral inequalities for mm-modified conformal vector fields on compact Riemannian manifolds and derive corresponding triviality and rigidity results. We further obtain integrability conditions for such vector fields and establish an extended Bourguignon-Ezin identity. We then prove that a compact Riemannian manifold admitting a non-trivial closed mm-modified conformal vector field is isometric to a round sphere under either a constant scalar curvature and positive Ricci curvature assumption or a suitable bound on the Ricci curvature involving the first nonzero eigenvalue of the Laplace operator. As a consequence, we obtain an extension of a classical sphere characterization result of Yano and Nagano to the corresponding result for a non trivial mm-modified conformal vector field assumed closed. Finally, we consider a complete Kähler manifold admitting a closed mm-modified conformal vector field VV, and show that VV is mm-modified Killing and around a non-zero point of VV the manifold splits as the product of the complex Euclidean plane C\mathbb{C} and a complete Kähler manifold.

2020 Mathematics Subject Classifications: 53C21, 53C25

Introduction

A Riemannian manifold (M,g)(M,g) is said to be a Ricci soliton if it satisfies the equation

12LVg+Ric⁡=λg,(1)\frac{1}{2}\mathcal{L}_{V}g+\operatorname{Ric}=\lambda g, \tag*{(1)}

where L\mathcal{L} denotes the Lie-derivative operator, VV is a smooth vector field on MM, gg is the Riemannian metric, RicRic is the Ricci tensor of gg, and λ\lambda is a real constant. If VV is the gradient of a smooth function ff, i.e., V=∇fV = \nabla f, then (M,g,f,λ)(M,g,f,\lambda) is called a gradient Ricci soliton. As a natural generalization of Einstein metrics and gradient Ricci solitons, Catino [4] introduced the notion of a generalized quasi-Einstein manifold. More precisely, an nn-dimensional Riemannian manifold (M,g)(M,g) is said to be generalized quasi-Einstein if it satisfies

Ric+Hess f−μ df⊗df=λg,(2)Ric + Hess\,f - \mu\,df \otimes df = \lambda g, \tag*{(2)}

where μ\mu is a constant and λ\lambda is a smooth function on MM. Catino [4] showed that a complete generalized quasi-Einstein manifold with harmonic Weyl tensor and zero radial Weyl curvature is locally a warped product with an (n−1)(n-1)-dimensional Einstein fiber. In particular, taking

μ=1m,0<m≤∞,\mu= \frac{1}{m}, \qquad0 < m \leq\infty,

equation (2) becomes

Ric+Hess f−1mdf⊗df=λg.(3)Ric + Hess\,f - \frac{1}{m}df \otimes df = \lambda g. \tag*{(3)}

The tensor

Ricfm=Ric+Hess f−1mdf⊗dfRic^{m}_{f} = Ric + Hess\,f - \frac{1}{m}df \otimes df

is referred to as the mm-Bakry-Émery Ricci tensor. When λ\lambda is constant, equation (3) defines an mm-quasi-Einstein manifold; see Case, Shu and Wei [3] and He, Petersen and Wylie [8]. A vector-field analogue of (3) was subsequently considered by Limoncu [9] and Barros and Ribeiro, Jr. [1]. More precisely, if vv denotes the 1-form metrically dual to the vector field VV, they considered

12LVg+Ric−1mv⊗v=λg,(4)\frac{1}{2}\mathcal{L}_{V}g + Ric - \frac{1}{m}v \otimes v = \lambda g, \tag*{(4)}

where λ\lambda is a constant. Moreover, in the limiting case m=∞m = \infty, equation (4) reduces to the classical Ricci soliton equation (1). For a Riemannian manifold (M,g)(M,g), it is interesting to know what happens when gg is both Einstein and mm-quasi Einstein. In this case, there exists a vector field VV on MM such that

12LVg−1mv⊗v=cg,\frac{1}{2}\mathcal{L}_{V}g - \frac{1}{m}v \otimes v = cg,

where cc is a constant. Zhang and Chen [15] generalized the above setting by replacing the constant cc with a smooth function σ∈C∞(M)\sigma\in C^{\infty}(M), thereby introducing the following definition.

Definition: A vector field VV on a Riemannian manifold (M,g)(M,g) is called an mm-modified conformal vector field if

12LVg−1mv⊗v=σg(5)\frac{1}{2}\mathcal{L}_{V}g-\frac{1}{m}v\otimes v=\sigma g \tag*{(5)}

where mm is a positive constant and σ∈C∞(M)\sigma\in C^{\infty}(M).

When σ\sigma is constant and 0<m<∞0<m<\infty, a vector field VV satisfying (1.5) is called an mm-modified homothetic vector field; in the special case σ=0\sigma=0, it is called an mm-modified Killing vector field. VV is said to be trivial when V=0V=0. In the limiting case m=∞m=\infty, equation (1.5) reduces to: LVg=2σg\mathcal{L}_{V}g=2\sigma g, and hence recovers the usual notion of a conformal vector field. Zhang and Chen [15] established several rigidity results for mm-modified conformal vector fields. In particular, they proved that a connected compact Riemannian manifold of dimension at least 2 admits no mm-modified conformal vector field of constant length. They further showed in Riemannian case that, for Lie groups admitting left-invariant mm-modified conformal vector fields, the metric is Einstein and non-trivial mm-quasi-Einstein. Moreover, such Lie groups are solvable and isometric to a real hyperbolic space. In the compact setting, Poddar and Sharma [10] investigated mm-modified conformal vector fields and established several rigidity results. In particular, they proved that:

  1. a compact Riemannian manifold admits no non-trivial mm-modified homothetic vector field; and

  1. if VV is an mm-modified conformal vector field, with 0<m<∞0<m<\infty, on a compact Riemannian manifold satisfying

Ric⁡(V,V)≤2(1−n)nm2(max⁡M∣V∣)4,\operatorname{Ric}(V,V)\leq\frac{2(1-n)}{nm^{2}}\left(\max_{M}|V|\right)^{4},

then VV is trivial.

More recently, Deshmukh et al. [6] obtained further triviality results for mm-modified conformal vector fields in both compact and non-compact settings. They also constructed an example of an mm-modified conformal vector field on a round sphere.

m-Modified Conformal Vector Fields On Compact Riemannian Manifolds

First, we establish the following generalization of the result (i) of [10] stated earlier.

Theorem 2.1 If VV is an mm-modified conformal vector field on a compact (without boundary) Riemannian manifold (M,g)(M,g), then ∫M(LVσ) dvg≤0\int_M(\mathcal{L}_V \sigma)\,dv_g \le0, equality holds when VV is Killing for m=∞m=\infty, and VV vanishes for m≠∞m\ne\infty.

Remark 2.1 For σ\sigma constant, this theorem reduces to the result (i) of [10].

Proof. First, tracing equation (5) gives

div⁡V=1m∣V∣2+nσ.(6)\operatorname{div}V=\frac{1}{m}|V|^2+n\sigma. \tag*{(6)}

Equation (5) can also be exhibited as

g(∇XV,Y)+g(∇YV,X)=2m2g(V,X)g(V,Y)+2σg(X,Y).(7)g(\nabla_XV,Y)+g(\nabla_YV,X)=\frac{2}{m^2}g(V,X)g(V,Y)+2\sigma g(X,Y). \tag*{(7)}

The exterior derivative dvdv of the 1-form vv is given by

g(∇XV,Y)−g(∇YV,X)=2(dv)(X,Y).(8)g(\nabla_XV,Y)-g(\nabla_YV,X)=2(dv)(X,Y). \tag*{(8)}

As dvdv is skew-symmetric, we define a tensor field of type (1,1)(1,1) given by

dv(X,Y)=g(FX,Y).(9)dv(X,Y)=g(FX,Y). \tag*{(9)}

Evidently, FF is skew self-adjoint, i.e., g(FX,Y)=−g(X,FY)g(FX,Y)=-g(X,FY). Thus, (9) assumes the form

g(∇XV,Y)−g(∇YV,X)=2g(FX,Y).(10)g(\nabla_XV,Y)-g(\nabla_YV,X)=2g(FX,Y). \tag*{(10)}

Adding equations (2.2 and (10) side by side, and factoring out YY gives

∇XV=1mg(V,X)V+σX+FX.(11)\nabla_XV=\frac{1}{m}g(V,X)V+\sigma X+FX. \tag*{(11)}

Now, we compute the covariant derivative of the squared gg-norm of VV using (11) as follows

∇X∣V∣2=2g(∇XV,V)=2mg(V,X)∣V∣2+2σg(X,V)+2g(FX,V).(12)\nabla_X|V|^2=2g(\nabla_XV,V)=\frac{2}{m}g(V,X)|V|^2+2\sigma g(X,V)+2g(FX,V). \tag*{(12)}

Using this we compute

Δ∣V∣2=2mdiv⁡(∣V∣2V)+2(LVσ+σdiv⁡V)−2div⁡(FV).(13)\Delta|V|^2=\frac{2}{m}\operatorname{div}(|V|^2V)+2(\mathcal{L}_V\sigma+\sigma\operatorname{div}V)-2\operatorname{div}(FV). \tag*{(13)}

Next, multiplying both sides of (6) with div⁡V\operatorname{div}V, gives (div⁡V)2=nσdiv⁡V+1m∣V∣2div⁡V(\operatorname{div}V)^2=n\sigma\operatorname{div}V+\frac{1}{m}|V|^2\operatorname{div}V. Integrating it over the compact MM, using divergence theorem, and the identity: div⁡(∣V∣2V)=∇V∣V∣2+∣V∣2div⁡V\operatorname{div}(|V|^2V)=\nabla_V|V|^2+|V|^2\operatorname{div}V, we obtain

∫M(div⁡V)2dvg=∫M[nσdiv⁡V−1m(∇V∣V∣2)]dvg(14)\int_M(\operatorname{div}V)^2dv_g=\int_M\left[n\sigma\operatorname{div}V-\frac{1}{m}(\nabla_V|V|^2)\right]dv_g \tag*{(14)}

where dvgdv_g is the gg-volume elemment of (M,g)(M,g). Substituting VV for XX in (12) we have

∇V∣V∣2=2m∣V∣4+2σ∣V∣2.(15)\nabla_V|V|^2=\frac{2}{m}|V|^4+2\sigma|V|^2. \tag*{(15)}

Using this equation, along with the identity: div⁡(∣V∣2V)=∇V∣V∣2+∣V∣2div⁡V\operatorname{div}(|V|^2V)=\nabla_V|V|^2+|V|^2\operatorname{div}V, in equation (13) provides

12Δ∣V∣2=1m[2m∣V∣4+2σ∣V∣2+∣V∣2div⁡V]+LVσ+σdiv⁡V−div⁡(FV).(16)\frac{1}{2}\Delta|V|^2=\frac{1}{m}\left[\frac{2}{m}|V|^4+2\sigma|V|^2+|V|^2\operatorname{div}V\right]+\mathcal{L}_V\sigma+\sigma\operatorname{div}V-\operatorname{div}(FV). \tag*{(16)}

Eliminating ∣V∣2|V|^2 between (6) and (16) shows that

12Δ∣V∣2=2m∣V∣2(div⁡V−nσ)+(2σ+div⁡V)(div⁡V−nσ)+LVσ+σdiv⁡V−div⁡(FV).\frac{1}{2}\Delta|V|^2=\frac{2}{m}|V|^2(\operatorname{div}V-n\sigma)+(2\sigma+\operatorname{div}V)(\operatorname{div}V-n\sigma)+\mathcal{L}_V\sigma+\sigma\operatorname{div}V-\operatorname{div}(FV).

Now using (2.1) in the above, and simplifying we find that

12Δ∣V∣2=3(div⁡V)2+(σdiv⁡V)(3−5n)+2n(n−1)σ2+LVσ−div⁡(FV).\frac{1}{2}\Delta|V|^2 = 3(\operatorname{div}V)^2 + (\sigma\operatorname{div}V)(3-5n) + 2n(n-1)\sigma^2 + \mathcal{L}_V\sigma- \operatorname{div}(FV).

Integrating it over compact MM provides

∫M[3(div⁡V)2+(3−5n)(σdiv⁡V)+2n(n−1)σ2+LVσ] dvg=0.(17)\int_M [3(\operatorname{div}V)^2 + (3-5n)(\sigma\operatorname{div}V) + 2n(n-1)\sigma^2 + \mathcal{L}_V\sigma]\,dv_g = 0. \tag*{(17)}

Going back to equation (14), using (2.10) and (2.1) we have that

∫M[3(div⁡V)2+(2−5n)(σdiv⁡V)−2n(1−n)σ2] dvg=0.(18)\int_M [3(\operatorname{div}V)^2 + (2-5n)(\sigma\operatorname{div}V) - 2n(1-n)\sigma^2]\,dv_g = 0. \tag*{(18)}

At this point, multiplying the above equation with 3−5n3-5n, and (2.12) with 2−5n2-5n, and subsequently subtracting the resulting equations we obtain

∫M[3(div⁡V)2+2n(n−1)σ2] dvg=(2−5n)∫M(LVσ) dvg.(19)\int_M [3(\operatorname{div}V)^2 + 2n(n-1)\sigma^2]\,dv_g = (2-5n)\int_M (\mathcal{L}_V\sigma)\,dv_g. \tag*{(19)}

The assertions of the theorem follow from equation (2.14). This completes the proof.

We turn our attention back to the result (ii) of [10] that provides a generalization of the following result [12] of Yano “If a compact Riemannian manifold (M,g)(M,g) admits a conformal vector field VV satisfying Ric⁡(V,V)≤0\operatorname{Ric}(V,V) \le0, then VV is parallel, and if the Ricci curvature is negative definite, then the conformal vector field VV is zero”. Improving this result further and extending the aforementioned result of Yano to the case of an mm-modified vector field, we prove the following result.

Theorem 2.2 Let (M,g)(M,g) be a compact Riemannian manifold admitting an mm-modified conformal vector field VV. If Ric⁡(V,V)≤0\operatorname{Ric}(V,V) \le0, then VV is parallel if m=∞m=\infty, and VV vanishes identically if m≠∞m\ne\infty.

Proof. Recalling (2.6) and using it we compute

Riem⁡(Y,X)V=(Yσ)X−(Xσ)Y+1m[g(∇YV,X)V−g(∇XV,Y)V+v(X)Y−v(Y)X+v(X)FY−v(Y)FX]+(∇YF)X−(∇XF)Y,(20)\begin{aligned} \operatorname{Riem}(Y,X)V &= (Y\sigma)X - (X\sigma)Y + \frac{1}{m}[g(\nabla_YV,X)V - g(\nabla_XV,Y)V \\ &\quad+ v(X)Y - v(Y)X + v(X)FY \\ &\quad- v(Y)FX] + (\nabla_YF)X - (\nabla_XF)Y, \tag*{(20)} \end{aligned}

where RiemRiem denotes the curvature tensor of type (1,3)(1,3) for gg, defined by: Riem(X,Y)Z=∇X∇YZ−∇Y∇XZ−∇[X,Y]ZRiem(X,Y)Z = \nabla_X\nabla_Y Z-\nabla_Y\nabla_X Z-\nabla_{[X,Y]}Z for arbitrary X,Y,Z∈X(M)X,Y,Z \in\mathfrak{X}(M). Contracting equation (20) at YY we have

Ric(X,V)=(1−n)Xσ+1m[g(∇VV,X)−g(∇XV,V)]+(n−1)σv(X)−g(FX,V)+(divF)X.Ric(X,V)=(1-n)X\sigma+\frac{1}{m}\left[g(\nabla_VV,X)-g(\nabla_XV,V)\right]+(n-1)\sigma v(X)-g(FX,V)+(divF)X.

Next, substituting VV for XX in the above equation gives

Ric(V,V)=(1−n)Vσ+1m(n−1)σ∣V∣2+(divF)V(21)Ric(V,V)=(1-n)V\sigma+\frac{1}{m}(n-1)\sigma|V|^2+(divF)V \tag*{(21)}

Through the computation:

∇X(FV)=(∇XF)V+F∇XV=(∇XF)V+σFX+1mv(X)FV+F2X\begin{aligned} \nabla_X(FV)&=(\nabla_XF)V+F\nabla_XV\\ &=(\nabla_XF)V+\sigma FX+\frac{1}{m}v(X)FV+F^2X \end{aligned}

and its trace at XX we obtain

div(FV)=(divF)V−∣F∣2,div(FV)=(divF)V-|F|^2,

where we have used (2.6) and the skew self adjointness of FF. Integrating the preceding equation over the compact MM, we have

∫M(divF)V dvg=−∫M∣F∣2dvg(22)\int_M(divF)V\,dv_g=-\int_M|F|^2dv_g \tag*{(22)}

Integrating equation (21) over the compact MM and making use of equations (22) and (2.14) we obtain the formula

∫MRic(V,V)dvg=n−15n−2∫M[3(divV)2+2n(n−1)σ2]dvg+n−1m∫M∣V∣2dvg+∫M∣F∣2dvg.(23)\begin{aligned} \int_M Ric(V,V)dv_g&=\frac{n-1}{5n-2}\int_M\left[3(divV)^2+2n(n-1)\sigma^2\right]dv_g\\ &\quad+\frac{n-1}{m}\int_M|V|^2dv_g+\int_M|F|^2dv_g. \tag*{(23)} \end{aligned}

The conclusions of the theorem follow immediately from this consequence, completing the proof.

Integrability Conditions for mm-Modified Conformal Vector Field

In this section, we derive integrability conditions for an mm-modified conformal vector field VV on a Riemannian manifold (M,g)(M,g). By definition, the equation (5) is written in local co-ordinates as

LVgij=2σgij+2mViVj.(24)\mathcal{L}_V g_{ij}=2\sigma g_{ij}+\frac{2}{m}V_iV_j. \tag*{(24)}

Lie-differentiating the relation: gijgjk=δikg_{ij}g^{jk}=\delta_i^k along VV gives

(LVgij)gjk+gijLVgjk=0.(\mathcal{L}_V g_{ij})g^{jk}+g_{ij}\mathcal{L}_V g^{jk}=0.

Using (24) in it, and taking inner product with gilg^{il} yields that

LVgkl=−2σgkl−2mVkVl.\mathcal{L}_V g^{kl}=-2\sigma g^{kl}-\frac{2}{m}V^kV^l.

At this point we denote ViVjV_iV_j by pijp_{ij} and hence Tr.p=∣V∣2Tr.p=|V|^2. Then we have

LVgij=2σgij+2mpij,LVgij=−2σgij−2mpij.(25)\mathcal{L}_V g_{ij}=2\sigma g_{ij}+\frac{2}{m}p_{ij},\qquad\mathcal{L}_V g^{ij}=-2\sigma g^{ij}-\frac{2}{m}p^{ij}. \tag*{(25)}

Use of the formula (Yano [13]):

LVΓjki=12gil[∇jLVgkl+∇kLVgjl−∇lLVgjk]\mathcal{L}_V\Gamma^i_{jk}=\frac{1}{2}g^{il}\left[\nabla_j\mathcal{L}_V g_{kl}+\nabla_k\mathcal{L}_V g_{jl}-\nabla_l\mathcal{L}_V g_{jk}\right]

in (25) provides

LVΓjki=(∇jσ)δki+(∇kσ)δji−(∇iσ)gjk+1m[∇jpki+∇kpji−∇ipjk](26)\mathcal{L}_V\Gamma^i_{jk}=(\nabla_j\sigma)\delta^i_k+(\nabla_k\sigma)\delta^i_j-(\nabla^i\sigma)g_{jk} +\frac{1}{m}\left[\nabla_jp^i_k+\nabla_kp^i_j-\nabla^ip_{jk}\right] \tag*{(26)}

where Γjki\Gamma^i_{jk} denote the connection coefficients of gijg_{ij}. Now combined use of the formula [13]: LVRiklj=∇iLVΓklj−∇kLVΓilj\mathcal{L}_V R_{ikl}{}^j=\nabla_i\mathcal{L}_V\Gamma^j_{kl}-\nabla_k\mathcal{L}_V\Gamma^j_{il} and equation (26), gives

LVRiklj=(∇i∇lσ)δkj−(∇k∇lσ)δij−(∇i∇jσ)gkl+(∇k∇jσ)gil+1m[∇i∇kplj−∇k∇iplj+∇i∇lpkj−∇k∇lpij−∇i∇jpkl+∇k∇jpil](27)\begin{aligned} \mathcal{L}_V R_{ikl}{}^j={}&(\nabla_i\nabla_l\sigma)\delta^j_k-(\nabla_k\nabla_l\sigma)\delta^j_i-(\nabla_i\nabla^j\sigma)g_{kl}+(\nabla_k\nabla^j\sigma)g_{il}\\ &+\frac{1}{m}\left[\nabla_i\nabla_kp^j_l-\nabla_k\nabla_ip^j_l+\nabla_i\nabla_lp^j_k-\nabla_k\nabla_lp^j_i-\nabla_i\nabla^jp_{kl}+\nabla_k\nabla^jp_{il}\right] \tag*{(27)} \end{aligned}

where RikljXl=∇i∇kXj−∇k∇iXjR_{ikl}{}^{j}X^{l}=\nabla_i\nabla_kX^{j}-\nabla_k\nabla_iX^{j} and XlX^{l} are components of an arbitrary vector field X∈X(M)X\in\mathfrak{X}(M). Contracting (3.4) with respect to ii and jj we get

LVRkl=(2−n)∇k∇lσ−(Δσ)gkl+1m[∇i∇kpli+∇i∇lpki−∇k∇l(Tr⁡.p)](28)\mathcal{L}_V R_{kl}=(2-n)\nabla_k\nabla_l\sigma-(\Delta\sigma)g_{kl} +\frac{1}{m}\left[\nabla_i\nabla_k p_l{}^i+\nabla_i\nabla_l p_k{}^i-\nabla_k\nabla_l(\operatorname{Tr}.p)\right] \tag*{(28)}

where RklR_{kl} are the components of the Ricci tensor of gijg_{ij}. Finally, we compute LVR=LV(Rklgkl)\mathcal{L}_V R=\mathcal{L}_V(R_{kl}g^{kl}) using (3.2) and (3.5) and thus obtain

LVR=2(1−n)Δσ−2σR+2m[∇i∇lpli−Δ(Tr⁡.p)−Rklpkl].(29)\mathcal{L}_V R=2(1-n)\Delta\sigma-2\sigma R+\frac{2}{m}\left[\nabla_i\nabla^l p_l{}^i-\Delta(\operatorname{Tr}.p)-R_{kl}p^{kl}\right]. \tag*{(29)}

Equations (3.3), (3.4), (3.5), (3.6) provide the evolutions of the connection coefficients, curvature tensor, Ricci tensor and scalar curvature along an mm-modified conformal vector field VV.

Extended Bourguignon-Ezin Identity and Scalar Curvature Constraints

Here we assume (M,g)(M,g) compact and admitting an mm-modified vector field VV. Integrating (3.6) over the compact MM we find that

∫M(LVR) dvg=−2∫M[σR+1mRic⁡(V,V)]dvg.(30)\int_M(\mathcal{L}_V R)\,dv_g=-2\int_M\left[\sigma R+\frac{1}{m}\operatorname{Ric}(V,V)\right]dv_g. \tag*{(30)}

Using the equation (2.1) and the identity: div⁡(RV)=LVR+Rdiv⁡(V)\operatorname{div}(RV)=\mathcal{L}_V R+R\operatorname{div}(V) in (4.1) we obtain

∫M(LVR) dvg=−2nm(n−2)∫M[(Ric⁡−Rng)(V,V)]dvg.(31)\int_M(\mathcal{L}_V R)\,dv_g=-\frac{2n}{m(n-2)}\int_M\left[\left(\operatorname{Ric}-\frac{R}{n}g\right)(V,V)\right]dv_g. \tag*{(31)}

We note that this reduces to the Bourguignon-Ezin identity [2] for a conformal vector field (for which m=∞m=\infty) on a compact Riemannian manifold. We state the above consequence as the following result.

Theorem 4.1 For an mm-modified conformal vector field VV on a compact Riemannian manifold of dimension n>2n>2, we have the extended Bourguignon-Ezin identity given by (4.2).

Remark 4.1 (31) is trivially satisfied by a compact Einstein manifold.

We close this section with the scalar curvature constraint. Using equation (23) in (31) and simplifying gives the following constraint on the scalar curvature RR:

−m(n−2)2n∫M(LVR) dvg=n−15n−2∫M[3(div⁡V)2+2n(n−1)σ2] dvg+∫M∣F∣2 dvg+∫Mn−1m∣V∣2 dvg−1n∫MR∣V∣2 dvg.(32)-\frac{m(n-2)}{2n}\int_M (\mathcal{L}_V R)\,dv_g = \frac{n-1}{5n-2}\int_M [3(\operatorname{div}V)^2+2n(n-1)\sigma^2]\,dv_g +\int_M |F|^2\,dv_g+\int_M \frac{n-1}{m}|V|^2\,dv_g-\frac{1}{n}\int_M R|V|^2\,dv_g. \tag*{(32)}

It follows immediately from the above equation that

∫M(LVR) dvg≤2m(n−2)∫M(R−n(n−1)m)∣V∣2.(33)\int_M (\mathcal{L}_V R)\,dv_g\leq\frac{2}{m(n-2)}\int_M\left(R-\frac{n(n-1)}{m}\right)|V|^2. \tag*{(33)}

In particular, if RR is constant, and VV is non-zero and non-conformal, then the in-equality (33) shows that R≥n(n−1)mR\geq\frac{n(n-1)}{m}, because VV is not conformal, i.e., m≠∞m\ne\infty. However, if R=n(n−1)mR=\frac{n(n-1)}{m}, then (32) implies that div⁡V=0\operatorname{div}V=0, σ=0\sigma=0, and hence from (6) we get V=0V=0, a contradiction. Consequently, R>n(n−1)mR>\frac{n(n-1)}{m}. Hence we have proved the following result.

Proposition 4.1 If a compact Riemannian manifold of dimension n>2n>2 has constant scalar curvature RR, and admits a non-zero non-conformal mm-modified conformal vector field, then RR is bounded below by the positive constant n(n−1)m\frac{n(n-1)}{m}.

Closed mm-modified Conformal Vector Fields On Compact Riemannian Manifolds

We begin this section by recalling the following rigidity theorems for conformal vector fields on compact Riemannian manifolds.

Theorem 5.1 (Yano [13]) Let (M,g)(M,g) be an (n≥3)(n\geq3)-dimensional closed Riemannian manifold with constant scalar curvature admitting a nontrivial gradient conformal vector field, then MM is isometric to a round sphere.

Theorem 5.2 (Deshmukh and Al-Solamy [5]) Let (M,g)(M,g) be an nn-dimensional compact Riemannian manifold whose Ricci curvature satisfies

0<Ric⁡≤(n−1)(2−ncλ1)c,0<\operatorname{Ric}\leq(n-1)\left(2-\frac{nc}{\lambda_1}\right)c,

where cc is a constant and λ1\lambda_1 denotes the first nonzero eigenvalue of the Laplace operator. If MM admits a nonzero conformal gradient vector field, then MM is isometric to the sphere Sn(c)\mathbb{S}^n(c).

Motivated by the aforementioned results, we now establish the corresponding rigidity result (a generalization of Theorem 5.2) for closed mm-modified conformal vector fields.

Theorem 5.3 Let (Mn,g)(M^n,g) be a compact Riemannian manifold with positive Ricci curvature admitting a non-trivial closed mm-modified conformal vector field. Then (M,g)(M,g) is isometric to a round sphere if one of the following conditions is satisfied:

(i) MM has constant scalar curvature and n≥3n \ge3.

(ii) The Ricci curvature satisfies

Ric⁡≤(n−1)(2−ncλ1)c,\operatorname{Ric} \le(n-1)\left(2-\frac{nc}{\lambda_1}\right)c,

where cc is a constant and λ1\lambda_1 denotes the first nonzero eigenvalue of the Laplace operator.

Proof. Let VV be a non-trivial closed mm-modified conformal vector field on MM, and let v=V♭v=V^\flat denote its metrically dual 1-form. Since MM is compact, the Hodge decomposition gives

v=dh+η,(34)v=dh+\eta, \tag*{(34)}

where h∈C∞(M)h\in C^\infty(M) and η\eta is a 1-form such that δη=0\delta\eta=0. We are going to show that the harmonic component in (34) vanishes under our hypothesis. Since VV is closed, its dual 1-form vv is closed, and hence

0=dv=dη.0=dv=d\eta.

On the other hand, the Hodge decomposition gives

δη=0.\delta\eta=0.

Thus

ΔHη=(dδ+δd)η=0,\Delta_H\eta=(d\delta+\delta d)\eta=0,

so that η\eta is harmonic. The Bochner formula for a harmonic 1-form yields

12Δ∣η∣2=∣∇η∣2+Ric⁡(η♯,η♯),(35)\frac{1}{2}\Delta|\eta|^2=|\nabla\eta|^2+\operatorname{Ric}(\eta^\sharp,\eta^\sharp), \tag*{(35)}

where η♯\eta^\sharp denotes the vector field metrically dual to η\eta. Integrating (35) over the compact manifold MM and applying the divergence theorem, we obtain

0=∫M(∣∇η∣2+Ric⁡(η♯,η♯)) dμ.(36)0=\int_M\left(|\nabla\eta|^2+\operatorname{Ric}(\eta^\sharp,\eta^\sharp)\right)\,d\mu. \tag*{(36)}

Since Ric is positive definite, both terms in the integrand are non-negative and

Ric⁡(η♯,η♯)>0\operatorname{Ric}(\eta^\sharp,\eta^\sharp)>0

whenever η♯≠0\eta^\sharp\ne0. Consequently, (5.3) implies

η≡0.\eta\equiv0.

It follows from (5.1) that

v=dh,v=dh,

and hence

V=∇h.(37)V=\nabla h. \tag*{(37)}

We next exploit the mm-modified conformal equation. As V=∇hV=\nabla h, equation (1.5) becomes

Hess⁡h−1mdh⊗dh=σg.(38)\operatorname{Hess} h-\frac{1}{m}dh\otimes dh=\sigma g. \tag*{(38)}

Let us define: u=e−h/mu=e^{-h/m}. Then

du=−1me−h/mdhdu=-\frac{1}{m}e^{-h/m}dh

and a direct computation gives

Hess⁡u=e−h/mm2dh⊗dh−e−h/mmHess⁡h.(39)\operatorname{Hess} u=\frac{e^{-h/m}}{m^2}dh\otimes dh-\frac{e^{-h/m}}{m}\operatorname{Hess} h. \tag*{(39)}

Using (5.5) in (5.6), we obtain

Hess⁡u=e−h/mm2dh⊗dh−e−h/mm(1mdh⊗dh+σg)=−1me−h/mσg.\begin{aligned} \operatorname{Hess} u&=\frac{e^{-h/m}}{m^2}dh\otimes dh-\frac{e^{-h/m}}{m}\left(\frac{1}{m}dh\otimes dh+\sigma g\right)\\ &=-\frac{1}{m}e^{-h/m}\sigma g. \end{aligned}

Therefore,

Hess⁡u=λg,λ=−1me−h/mσ.(40)\operatorname{Hess} u=\lambda g,\qquad\lambda=-\frac{1}{m}e^{-h/m}\sigma. \tag*{(40)}

Thus ∇u\nabla u is a gradient conformal vector field. It remains to verify that ∇u\nabla u is non-trivial. Suppose, to the contrary, that σ=0\sigma=0. Then integrating equation (2.1) and using the divergence theorem would imply V≡0V\equiv0, a contradiction to the nontriviality of VV. Therefore, σ\sigma is a non-zero function on MM. Hence (M,g)(M,g) admits a non-trivial gradient conformal vector field. To prove part (i), we assume that MM has constant scalar curvature. Since n≥3n \ge3, Theorem 5.1 applies and yields that (M,g)(M,g) is isometric to a round sphere. This completes the proof of part (i).

For part (ii), the positivity of the Ricci curvature, together with the assumed upper bound on the Ricci curvature, allows us to apply Theorem 5.2. Consequently, (M,g)(M,g) is isometric to a round sphere. This completes the proof.

Corollary 5.1 Let (Mn,g)(M^n,g) be an nn-dimensional compact Riemannian manifold, whose Ricci curvature satisfies

0<Ric⁡≤(n−1)(2−ncλ1)c0 < \operatorname{Ric} \le(n-1)\left(2-\frac{nc}{\lambda_1}\right)c

for a constant cc, where λ1\lambda_1 is the first nonzero eigenvalue of the Laplace operator. If MM admits a non-trivial closed conformal vector field, then MM is isometric to a round sphere.

Proof. The result follows immediately by letting m=∞m=\infty in (1.5) and applying Theorem 5.3 (ii).

Corollary 5.2 Let (Mn,g)(M^n,g), n≥3n \ge3, be a compact Einstein manifold admitting a non-trivial closed mm-modified conformal vector field (m≠∞m \ne\infty). Then (M,g)(M,g) is isometric to a round sphere.

Proof. The result follows directly from Theorem 5.3 (i) and Proposition 4.1 (here non-conformal means (m≠∞m \ne\infty)), together with the fact that an Einstein manifold has constant scalar curvature for n>2n > 2, as a consequence of Schur’s lemma.

Remark 5.1 Corollary 5.2 raises the following question “What can we say about a compact Einstein manifold admitting a non-trivial non-closed mm-modified conformal vector field (m≠∞m \ne\infty)?” For m=∞m=\infty (VV conformal) we know, by a classical result of Yano and Nagano [14], that a compact Einstein manifold MM admitting a non-Killing conformal vector field VV is isometric to a Euclidean sphere (this result was proved more generally in [14], for a complete MM under the condition that VV is complete. For MM compact, VV is automatically complete.)

Closed mm-modified Conformal Vector Fields On Kähler Manifolds

An almost Hermitian manifold MM is a real 2n2n-dimensional smooth manifold with a (1,1)(1,1) tensor field JJ and a Riemannian metric gg such that J2=−IJ^2=-I and g(JX,JY)=g(JX,JY) = g(X,Y)g(X,Y) for arbitrary vector fields X,YX,Y on MM. If JJ is integrable, i.e., the Nijenhuis tensor NN defined by N[X,Y]=[JX,JY]−[X,Y]−J[JX,Y]−J[X,JY]N[X,Y]=[JX,JY]-[X,Y]-J[JX,Y]-J[X,JY] vanishes, then MM is a Hermitian manifold. An almost Hermitian manifold is said to be an almost Kähler manifold if the fundamental 2-form Ω\Omega defined by Ω(X,Y)=g(JX,Y)\Omega(X,Y)=g(JX,Y) is closed i.e., dΩ=0d\Omega=0 (dd is the exterior derivative operator). An almost Kähler manifold whose underlying almost complex structure JJ is integrable i.e., the Nijenhuis tensor of JJ vanishes on MM, is known as a Kähler manifold. It is known that for a Kähler manifold ∇J=0\nabla J=0, where ∇\nabla is the Riemannian connection of gg. Conformal vector fields on Kähler manifolds are known to exhibit strong rigidity phenomena. A classical result of Goldberg [7] says that a closed conformal vector field (i.e., the metrically equivalent 1-form is closed) on a Kähler manifold is necessarily homothetic. In this section, we examine the impact of the existence of a non-zero closed mm-modified conformal vector field VV on a Kähler manifold, and prove the following result.

Theorem 6.1 If a complete Kähler manifold (M,g,J)(M,g,J) of real dimension 2n>22n>2 admits a real non-trivial mm-modified (m≠∞m\ne\infty) closed conformal vector field VV, then VV reduces to mm-modified (m≠∞m\ne\infty) Killing vector field, and around any non-zero of VV, MM splits as the product C×N\mathbb{C}\times N, where C\mathbb{C} is the Euclidean complex plane and NN is a complete Kähler manifold of real dimension 2n−22n-2.

Proof. The mm-modified equation can be written in view (2.6) and (2.4) as

∇XV=1mg(V,X)V+σX(41)\nabla_XV=\frac{1}{m}g(V,X)V+\sigma X \tag*{(41)}

for arbitrary X∈X(M)X\in\mathfrak{X}(M), because VV is closed, i.e. dv=0dv=0 and hence F=0F=0. Operating (6.1) by the complex structure tensor JJ we have

J∇XV=1mg(V,X)JV+σJX(42)J\nabla_XV=\frac{1}{m}g(V,X)JV+\sigma JX \tag*{(42)}

Next, substituting JXJX for XX in (6.1) we also have

∇JXV=1mg(V,JX)JV+σJX(43)\nabla_{JX}V=\frac{1}{m}g(V,JX)JV+\sigma JX \tag*{(43)}

Before proceeding ahead, we denote the 1-form metrically equivalent to JVJV by (JV)b(JV)^b, i.e., (JV)b(X)=g(JV,X)(JV)^b(X)=g(JV,X). Using (6.2) and (6.3) we find that

2d(JV)b(X,Y)=g(∇XJV,Y)−g(∇YJV,X)=g(J∇XV,Y)−g(J∇YV,X)2d(JV)^b(X,Y)=g(\nabla_XJV,Y)-g(\nabla_YJV,X) = g(J\nabla_XV,Y)-g(J\nabla_YV,X)

The use of equation (6.2) in the above equation entails

2(d(JV)♭)(X,Y)=1m[g(V,X)g(JV,Y)−g(JV,X)g(V,Y)]+2σg(JX,Y),(44)2\bigl(d(JV)^\flat\bigr)(X,Y)=\frac{1}{m}\bigl[g(V,X)g(JV,Y)-g(JV,X)g(V,Y)\bigr]+2\sigma g(JX,Y), \tag*{(44)}

where dd is the exterior derivative operator and X,YX,Y arbitrary vector fields on MM. Skew-symmetrizing it with respect to X,YX,Y shows that

2d(JV)♭=1mV♭∧(JV)♭+2σΩ.(45)2d(JV)^\flat=\frac{1}{m}V^\flat\wedge(JV)^\flat+2\sigma\Omega. \tag*{(45)}

Operating (6.5) with dd gives

1mV♭∧d(JV)♭=2dσ∧Ω(46)\frac{1}{m}V^\flat\wedge d(JV)^\flat=2d\sigma\wedge\Omega \tag*{(46)}

where we used the co-boundary formula: d2=0d^2=0, dv=0dv=0, and the property: d(α∧β)=dα∧β+(−1)pα∧dβd(\alpha\wedge\beta)=d\alpha\wedge\beta+(-1)^p\alpha\wedge d\beta for pp and qq forms α,β\alpha,\beta respectively. Expanding (6.6) for arbitrary X,Y,Z∈X(M)X,Y,Z\in\mathfrak{X}(M) we have

v(X)d(JV)♭(Y,Z)+v(Y)d(JV)♭(Z,X)+v(Z)d(JV)♭(X,Y)=2m[(Xσ)Ω(Y,Z)+(Yσ)Ω(Z,X)+(Zσ)Ω(X,Y)].\begin{aligned} v(X)d(JV)^\flat(Y,Z)+v(Y)d(JV)^\flat(Z,X)+v(Z)d(JV)^\flat(X,Y) &=2m\bigl[(X\sigma)\Omega(Y,Z)+(Y\sigma)\Omega(Z,X)\\ &\qquad+(Z\sigma)\Omega(X,Y)\bigr]. \end{aligned}

Using (6.4) in the foregoing equation and factoring out ZZ provides

σ[g(V,X)JY−g(V,Y)JX+g(JX,Y)V]=2m[(Xσ)JY−(Yσ)JX+g(JX,Y)∇σ].\sigma\bigl[g(V,X)JY-g(V,Y)JX+g(JX,Y)V\bigr] =2m\bigl[(X\sigma)JY-(Y\sigma)JX+g(JX,Y)\nabla\sigma\bigr].

Contracting it with respect to YY we get

∇σ=σ2mV(47)\nabla\sigma=\frac{\sigma}{2m}V \tag*{(47)}

because 2n>22n>2, by hypothesis. Now, using (6.1) we compute and find that

Riem(Y,X)V=(Yσ)X−(Xσ)Y+σm[v(X)Y−v(Y)X].(48)\mathit{Riem}(Y,X)V=(Y\sigma)X-(X\sigma)Y+\frac{\sigma}{m}\bigl[v(X)Y-v(Y)X\bigr]. \tag*{(48)}

Contracting it at YY and using (6.7) we find that

Ric(V)=(2n−1)∇σ.\mathit{Ric}(V)=(2n-1)\nabla\sigma.

where we have denoted the Ricci tensor of type (1,1)(1,1) also by Ric\mathit{Ric}. As Ric\mathit{Ric} commutes with JJ on a Kähler manifold, the foregoing equation implies that

Ric(JV)=(2n−1)J∇σ.(49)\mathit{Ric}(JV)=(2n-1)J\nabla\sigma. \tag*{(49)}

Going back to (6.8), operating by JJ and using the Kähler property: JRiem⁡(X,Y)Z=Riem⁡(X,Y)JZJ\operatorname{Riem}(X,Y)Z=\operatorname{Riem}(X,Y)JZ for all X,Y,Z∈X(M)X,Y,Z\in\mathfrak{X}(M), and then contracting at YY gives

Ric⁡(JV)=−J∇σ+σmJV(50)\operatorname{Ric}(JV)=-J\nabla\sigma+\frac{\sigma}{m}JV \tag*{(50)}

Comparing equations (6.9) and (6.10) and noting that 2n>22n>2 and m≠∞m\ne\infty by hypothesis, shows that

σV=0.(51)\sigma V=0. \tag*{(51)}

Using this in (6.7) at once implies that σ\sigma is constant. Therefore, as V≠0V\ne0, (6.11) implies that σ=0\sigma=0. Hence (6.1) reduces to

∇XV=1mg(V,X)V.(52)\nabla_XV=\frac{1}{m}g(V,X)V. \tag*{(52)}

Now, as gg is Hermitian, g(JX,JY)=g(X,Y)g(JX,JY)=g(X,Y), we have g(JX,Y)=−g(X,JY)g(JX,Y)=-g(X,JY) for arbitrary vector fields X,Y∈X(M)X,Y\in\mathfrak{X}(M). So g(JV,V)=−g(V,JV)=0g(JV,V)=-g(V,JV)=0, and hence JVJV is orthogonal to VV. On an open dense subset O\mathcal{O} of MM, let V≠0V\ne0, and (V,JV)(V,JV) denote the 2-dimensional distribution spanned by VV and JVJV. To show that (V,JV)(V,JV) is integrable, it suffices to show that [V,JV]∈(V,JV)[V,JV]\in(V,JV). For this, [V,JV]=∇VJV−∇JVV=J∇VV−∇JVV=1m∣V∣2JV[V,JV]=\nabla_VJV-\nabla_{JV}V=J\nabla_VV-\nabla_{JV}V=\frac{1}{m}|V|^2JV, where we used (6.12). Now, for real functions a,ba,b on O\mathcal{O}, computing

∇aV+bJV(cV+dJV)=a[(Vc)V+cm∣V∣2V+(Vd)JV+dm∣V∣2JV]+b[(JV)cV+(JV)d(JV)]\nabla_{aV+bJV}(cV+dJV)=a\left[(Vc)V+\frac{c}{m}|V|^2V+(Vd)JV+\frac{d}{m}|V|^2JV\right]+b\left[(JV)cV+(JV)d(JV)\right]

shows that (V,JV)(V,JV) is totally geodesic in O\mathcal{O}. Next we consider the distribution (V,JV)⊥(V,JV)^\perp orthogonal to (V,JV)(V,JV) and complementary in the tangent bundle on O\mathcal{O}. Let X,Y∈(V,JV)⊥X,Y\in(V,JV)^\perp, i.e., g(X,W)=0g(X,W)=0, g(Y,W)=0g(Y,W)=0 for any W∈(V,JV)W\in(V,JV), i.e. W=aV+bJVW=aV+bJV, for arbitrary real functions a,ba,b on O\mathcal{O}. Then we have g(∇YX,W)+g(X,∇YW)=0g(\nabla_YX,W)+g(X,\nabla_YW)=0 and g(∇XY,W)+g(Y,∇XW)=0g(\nabla_XY,W)+g(Y,\nabla_XW)=0. Subtracting them and using (6.12), we get g([X,Y],W)=0g([X,Y],W)=0. Thus [X,Y]∈(V,JV)⊥[X,Y]\in(V,JV)^\perp. Finally, to show that (V,JV)⊥(V,JV)^\perp is totally geodesic in O\mathcal{O}, we compute

g(∇XY,aV+bJV)=−ag(∇XV,Y)−bg(J∇XV,Y)g(\nabla_XY,aV+bJV)=-ag(\nabla_XV,Y)-bg(J\nabla_XV,Y)
−amv(X)v(Y)−bmv(X)g(JV,Y)=0,-\frac{a}{m}v(X)v(Y)-\frac{b}{m}v(X)g(JV,Y)=0,

because X,YX,Y are ⊥\perp to V,JVV,JV. Also, from (6.8) and the consequence σ=0\sigma=0, we have Riem⁡(X,Y)V=0\operatorname{Riem}(X,Y)V=0 [and hence Riem⁡(X,Y)JV=JRiem⁡(X,Y)V=0\operatorname{Riem}(X,Y)JV=J\operatorname{Riem}(X,Y)V=0] for any X,Y∈X(M)X, Y \in\mathcal{X}(M). Thus the leaves of the foliation (V,JV)(V, JV) are flat. We also note that a totally geodesic submanifold in Kähler MM of codimension 2 is also Kähler. Summing up, we have shown that around any non-zero of VV, MM splits locally as C×N\mathbb{C} \times N where NN is a complete (2n−2)(2n - 2)-dimensional Kähler manifold. This completes the proof.

Statements and Declarations

Competing Interests: On behalf of all the authors, the corresponding author states that there is no conflict of interest.

Data Availability: Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.

No Use Of Artificial Intelligence: No Artificial Intelligence has been used in the preparation of this article.

References

  1. [1]Barros, A. and Ribeiro, E. Integral formulae on quasi-Einstein manifolds and applications. Glasgow Math. J. 54 (2012), 213-223.DOI
  2. [2]Bourguignon, J.-P. and Ezin, J.-P. Scalar curvature functions in a conformal class of metrics and conformal transformations. A.M.S. 301 (1987), 723-736.
  3. [3]Case, J., Shu, Y. and Wei, G. Rigidity of quasi-Einstein metrics. Diff. Geom. Appln. 29 (2011), 93-100.arxiv.org/abs/0805.3132
  4. [4]Catino, G. Generalized quasi-Einstein manifolds with harmonic Weyl tensor. Math. Z. 271 (2012), 751-756.
  5. [5]Deshmukh, S. and Al-Solamy, F. Conformal gradient vector fields on a compact Riemannian manifold. Colloq. Math. 114 (2009), 71-80.DOI
  6. [6]Deshmukh, S., Turki, N. B., Chen, B-Y. and Fatima, T. Properties of m-modified conformal vector fields on a Riemannian manifold. Mathematics. 11(9) (2026), 29774-29795.DOI
  7. [7]Goldberg, S.I.: Curvature and Homology, Academic Press, N.Y. 1964.
  8. [8]He, C., Petersen, P. and Wylie, W. On the classification of warped product Einstein metrics. Comm. Anal. Geom. 20 (2012), 271-311.arxiv.org/abs/1010.5488
  9. [9]Limoncu, M. Modifications of the Ricci tensor and applications. Arch. Math. 95 (2010)DOI
  10. [10]Poddar, R. and Sharma, R. On the triviality of m-modified conformal vector fields. Indag. Math. (2025)arxiv.org/abs/2409.07607
  11. [12]Yano, K. On harmonic and Killing vector fields. Ann. Math. 55 (1952), 38-46.
  12. [13]Yano, K. Integral formulas in Riemannian geometry, Marcel Dekker, 1970.
  13. [14]Yano, K. and Nagano, T. Einstein spaces admitting a one-parameter group of conformal transformations. Ann. Math. 69 (1959), 451-461.
  14. [15]Zhang, H. and Chen, Z. On m-Modified Conformal Vector Fields. J. Geom. Anal. 33 (2023), 258.DOI

Paper details

vlnc:W_6kr49q7s
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents