Manifolds with harmonic Weyl curvature and curvature operator of the second kind Thanks: Supported in part by National Natural Science Foundations of China #12461008 and 12271069, Jiangxi Province Natural Science Foundation of China #20202ACB201001, Jiangxi Province Graduate Student Innovation Special Fund Project #YC2025-B037. Haiping Fu, Yao Lu Abstract . We prove that a compact Riemannian manifold of dimension n ≥ 8 n\geq 8 with harmonic Weyl curvature and 3 ​ ( n − 1 ) ​ ( n + 2 ) 4 ​ ( 3 ​ n − 1 ) \frac{3(n-1)(n+2)}{4(3n-1)} -nonnegative curvature operator of the second kind is either globally conformally equivalent to a space of positive constant curvature or is isometric to a flat manifold. In particular, We also give a classification of four-dimensional manifolds with harmonic Weyl curvature satisfying a cone condition. This result generalizes the work in [ 9 , 15 , 20 ] . 2010 Mathematics Subject Classification Primary 53C20, 53C24, 53C25. 1. introduction A central theme in Riemannian geometry is the study of the relationship between curvature operator and the topological structure of Riemannian manifolds. A well-known theorem of Tachibana [ 30 ] showed that any compact Riemannian manifold of dimension n\geq 3 with harmonic curvature and positive(resp., nonnegative) curvature operator is isometric to a quotient of the standard sphere(resp., locally symmetric). This result was improved by Tran [ 31 ] , who proved that a compact Riemannian manifold of dimension n\geq 4 with harmonic Weyl curvature and positive curvature operator is locally conformally flat. In [ 29 ] , Petersen and Wink showed that a compact Riemannian manifold with harmonic Weyl tensor and \left[\frac{n-1}{2}\right] -nonnegative curvature operator is either globally conformally equivalent to a quotient of the standard sphere or locally symmetric; the former case is guaranteed when the curvature operator is \left[\frac{n-1}{2}\right] -positive. Subsequently, in [ 28 ] , they established a related Tachibana-type theorem under the stronger Einstein condition. More recently, Colombo, Mariani, and Rigoli [ 7 ] proved that a compact Riemannian manifold of dimension n\geq 3 with harmonic curvature and \left[\frac{n-1}{2}\right] -positive curvature operator must have constant sectional curvature, generalizing Tachibana’s classical result. This finding was further refined by Bettiol and Goodman [ 2 ] , who showed the condition can be relaxed to \frac{n-1}{2} -positive curvature operator for n\geq 5 , and to \frac{n}{2} -positive curvature operator for n=3,4 . There is another way to define the curvature operator induced by Riemannian curvature tensor, known as the curvature operator of the second kind, see section 2.1 blow for details. In 1986, Nishikawa conjecture [ 25 ] states that a closed Riemannian manifold with positive (resp., nonnegative) curvature operator of the second kind is diffeomorphic to a spherical space form (resp., a Riemannian locally symmetric space). This conjecture was later proved by Cao-Gursky-Tran [ 5 ] and further refined by Li [ 20 , 22 ] . Both proofs relied on Brendle’s convergence result [ 3 ] for closed Ricci flows with PIC1 toward constant sectional curvature. For certain special classes of manifolds, Kashiwada [ 19 ] showed that Riemannian manifolds with harmonic curvature and nonnegative (resp., positive) curvature operator of the second kind are locally symmetric spaces (resp., constant curvature spaces) by using Bochner tenchique. Later, Nienhaus, Petersen, and Wink [ 26 ] derived a new Bochner formula for the curvature operator of the second kind and used it to prove that n -dimensional compact Einstein manifolds with k(<\frac{3n(n+2)}{2(n+4)}) -nonnegative curvature operators of the second kind must be either flat or a rational homology sphere. Subsequently, Dai and Fu [ 8 ] also obtained a new Bochner formula and employed it to prove that any compact Einstein manifold must have constant curvature if its curvature operator of the second kind is k -nonnegative, where k=1 for 4\leq n\leq 7 , k=2 for 8\leq n\leq 10 and k=[\frac{n+2}{4}] for n\geq 11 . In 2024, Dai-Fu-Yang [ 9 ] improved upon Kashiwada’s result by proving that a compact manifold with harmonic Weyl tensor and nonnegative curvature operator of the second kind is either globally conformally equivalent to a space of positive constant curvature or is isometric to a flat manifold. Recently, Fu, Lu and Dai [ 15 ] show a rigidity theorem for manifolds with harmonic curvature, which generalizes the results of [ 19 , 8 ] . For further developments on this topic, we refer the reader to [ 21 , 23 , 24 , 17 , 14 ] . Let the nondecreasing sequence \lambda_{i} be the eigenvalues of the curvature operator. If {\lambda}_{1}+{\lambda}_{2}+\cdots{\lambda}_{[k]}+(k-[k])\lambda_{[k]+1}>0(\geq 0), we say that the curvature operator is k -positive(resp., nonnegative). In this paper, we study Riemannian manifolds of dimension n\geq 4 that possess harmonic Weyl curvature and satisfy conditions on the curvature operator of the second kind. Our main results are the following three theorems. Theorem A . Let (M,g) be an n(\geq 8) -dimensional compact Riemannian manifold with harmonic Weyl curvature tensor. If the curvature operator of the second kind \mathring{R} is \frac{3(n-1)(n+2)}{4(3n-1)} -nonnegative, then M is either globally conformally equivalent to a space of positive constant curvature or is isometric to a flat manifold. Corollary 1.1 . Let (M,g) be an n(\geq 8) (resp., 11 )-dimensional compact Riemannian manifold with harmonic Weyl curvature tensor. If the curvature operator of the second kind \mathring{R} is 2 (resp., 3 )-nonnegative, then M is either globally conformally equivalent to a space of positive constant curvature or is isometric to a flat manifold. Theorem B . Let (M,g) be an n(=5,7) -dimensional compact Riemannian manifold with harmonic Weyl curvature tensor. \mathring{R} be the curvature operator of the second kind. If (i) n=5 and \mathring{R} is \frac{252}{169} -nonnegative; (ii) n=7 and \mathring{R} is \frac{54}{35} -nonnegative, then M is either globally conformally equivalent to a space of positive constant curvature or is isometric to a flat manifold. Remark 1.2 . Theorem A and B generalize the result of Dai, Fu and Yang [ 9 ] , who obtained the same conclusion under non-negative conditions. However, Our method does not apply to the case n=6 . Whether Dai-Fu-Yang’s result can be improved to achieve optimality remains an interesting and open problem. For n(\geq 8) (resp., 11 ), Theorem A can be regarded as a refined version of their theorem due to [ 5 , 22 ] . Furthermore, we obtain a classification of four-dimensional Riemannian manifolds with harmonic Weyl curvature. Theorem C . Let (M,g) be a 4 -dimensional compact manifold with harmonic Weyl curvature whose the curvature operator of the second kind \mathring{R}\, satisfies (1.1) \lambda_{1}+\lambda_{2}+\lambda_{3}\geq-3\bar{\lambda}, where \bar{\lambda} represents the average value of all the eigenvalues of \mathring{R}\, . Then one of the following must be true: (1) M is conformal flat. (2) up to rescaling, M is \mathbb{CP}^{2} with the Fubini–Study metric. (3) up to rescaling, M is isometric to a quotient of \Sigma_{1}^{2}\times\Sigma_{2}^{2} with the product metric, where k_{i}(i=1,2) is constant Gaussian curvature of \Sigma_{i}^{2} and \frac{s}{2}=k_{1}+k_{2}>0 . Remark 1.3 . For 4 -dimensional manifolds, this theorem improves upon [ 14 , Theorem D] and their conclusions due to [ 15 ] . This theorem can be regarded as an optimal rigidity result under certain conditions. The paper is organized as follows. In Sect. 2, we give an introduction to the curvature operator of the second kind and some Lemmas. In Sect. 3, we present a refined Bochner formula for manifold with harmonic Weyl curvature and prove Theorem A. In Sect. 4, we prove Theorem B. The proofs of Theorems C are presented in Sect. 5. 2. Preliminaries 2.1. Notion and Lemmas In a Riemannian manifold (M^{n},g) , the Riemannian curvature tensor R induces two symmetric linear operators \hat{R} and \mathring{R} . Denote by V be the tangent space of this Riemannian manifold, by {\Lambda}^{2}V the space of skew symmetric 2 -tensor over V , and by S_{0}^{2}(V) the space of trace-free symmetric 2 -tensor over V . The curvature operator of the first kind is defined by \displaystyle\hat{R}:{\Lambda}^{2}V\to{\Lambda}^{2}V \displaystyle\hat{R}(e_{i}\wedge e_{j})=\frac{1}{2}\sum\limits_{k,l}{R_{ijkl}e_{k}\wedge e_{l}}, and the curvature operator of the second kind is defined in [ 5 , 26 ] by \mathring{R}={\Pr}_{s_{0}^{2}(V)}\circ\left.\bar{R}\right|_{S_{0}^{2}(V)}, where the operator \bar{R} act on S^{2}(V) induced from Riemannian curvature tensor R is defined by \bar{R}(e_{i}\odot e_{j})=\sum\limits_{k,l}{R_{kijl}e_{k}\odot e_{l}}, where e_{i}\odot e_{j}=e_{i}\otimes e_{j}+e_{j}\otimes e_{i} . The inner product on S^{2}(V) is defined by \left\langle A,B\right\rangle=tr(A^{T}B), Then \frac{1}{\sqrt{2}}\{e_{i}\odot e_{j}\}_{1\leq i0 , then \displaystyle[\mathring{R},c\Omega,cS]=c\cdot[\mathring{R},\Omega,S]. (b) If \Omega\leq\widetilde{\Omega}, then \displaystyle[\mathring{R},\Omega,S]\geq[\mathring{R},\widetilde{\Omega},S]. (c) \displaystyle[\mathring{R},\Omega,S]+[\mathring{R},\widetilde{\Omega},\widetilde{S}]\geq[\mathring{R},\Omega+\widetilde{\Omega},S+\widetilde{S}]. 3. Bochner formula for manifold with harmonic Weyl curvature Let M be an n -dimensional complete manifold with harmonic Weyl curvature, endowed with a standard Riemannian metric \left\langle\text{,}\right\rangle . Denote by W its Weyl tensor and by s its scalar curvature. For simplicity, we set R_{sjti}W_{sjkl}W_{tikl}=\alpha(W)\quad\text{and}\quad R_{sikt}W_{sjkl}W_{ijtl}=\beta(W). Lemma 3.1 . Let M be an n(\geq 4) -dimensional complete Riemannian manifold with harmonic Weyl curvature. Then (3.1) \displaystyle\left\langle\Delta W,W\right\rangle=2R_{lt}W_{ijkl}W_{ijkt}-\alpha(W)-4\beta(W). Remark 3.2 . Although a similar formula has been proven in [ 13 ] , its form of expression differs. For the sake of completeness, we still present its proof here. Proof. Since W is harmonic, by the second Bianchi identity we obtain (see [ 13 ] ) W_{ijkl,h}+W_{ijlh,k}+W_{ijhk,l}=0. Combining with the above, the Ricci identity gives \displaystyle\left\langle\Delta W,W\right\rangle=W_{ijkl,ss}W_{ijkl}=W_{ijkl}(W_{ijsl,ks}+W_{ijks,ls}) \displaystyle=2W_{ijkl}(W_{ijks,sl}+R_{lsti}W_{tjks}+R_{lstj}W_{itks}+R_{lstk}W_{ijts}+R_{lsts}W_{ijkt}) \displaystyle=4R_{lsti}W_{ijkl}W_{tjks}+2R_{lstk}W_{ijkl}W_{ijts}+2R_{lt}W_{ijkl}W_{ijkt} \displaystyle=2R_{lt}W_{ijkl}W_{ijkt}-\alpha(W)-4\beta(W). This completes the proof of Lemma 3.1. ∎ Let S^{ij}=\frac{1}{2}\sum\limits_{k,l}{(W_{iklj}+W_{ilkj})e_{k}\odot e_{l}}. Then S^{ij} is a trace-less symmetric tensor since Weyl tensor is trace-free. Therefore, we can compute: \displaystyle\sum\limits_{i,j} \displaystyle{<\mathring{R}(S^{ij}),S^{ij}>}=\sum\limits_{i,j}{<\bar{R}(S^{ij}),S^{ij}>} \displaystyle= \displaystyle\frac{1}{2}\sum\limits_{i,j,k,l,s,t}{(W_{iklj}+W_{ilkj})(W_{istj}+W_{itsj})(R_{kstl}+R_{ktsl})} (3.2) \displaystyle= \displaystyle\frac{1}{2}(W_{iklj}W_{istj}R_{kstl}+W_{iklj}W_{istj}R_{ktsl}+W_{iklj}W_{itsj}R_{kstl}+W_{iklj}W_{itsj}R_{ktsl} \displaystyle+W_{ilkj}W_{istj}R_{kstl}+W_{ilkj}W_{istj}R_{ktsl}+W_{ilkj}W_{itsj}R_{kstl}+W_{ilkj}W_{itsj}R_{ktsl}) \displaystyle= \displaystyle 2W_{iklj}W_{istj}R_{kstl}+2W_{iklj}W_{istj}R_{ktsl} \displaystyle= \displaystyle\frac{1}{2}\alpha(W)-4\beta(W) where the last equality follows from the first Bianchi identity, which gives \displaystyle W_{iklj}W_{istj}R_{ktsl}=W_{iklj}W_{istj}(R_{klst}+R_{kstl})=W_{iklj}W_{istj}R_{klst}-\beta(W) and \displaystyle\alpha(W)=W_{ijkl}W_{ijst}R_{klst}=(W_{ilkj}+W_{ikjl})(W_{itsj}+W_{isjt})R_{klst} \displaystyle=W_{ilkj}W_{itsj}R_{klst}+W_{ilkj}W_{isjt}R_{klst}+W_{ikjl}W_{itsj}R_{klst}+W_{ikjl}W_{isjt}R_{klst} \displaystyle=4W_{iklj}W_{istj}R_{klst}. In order to derive a Bochner formula for Riemannian manifolds with harmonic Weyl curvature, we require the following proposition, which is taken from [ 8 ] . Proposition 3.3 ( [ 8 ] ) . Let T and \mathring{R} denote a smooth algebraic curvature tensor and a curvature operator of the second kind on a Riemannian manifold (M,g) of dimension n\geq 3 , respectively. Then <\mathring{R}(T^{S_{0}^{2}}),T^{S_{0}^{2}}>=\frac{2n+32}{n}R_{st}T_{sjkl}T_{tjkl}-5\alpha(T)+4\beta(T)-\frac{16}{n^{2}}s\left|T\right|^{2}. With the aforementioned preparations, we arrive at the following Bochner formula. Proposition 3.4 . Let M be an n(\geq 4) -dimensional complete Riemannian manifold with harmonic Weyl curvature. Then 3\left\langle\Delta W,W\right\rangle=\left\langle\mathring{R}\,(W^{S_{0}^{2}}),W^{S_{0}^{2}}\right\rangle+\frac{4(n-8)}{n}R_{lt}W_{ijkl}W_{ijkt}+4\sum\limits_{i,j}{<\mathring{R}(S^{ij}),S^{ij}>}+\frac{16}{n^{2}}s|W|^{2}. Proof. From Proposition 3.3 , we have \displaystyle<\mathring{R}(W^{S_{0}^{2}}),W^{S_{0}^{2}}>=\frac{2n+32}{n}{R_{st}}{W_{sjkl}}{W_{tjkl}}-5\alpha(W)+4\beta(W)-\frac{16}{n^{2}}s{\left|W\right|}^{2}. Combining this with ( 3.1 ) and ( 3 ), we obtain \displaystyle 3\left\langle\Delta W,W\right\rangle \displaystyle=\left\langle\mathring{R}\,(W^{S_{0}^{2}}),W^{S_{0}^{2}}\right\rangle+\frac{4(n-8)}{n}R_{lt}W_{ijkl}W_{ijkt}+2\alpha(W)-16\beta(W)+\frac{16}{n^{2}}s|W|^{2} \displaystyle=\left\langle\mathring{R}\,(W^{S_{0}^{2}}),W^{S_{0}^{2}}\right\rangle+\frac{4(n-8)}{n}R_{lt}W_{ijkl}W_{ijkt}+4\sum\limits_{i,j}{<\mathring{R}(S^{ij}),S^{ij}>}+\frac{16}{n^{2}}s|W|^{2}. This completes the proof of Proposition 3.4 . ∎ Let \{S^{\alpha}\} is an orthonormal eigenbasis for \mathring{R} with corresponding eigenvalues \{\lambda_{\alpha}\} . Since S^{ij} is a traceless symmetric (0,2) -tensor, we have \sum_{i,j}\left\langle\mathring{R}(S^{ij}),S^{ij}\right\rangle=\sum_{\alpha}\left[\lambda_{\alpha}\sum_{i,j}{(S_{\alpha}^{ij})^{2}}\right], where S_{\alpha}^{ij} is the component of S^{ij} under the basis \{S^{\alpha}\} . Hence Proposition 3.4 can be written as (3.3) 3\left\langle\Delta W,W\right\rangle=\sum_{\alpha}{\lambda_{\alpha}\left|S^{\alpha}W\right|^{2}}+\frac{4(n-8)}{n}R_{lt}W_{ijkl}W_{ijkt}+4\sum_{\alpha}\left[\lambda_{\alpha}\sum_{i,j}{(S_{\alpha}^{ij})^{2}}\right]+\frac{16}{n^{2}}s|W|^{2}. Based on the calculations presented in [ 8 ] , it is shown that \displaystyle\sum\limits_{\alpha=1}^{N}{\left|S^{\alpha}W\right|^{2}}=\frac{2(n^{2}+n-8)}{n}{\left|W\right|^{2}}, and \left|S^{\alpha}W\right|^{2}\leq\frac{8(n-2)}{n}\left|W\right|^{2}. Therefore, (3.4) \sum\limits_{\alpha}{{\lambda}_{\alpha}\left|S^{\alpha}W\right|^{2}}\geq[\mathring{R},\frac{8(n-2)}{n},\frac{2(n^{2}+n-8)}{n}]|W|^{2}. According to [ 26 , Example 3.2] , the scalar curvature s of M satisfies (3.5) \displaystyle s\geq\frac{2n}{n+2}[\mathring{R},1,\frac{(n-1)(n+2)}{2}], and from [ 26 , Lemma 3.14] , the Ricci curvature Ric of M satisfies \displaystyle\operatorname{Ric}\geq\frac{n-1}{n+1}\left[\mathring{R},1,n\right]+\frac{1}{n(n+1)}s. Combining ( 3.5 ) with Lemma 2.4 (c), we obtain \displaystyle\operatorname{Ric} \displaystyle\geq\frac{n-1}{n+1}\left[\mathring{R},1,n\right]+\frac{2}{(n+1)(n+2)}[\mathring{R},1,\frac{(n-1)(n+2)}{2}] (3.6) \displaystyle\geq\left[\mathring{R},\frac{n}{n+2},n-1\right]. Furthermore, applying ( 3 ) to the curvature tensor R_{ijkl}=\delta_{ik}\delta_{jl}-\delta_{il}\delta_{jk} gives \sum_{\alpha}{\sum_{i,j}{(S_{\alpha}^{ij})^{2}}}=3|W|^{2}. From the weight principle, it follows that (3.7) \displaystyle\sum_{\alpha}\left[\lambda_{\alpha}\sum_{i,j}{(S_{\alpha}^{ij})^{2}}\right]\geq 3\left[\mathring{R},1,1\right]\left|W\right|^{2}. Proposition 3.5 . Let M be an n(\geq 8) -dimensional complete Riemannian manifold with harmonic Weyl curvature. Then 3\left\langle\Delta W,W\right\rangle\geq[\mathring{R},\frac{8(3n-1)}{n+2},6(n-1)]\left|W\right|^{2}. Proof. Combining Proposition 3.4 with ( 3.5 ),( 3.6 ),( 3.7 ) and Lemma 2.4 yields \displaystyle 3 \displaystyle\left\langle\Delta W,W\right\rangle=\left\langle\mathring{R}\,(W^{S_{0}^{2}}),W^{S_{0}^{2}}\right\rangle+\frac{4(n-8)}{n}R_{lt}W_{ijkl}W_{ijkt}+4\sum\limits_{i,j}{<\mathring{R}(S^{ij}),S^{ij}>}+\frac{16}{n^{2}}s|W|^{2} \displaystyle\geq \displaystyle[\mathring{R},\frac{8(n-2)}{n},\frac{2(n^{2}+n-8)}{n}]\left|W\right|^{2}+\frac{4(n-8)}{n}[\mathring{R},\frac{n}{n+2},n-1]\left|W\right|^{2}+12[R,1,1]\left|W\right|^{2} \displaystyle+\frac{32}{n(n+2)}[R,1,\frac{(n-1)(n+2)}{2}]\left|W\right|^{2} \displaystyle\geq \displaystyle[\mathring{R},\frac{8(n-2)}{n}+\frac{4(n-8)}{n+2}+12+\frac{32}{n(n+2)},\frac{2(n^{2}+n-8)}{n}+\frac{4(n-8)(n-1)}{n}+12+\frac{16(n-1)}{n}]\left|W\right|^{2} \displaystyle= \displaystyle[\mathring{R},\frac{8(3n-1)}{n+2},6(n-1)]\left|W\right|^{2}. ∎ Proof of Theorem A . If \mathring{R} is \frac{3(n-1)(n+2)}{4(3n-1)} -nonnegative, then by using Proposition 3.5 and Lemma 2.3 , we have \frac{1}{2}\Delta\left|W\right|^{2}=\left|\nabla W\right|^{2}+\left\langle\Delta W,W\right\rangle\geq 0. By the maximum principle, from the above we have \nabla W=0. According to Derdziński and Roter’s result [ 11 ] that Riemannian manifolds with \nabla W=0 are either locally conformally flat or locally symmetric, we see that (M,g) is either locally conformally flat or locally symmetric. By Theorem 1.8 in [ 22 ] , (M,g) is either locally irreducible or flat. If (M,g) is locally conformally flat, then by the classification theorem in [ 6 , 32 ] and the fact that (M,g) is locally irreducible, it follows that (M,g) is globally conformally equivalent to a space of positive constant curvature or is isometric to a flat manifold. If (M,g) is locally symmetric, then it has parallel Ricci curvature. Since (M,g) is locally irreducible, it is an Einstein manifold. Results from [ 8 , 15 ] imply that M is necessarily of constant curvature. Therefore, (M,g) is a space of constant curvature. ∎ 4. The case for n=5 and n=7 In this section, we prove Theorem B . First, we establish an upper bound for the Ricci curvature in terms of the scalar curvature and the curvature operator of the second kind. Proposition 4.1 . Let R be an algebraic curvature tensor on a Euclidean vector space V of dimension n\geq 3 . If R has \frac{(n-1)(n-2)}{2} -nonnegative curvature operator of the second kind, then the Ricci curvature of R satisfies Ric\leq\frac{s}{2}. Remark 4.2 . This proposition follows from the proof of [ 20 , Theorem 1.6] . Proof. Let \{e_{1},...,e_{n}\} be an orthonormal basis of V and define: {\psi}_{kl}=\frac{1}{\sqrt{2}}e_{k}\odot e_{l},\quad\text{for}\quad 2\leq k0 . The expression for f can be rewritten as \displaystyle f=2\left[s(a_{1}^{2}+a_{2}^{2}+a_{3}^{2})-12(a_{1}^{3}+a_{2}^{3}+a_{3}^{3})\right] \displaystyle=2\left[s\sum\limits_{i=1}^{3}{(\frac{s}{6}-a_{i}-\frac{s}{6})^{2}}+12\sum\limits_{i=1}^{3}{(\frac{s}{6}-a_{i}-\frac{s}{6})^{3}}\right] \displaystyle=2\left[12\sum\limits_{i=1}^{3}{(\frac{s}{6}-a_{i})^{3}}-5s\sum\limits_{i=1}^{3}{(\frac{s}{6}-a_{i})^{2}}+\frac{s^{3}}{4}\right] \displaystyle=\frac{125}{72}s^{3}\left[\sum\limits_{i=1}^{3}{[\frac{12}{5s}(\frac{s}{6}-a_{i})]^{3}}-\sum\limits_{i=1}^{3}{[\frac{12}{5s}(\frac{s}{6}-a_{i})]^{2}}\right]+\frac{s^{3}}{2}. Since \frac{12}{5s}(\frac{s}{6}-a_{i})\geq 0\quad\text{and}\quad\sum\limits_{i}{\frac{12}{5s}(\frac{s}{6}-a_{i})}=\frac{6}{5}, it follows from [ 14 , Lemma 3.3] that the minimal points of f are (\frac{12}{5s}(\frac{s}{6}-a_{1}),\frac{12}{5s}(\frac{s}{6}-a_{2}),\frac{12}{5s}(\frac{s}{6}-a_{3}))=(\frac{2}{5},\frac{2}{5},\frac{2}{5}) and (\frac{12}{5s}(\frac{s}{6}-a_{1}),\frac{12}{5s}(\frac{s}{6}-a_{2}),\frac{12}{5s}(\frac{s}{6}-a_{3}))=(\frac{3}{5},\frac{3}{5},0), which correspond to (a_{1},a_{2},a_{3})=(0,0,0)\quad\text{and}\quad(a_{1},a_{2},a_{3})=(-\frac{s}{12},-\frac{s}{12},\frac{s}{6}), respectively. Consequently, we have f\geq f(0,0,0)=f(-\frac{s}{12},-\frac{s}{12},\frac{s}{6})=0, which in this case implies \frac{1}{2}\Delta\left|W^{+}\right|^{2}=\left|\nabla W^{+}\right|^{2}+\frac{s}{2}\left|W^{+}\right|^{2}-72\det W^{+}\geq 0. By the maximum principle, we obtain \nabla W^{+}=0 , and either (a_{1},a_{2},a_{3})=(0,0,0), or (a_{1},a_{2},a_{3})=(-\frac{s}{12},-\frac{s}{12},\frac{s}{6}). Similarly, we obtain \nabla W^{-}=0 , with (b_{1},b_{2},b_{3}) being either (0,0,0) or (-\frac{s}{12},-\frac{s}{12},\frac{s}{6}). It follows that either W=0 , or W\neq 0 and \nabla W=0 –the latter case implying that M is a Kähler manifold of constant scalar curvature [ 10 ] . Together with the harmonic Weyl curvature condition, this forces the Ricci tensor to be parallel. Consequently, we have \nabla R=0 ; that is, M is locally symmetric. So when the universal cover \tilde{M} of M is irreducible, \tilde{M} is \mathbb{CP}^{2} with the Fubini–Study metric by the classification of 4 -dimensional symmetric spaces; when the universal cover \tilde{M} of M is reducible, then \Sigma_{1}^{2}\times\Sigma_{2}^{2} with the product metric, where k_{i}(i=1,2) is constant Gaussian curvature of \Sigma_{i}^{2} and k_{1}+k_{2}=\frac{s}{2}>0 . ∎ Corollary 5.1 . Let (M,g) be a 4 -dimensional compact manifold with harmonic Weyl curvature and \chi(M)\geq 0 whose the curvature operator of the second kind satisfies \lambda_{1}+\lambda_{2}+\lambda_{3}\geq-3\bar{\lambda}, then one of the following must be true: (1) M is flat. (2) M is conformal to a quotient of the round sphere (S^{4},g_{0}) . (3) M is conformal to a quotient of \mathbb{S}^{1}\times\mathbb{S}^{3} with the product metric. (4) up to rescaling, M is \mathbb{CP}^{2} with the Fubini–Study metric. (5) up to rescaling, M is isometric to a quotient of \Sigma_{1}^{2}\times\Sigma_{2}^{2} with the product metric, where k_{i}(i=1,2) is constant Gaussian curvature of \Sigma_{i}^{2} and \frac{s}{2}=k_{1}+k_{2}>0 . Proof. If W=0 and s=0 , the Chern–Gauss–Bonnet theorem implies that \mathring{Ric}=0 , i.e., M is flat. In the case of W=0 and s>0 , then [ 12 , Proposition F] implies that the second Betti number b_{2}(M)=0 , and 0\leq\chi(M)\leq 2 , i.e., the first Betti number b_{1}(M)=0 or 1 . It is derived from the Chern–Gauss–Bonnet theorem that -2\int_{M}|\mathring{Ric}|^{2}+\frac{1}{6}\int_{M}s^{2}=32{\pi}^{2}\chi(M)\geq 0,\text{i.e.},\int_{M}|\mathring{Ric}|^{2}\leq\frac{1}{6}\int_{M}s^{2}. By [ 16 , Theorem 1.7] , M is conformal to a quotient of the stand sphere \mathbb{S}^{4} or \mathbb{S}^{1}\times\mathbb{S}^{3} with the product metric. Combining Theorem C , we complete the proof of Corollary 5.1 . ∎ Remark 5.2 . For 4 -dimensional manifolds with harmonic curvature, in (2) and (3) of Corollary 5.1 , the term “conformal” can be strengthened to “isometric”. In fact, M is isometric to a quotient of the stand sphere \mathbb{S}^{4} since the stand sphere \mathbb{S}^{4} is Einstein. On the other hand, \chi(M)=0 implies that \int_{M}|\mathring{Ric}|^{2}=\frac{1}{12}Y^{2}(M,[\tilde{g}]) since -2\int_{M}|\mathring{Ric}|^{2}+\frac{1}{6}\int_{M}s^{2} is conformal invariant. By [ 16 , Theorem 1.6] , M is isometric to a quotient of \mathbb{S}^{1}\times\mathbb{S}^{3} with the product metric. Theorem 5.3 . Let (M,g) be a 4 -dimensional compact manifold with harmonic Weyl curvature and 4\frac{1}{2} -nonnegative curvature operator of the second kind. Then one of the following must be true: (1) M is flat. (2) M is conformal to a quotient of the round sphere (\mathbb{S}^{4},g_{0}) . (3) M is conformal to a quotient of \mathbb{R}^{1}\times\mathbb{S}^{3} with the product metric. (4) up to rescaling, M is isometric to \mathbb{CP}^{2} with the Fubini–Study metric. Remark 5.4 . For general 4 -dimensional manifolds, [ 20 , Theorem 1.4] yields a similar but weaker result than the one stated in this theorem. For 4 -dimensional manifolds with harmonic curvature, in (2) and (3) of Theorem 5.3 , the term “conformal” can be strengthened to “isometric”. This can be obtained similarly to the latter part of the proof of Corollary 5.1 that M is isometric to a quotient of either the round sphere \mathbb{S}^{4} or \mathbb{R}^{1}\times\mathbb{S}^{3} . Proof. Suppose that {\lambda}_{1}\leq{\lambda}_{2}\leq\cdots\leq{\lambda}_{9} are the eigenvalues of \mathring{R}\, and \bar{\lambda} is their average. Then 4\frac{1}{2} -nonnegative curvature operator of the second kind \mathring{R}\, satisfies \displaystyle{\lambda}_{1}+{\lambda}_{2}+{\lambda}_{3}\geq-{\lambda}_{4}-\frac{1}{2}{\lambda}_{5} \displaystyle\geq-\frac{\frac{1}{2}{\lambda}_{5}+{\lambda}_{6}+{\lambda}_{7}+{\lambda}_{8}+{\lambda}_{9}}{3} \displaystyle\geq-\frac{{\lambda}_{1}+{\lambda}_{2}+{\lambda}_{3}+{\lambda}_{4}+{\lambda}_{5}+{\lambda}_{6}+{\lambda}_{7}+{\lambda}_{8}+{\lambda}_{9}}{3} \displaystyle=-3\bar{\lambda}. Combining Theorem C , M is either flat, conformal flat, or is isometric to \mathbb{CP}^{2} with the Fubini–Study metric because \mathbb{S}^{2}\times\mathbb{S}^{2} has 4\frac{1}{2} -negative curvature operator of the second kind. By [ 20 , Theorem 1.6] , M has nonnegative Ricci curvature. Then, applying [ 6 , Theorem A] and [ 32 , Theorem 1] , we conclude that M is conformal to a quotient of either the round sphere \mathbb{S}^{4} or \mathbb{R}^{1}\times\mathbb{S}^{3} . This completes the proof of Theorem 5.3 . ∎