---
title: Improved Chen–Ricci Inequalities
url: https://www.emergentmind.com/topics/improved-chen-ricci-inequalities
type: topic
---

# Improved Chen–Ricci Inequalities

Improved Chen–Ricci inequalities are sharp or sharpened intrinsic–extrinsic comparison estimates that bound Ricci-type curvature quantities of a geometric object by quadratic expressions in its mean-curvature-type data, together with ambient correction terms determined by the surrounding geometry. In the modern literature, the term encompasses several closely related developments: abstract inequalities for curvature-like tensors, refined Ricci bounds for submanifolds in complex, contact, and quaternionic geometries, and generalized versions for Riemannian submersions and Riemannian maps. A recurring feature is that the classical Chen–Ricci pattern is preserved while additional geometric structure—such as a complex structure, contact tensor, warping function, or map-induced second fundamental tensor—produces sharper coefficients, new correction terms, or more rigid equality cases [1104.3439], [1512.07647], [2509.15281].

## 1. Classical pattern and the meaning of “improved”

The classical point of departure is Chen’s Ricci inequality for an \(n\)-dimensional submanifold of a real space form \(R^m(c)\):
\[
n|H|^{2}\geq \{Ric(X)-(n-1)c\},
\]
for each unit tangent vector \(X\). The standard equality theory already exhibits the characteristic rigidity of Chen inequalities: if \(H(p)=0\), equality for a unit vector is equivalent to membership in the relative null space, and equality for all unit vectors occurs only at a geodesic point or, when \(n=2\), at an umbilical point [1104.3439].

In later work, an inequality is called “improved” when the right-hand side is sharpened in one of two ways. First, the coefficient of the extrinsic term may be reduced; for Riemannian maps, the general bound
\[
{\rm Ric}^{\cal H}(X)\leq {\rm Ric}^{\cal R}(F_{\ast }X) + \frac14\|\operatorname{trace}B^{\mathcal H}\|^2
\]
is improved to
\[
{\rm Ric}^{\cal H}\left( X\right) \leq {\rm Ric}^{\cal R}\left( F_{\ast}X\right) +\frac{(r-1)}{4r}\left\Vert {\rm trace~}B^{\cal H}\right\Vert^{2},
\]
and this is explicitly described as sharper because \(\frac{r-1}{4r}\le \frac14\) for \(r>1\) [2509.15281]. Second, the inequality may retain the classical Chen–Ricci shape while incorporating structure-sensitive correction terms. In generalized \((\kappa,\mu)\)-contact space forms, for example, the Ricci bound contains additional \(f_4\), \(f_{5,1}\), and \(f_{5,2}\) contributions involving \(h^T\) and \((\phi h)^T\), and the paper characterizes these as contact-structure corrections [1512.07647].

This suggests that “improvement” in the Chen–Ricci literature is not limited to numerical sharpening. It also includes refinement of the geometric content of the bound, so that the inequality detects more of the ambient and induced structure than the original real-space-form estimate.

## 2. Abstract algebraic framework

A decisive generalization is the curvature-like tensor approach. A tensor \(T\) is called curvature-like if it satisfies the standard algebraic curvature symmetries, and if \(T\) fits an algebraic Gauss equation
\[
T(X,Y,Z,W)=g_B(\zeta(X,W),\zeta(Y,Z))-g_B(\zeta(X,Z),\zeta(Y,W)),
\]
then \(\zeta\) plays the role of an abstract second fundamental form. In this setting, the associated \(T\)-Ricci curvature satisfies the general estimate
\[
Ric_T(X)\leq \frac{1}{2}|\operatorname{trace}\zeta|^2.
\]
Equality for a given unit vector \(X\) is characterized by the conditions \(\zeta(X,Y)=0\) for all \(Y\perp X\) and
\[
\zeta(X,X)=\frac{1}{2}\operatorname{trace}\zeta,
\]
while equality for all unit vectors occurs only when \(\zeta=0\) or when \(n=2\) and the indicated two-dimensional condition holds [1104.3439].

The improved version appears when \(\zeta\) satisfies an additional normal-frame condition denoted in the paper by equation \((3.1)\). Under that restriction, the bound becomes
\[
Ric_T(X)\leq \frac{n}{n-1}|\operatorname{trace}\zeta|^2.
\]
The proof uses algebraic lemmas of Deng to estimate the diagonal components under a trace constraint, and the strengthened coefficient is the algebraic source of several later geometric improvements [1104.3439].

The applications in the same work show how this abstract formalism specializes to concrete submanifold geometries. For a Lagrangian submanifold of a complex space form,
\[
Ric(X)\leq \frac{1}{4}\big(c+n|H(p)|^2\big).
\]
For an \(n\)-dimensional Kaehlerian slant submanifold of a \(2n\)-dimensional complex space form,
\[
Ric(X)\leq \frac{1}{4}\big((n-1)n|H|^2+(n-1)c+3c\cos^2\theta\big).
\]
For a \(C\)-totally real submanifold of a Sasakian space form,
\[
Ric(X)\leq \frac{1}{4}\big((n-1)(c+3)+n|H|^2\big).
\]
In each case, the ambient structure is encoded by modifying the curvature-like tensor inserted into the abstract theorem, so the improvement is transferred from an algebraic statement to a geometric one [1104.3439].

## 3. Improved Chen–Ricci inequalities in contact and related submanifold geometries

A major development concerns \(C\)-totally real submanifolds in generalized \((\kappa,\mu)\)-contact space forms with divided \(R_5\). The ambient curvature tensor is written as
\[
\widetilde R = f_1R_1+f_2R_2+f_3R_3+f_4R_4+f_{5,1}R_{5,1}+f_{5,2}R_{5,2}+f_6R_6,
\]
and the submanifold condition is that the structure vector field \(\xi\) is normal, equivalently \(\eta(X)=0\) for all \(X\in TM\). The central Ricci estimate is
\[
\begin{aligned}
\Ric(U)\le\; &\frac14 n^2\|H\|^2 +(n-1)f_1 +f_4\big(\tr(h^T)+(n-2)g(h^TU,U)\big)\\
&+f_{5,1}\big(\tr(h^T)\,g(h^TU,U)-\|h^TU\|^2\big)\\
&+f_{5,2}\big(\tr((\phi h)^T)\,g((\phi h)^TU,U)-\|(\phi h)^TU\|^2\big).
\end{aligned}
\]
This is explicitly presented as the paper’s main Chen–Ricci inequality in the generalized \((\kappa,\mu)\)-contact setting. Its equality case is determined by the relations
\[
\sigma_{11}^r=\sigma_{22}^r+\cdots+\sigma_{nn}^r,\qquad \sigma_{12}^r=\cdots=\sigma_{1n}^r=0,
\]
for \(r=n+1,\dots,2m+1\). If \(H(p)=0\), equality holds iff \(U\in\mathcal N_p\), and equality for all unit tangent vectors occurs only when the point is totally geodesic or, for \(n=2\), totally umbilical [1512.07647].

The same paper exhibits a Sasakian-type specialization, defined by
\[
f_3=f_1-1,
\]
for which the correction terms disappear and the sharp bound simplifies to
\[
\Ric(U)\le \frac14 n^2\|H\|^2+(n-1)f_1,
\]
together with the tensor inequality
\[
4S\le \big(n^2\|H\|^2+4(n-1)f_1\big)g.
\]
This specialization is described as a cleaner, sharper Chen–Ricci bound because the \(h\)- and \(\phi h\)-terms vanish [1512.07647].

A different ambient refinement appears for Lagrangian submanifolds of Kähler QCH-manifolds. With ambient curvature tensor
\[
R=a\,\pi+b\,\Phi+c\,\Psi,
\]
the upper Ricci estimate becomes
\[
\operatorname{Ric}(X) \le \frac{1}{4} \left\{ a+\bigl((n-2)\bigl[\eta(X)^2+\tilde\eta(X)^2\bigr]+1\bigr)b +\bigl|\eta(X)\tilde\eta-\tilde\eta(X)\eta\bigr|^2c \right\} +\frac{n-1}{4}n|H|^2.
\]
The paper states that this extends the known complex-space-form estimate to the broader QCH class, and when \(b=0\) and \(c=0\) it reduces to the improved Chen–Ricci inequality for complex space forms [1709.10002].

The paper also proves a lower Ricci bound,
\[
\operatorname{Ric}(X) \ge \frac{n-2}{4}a -\frac{1}{4}\Bigl\{ n-\bigl[n\eta(X)^2+n\tilde\eta(X)^2\bigr] \Bigr\}b -\frac{1}{4}\Bigl\{ (n-1)\bigl(|\xi|^2|\tilde\xi|^2-g(\xi^T,\tilde\xi^T)^2\bigr) -\bigl|\eta(X)\tilde\eta-\tilde\eta(X)\eta\bigr|^2 \Bigr\}c +\frac{(n-1)(n-2)}{n}|H|^2,
\]
with equality for every unit vector only at totally geodesic points. A plausible implication is that the modern Chen–Ricci program has broadened from one-sided upper estimates to a more flexible Ricci-comparison theory in structured Kähler settings [1709.10002].

## 4. Quaternionic and non-Levi-Civita extensions

In quaternionic Kähler geometry endowed with a Ricci quarter-symmetric metric connection, the Chen–Ricci philosophy persists but the curvature terms are altered by the connection tensor \(M\) and the quaternionic projections \(P_\alpha\). The ambient connection is
\[
\nabla_X Y = \nabla^*_X Y+\eta(Y)LX-S(X,Y)P,
\]
and the curvature tensor of \(\nabla\) is expressed by the corresponding deformation formula involving \(R^*\), \(M\), \(Q\), and \(P\). For an \(n\)-dimensional submanifold \(N'^n\), the Ricci estimate takes the form
\[
\operatorname{Ric}(X) \le c\Bigl((n-1)+3\sum_{\alpha=1}^3 g^2(J_\alpha X,e_j)\Bigr) -\frac{c}{n}\Bigl((n-1)+3\sum_{\alpha=1}^3 \|P_\alpha\|^2\Bigr)\,[m+(n-2)M(X,X)] +\frac{n^2}{4}\|H\|^2.
\]
The paper identifies this as the result most directly connected to improved Chen–Ricci inequalities in its setting [2011.08582].

Its equality case is likewise rigid: off-diagonal second fundamental form components must vanish,
\[
h_{ij}^r=0,\qquad i\ne j,
\]
and the diagonal components must satisfy the Chen-lemma equality condition. The conclusions are dimension-sensitive: for \(n\neq 2\), equality forces a totally geodesic point, whereas for \(n=2\) it yields a totally umbilical situation [2011.08582].

The same paper also derives scalar-curvature and \(\tau-K(T)\) Chen-type inequalities, as well as generalized normalized Casorati inequalities. Those results are not Ricci inequalities in the narrow sense, but they demonstrate a systematic extension of the Chen framework to non-Levi-Civita connections and quaternionic ambient structures. This suggests that improved Chen–Ricci inequalities are part of a larger optimization-based family of intrinsic–extrinsic estimates rather than an isolated theorem schema [2011.08582].

## 5. Warped products, leaf-wise invariants, and generalized \(\delta\)-frameworks

Although not every recent refinement is a Ricci inequality in the strict sense, several are structurally adjacent to the improved Chen–Ricci program because they replace full scalar or Ricci curvature by adapted invariants of distributions or factors. For CR-warped product submanifolds
\[
M^n = N_T \times_f N_\perp
\]
in a complex space form \(\tilde M^{2m}(c)\), the 2026 paper introduces a leaf-wise \(\delta\)-invariant
\[
\hat\delta(V)(x):=\tau^M(V)-\inf\{K^M(\pi):\pi\subseteq V,\ \dim\pi=2\}.
\]
The distinction between \(\hat\delta(T_xN_\perp)\) and the intrinsic Chen invariant of \(N_\perp\) is controlled by the Bishop–O’Neill formula,
\[
\hat\delta(T_xN_\perp)(x) = \frac{1}{f(p)^2}\,\delta_{N_\perp}(q) -\Bigl[\binom{n_2}{2}-1\Bigr]\frac{\|\nabla f(p)\|^2}{f(p)^2},
\]
while the mixed sectional curvature identity
\[
\sum_{a=1}^{n_1}\sum_{A=n_1+1}^{n} K^M(e_a\wedge e_A)=\frac{n_2\,\Delta f}{f}
\]
supplies the warping contribution to the inequality [2605.19601].

The resulting bounds are
\[
\hat\delta(T_xN_T)(x)\leq\frac{n^2}{2}\|\vec H\|^2-\frac{n_2\Delta f}{f}+\frac{n_1(n_1+2n_2+2)}{2}\cdot\frac{c}{4}-\tilde K_{\min}(T_xN_T),
\]
and
\[
\hat\delta(T_xN_\perp)(x)\leq\frac{n^2}{2}\|\vec H\|^2-\frac{n_2\Delta f}{f}+\frac{n_2(n_2+2n_1-1)}{2}\cdot\frac{c}{4}-\frac{c}{4}.
\]
The paper calls these the “first Chen inequalities” for CR-warped products and emphasizes that the bound is sharp and uniform in the sign of \(c\) [2605.19601].

A related earlier CR-warped-product result in Bochner Kähler geometry gives
\[
\|\omega\|^2 \ge \|\omega_{D D}\|^2 + q\,\|\nabla_D(\ln f)\|^2,
\]
and, under compactness and an additional compatibility condition,
\[
\rho \le 0,\qquad \rho=0 \iff \nabla_D(\ln f)=0.
\]
These are scalar-curvature and second-fundamental-form estimates rather than Chen–Ricci inequalities, but they show how warping data enters as an unavoidable extrinsic correction term in the same general comparison philosophy [1601.04130].

The generalized \(\delta\)-invariant program reaches beyond submanifolds. For Riemannian submersions and Riemannian maps, the 2026 paper defines
\[
2\delta(r_1,\dots,r_k)(p) =\tau(p)-\inf\{\tau(L_1)+\cdots+\tau(L_k)\},
\]
\[
2\widehat{\delta}(r_1,\dots,r_k)(p) =\tau(p)-\sup\{\tau(L_1)+\cdots+\tau(L_k)\},
\]
and derives vertical and horizontal analogues adapted to the map geometry. The main submersion inequality is
\[
\delta_{\cal V}^{\ker \pi_*(r_1,\dots,r_k)}(p) \ge \delta_{\cal V}^{M_1(r_1,\dots,r_k)}(p) -c(r_1,\dots,r_k)\|H\|^2,
\]
with
\[
c(r_1,\dots,r_k) =\frac{r^2(r+k-1-\sum_{j=1}^k r_j)}{2(r+k-\sum_{j=1}^k r_j)}.
\]
The corresponding Riemannian-map inequality is
\[
{\rm scal}^{\cal H}-\sum_{j=1}^k {\rm scal}^{\cal H}(L_j) \le {\rm scal}^{\cal R}-\sum_{j=1}^k {\rm scal}^{\cal R}(\pi_*L_j) +c(r_1,\dots,r_k)\,\|{\rm trace}\,B^{\cal H}\|^2.
\]
These statements are presented as generalized B.-Y. Chen inequalities and are explicitly linked to the improved Chen–Ricci tradition in spirit and method [2606.18725].

## 6. Equality, rigidity, and current directions

Across the literature, equality cases are not incidental; they encode the extremal geometry. In the curvature-like tensor framework, equality forces the vanishing of mixed \(\zeta\)-components and alignment of \(\zeta(X,X)\) with the trace, and global equality leads to either \(\zeta=0\) or a rigid two-dimensional model [1104.3439]. In the generalized \((\kappa,\mu)\)-contact setting, equality in the scalar-curvature inequality requires shape operators of block form
\[
A_{n+1}= \begin{bmatrix} a&0&0\\ 0&b&0\\ 0&0&(a+b)I_{n-2} \end{bmatrix},
\qquad
A_r= \begin{bmatrix} c_r&d_r&0\\ d_r&-c_r&0\\ 0&0&0_{n-2} \end{bmatrix},
\]
and the Ricci equality conditions impose the linear relations in equation \((38)\) of that paper [1512.07647].

In the QCH Lagrangian setting, equality for all unit tangent vectors holds only at totally geodesic points or, for \(n=2\), at \(H\)-umbilical points with
\[
h(e_1,e_1)=\lambda Je_1,\qquad
h(e_2,e_2)=3\lambda Je_1,\qquad
h(e_1,e_2)=\lambda Je_2.
\]
The quaternionic quarter-symmetric setting yields analogous conclusions: equality either forces total geodesicity or a two-dimensional totally umbilical configuration [1709.10002], [2011.08582].

The same rigidity appears in map-theoretic versions. For Riemannian maps, equality in the improved bound
\[
{\rm Ric}^{\cal H}\left( X\right) \leq {\rm Ric}^{\cal R}\left( F_{\ast}X\right) +\frac{(r-1)}{4r}\left\Vert {\rm trace~}B^{\cal H}\right\Vert^{2}
\]
holds for all unit horizontal \(X\) iff either \(B^{\mathcal H}=0\) or \(r=2\) and
\[
\left( \nabla F_{\ast }\right) \left( h_{1},h_{1}\right) = 3\mu V_{r+1}, \qquad
\left( \nabla F_{\ast }\right) \left( h_{2},h_{2}\right) = \mu V_{r+1}, \qquad
\left( \nabla F_{\ast }\right) \left( h_{1},h_{2}\right) =\mu V_{r+2}.
\]
The 2026 generalized-\(\delta\) paper describes equality in terms of block-balancing and vanishing off-block coefficients of the O’Neill tensor or \(B^{\cal H}\), which is the same rigid pattern in a higher-order invariant language [2509.15281], [2606.18725].

A common misconception is that improved Chen–Ricci inequalities are only coefficient refinements of the original real-space-form estimate. The literature shows a broader picture. Some improvements are genuinely coefficient-theoretic, as in the replacement of \(\frac14\) by \(\frac{r-1}{4r}\) for Riemannian maps [2509.15281]. Others are ambient-structure-sensitive refinements, where extra curvature terms from \(h\), \(\phi h\), \(b\Phi\), \(c\Psi\), quaternionic projections, or warping data sharpen the geometric information without necessarily changing the formal coefficient of the mean-curvature term [1512.07647], [1709.10002], [2605.19601].

Current directions recorded in the cited works include extending CR-warped-product inequalities to locally conformal Kähler space forms, answering open problems associated with the first Chen inequality in the CR-warped setting, and developing generalized Chen inequalities for submersions and maps in real and complex space forms with full equality characterizations [2605.19601], [2606.18725]. Taken together, these developments define improved Chen–Ricci inequalities as a mature and expanding branch of submanifold and mapping geometry, centered on optimal intrinsic–extrinsic comparison and the rigidity of its equality cases.

Source: https://www.emergentmind.com/topics/improved-chen-ricci-inequalities