Citation: Wei Xue, Pengcheng Wan, Qiao Li, Ping Zhong, Gaohang Yu, Tao Tao. An online conjugate gradient algorithm for large-scale data analysis in machine learning[J]. AIMS Mathematics, 2021, 6(2): 1515-1537. doi: 10.3934/math.2021092
[1] | Mohd Danish Siddiqi, Fatemah Mofarreh . Schur-type inequality for solitonic hypersurfaces in (k,μ)-contact metric manifolds. AIMS Mathematics, 2024, 9(12): 36069-36081. doi: 10.3934/math.20241711 |
[2] | Hamed Faraji, Shahroud Azami, Ghodratallah Fasihi-Ramandi . h-Almost Ricci solitons with concurrent potential fields. AIMS Mathematics, 2020, 5(5): 4220-4228. doi: 10.3934/math.2020269 |
[3] | Rabia Cakan Akpınar, Esen Kemer Kansu . Metallic deformation on para-Sasaki-like para-Norden manifold. AIMS Mathematics, 2024, 9(7): 19125-19136. doi: 10.3934/math.2024932 |
[4] | Yusuf Dogru . η-Ricci-Bourguignon solitons with a semi-symmetric metric and semi-symmetric non-metric connection. AIMS Mathematics, 2023, 8(5): 11943-11952. doi: 10.3934/math.2023603 |
[5] | Yanlin Li, Dipen Ganguly, Santu Dey, Arindam Bhattacharyya . Conformal η-Ricci solitons within the framework of indefinite Kenmotsu manifolds. AIMS Mathematics, 2022, 7(4): 5408-5430. doi: 10.3934/math.2022300 |
[6] | Yanlin Li, Aydin Gezer, Erkan Karakaş . Some notes on the tangent bundle with a Ricci quarter-symmetric metric connection. AIMS Mathematics, 2023, 8(8): 17335-17353. doi: 10.3934/math.2023886 |
[7] | Shahroud Azami, Mehdi Jafari, Nargis Jamal, Abdul Haseeb . Hyperbolic Ricci solitons on perfect fluid spacetimes. AIMS Mathematics, 2024, 9(7): 18929-18943. doi: 10.3934/math.2024921 |
[8] | Rajesh Kumar, Sameh Shenawy, Lalnunenga Colney, Nasser Bin Turki . Certain results on tangent bundle endowed with generalized Tanaka Webster connection (GTWC) on Kenmotsu manifolds. AIMS Mathematics, 2024, 9(11): 30364-30383. doi: 10.3934/math.20241465 |
[9] | Fahad Sikander, Tanveer Fatima, Sharief Deshmukh, Ayman Elsharkawy . Curvature analysis of concircular trajectories in doubly warped product manifolds. AIMS Mathematics, 2024, 9(8): 21940-21951. doi: 10.3934/math.20241066 |
[10] | Abdul Haseeb, Fatemah Mofarreh, Sudhakar Kumar Chaubey, Rajendra Prasad . A study of ∗-Ricci–Yamabe solitons on LP-Kenmotsu manifolds. AIMS Mathematics, 2024, 9(8): 22532-22546. doi: 10.3934/math.20241096 |
In his famous paper on Ricci flows [1], Richard S. Hamilton introduced Ricci solitons, which have played a crucial role in deepening our comprehension of the Ricci flow's long-term dynamics and its relationship with geometric structures. Ricci solitons represent self-similar solutions that provide valuable perspectives on the geometry of manifolds.
A Ricci soliton on a Riemannian manifold (N,g) is defined by the existence of a vector field η and a constant κ such that the following equation holds:
Ric+12Lηg=κg, | (1.1) |
where Lηg denotes the Lie derivative of the metric g in the direction of η, and Ric represents the Ricci curvature of (N,g). In this context, (N,g,η,κ) is referred to as a Ricci soliton, with η known as the potential vector field and κ as the Ricci soliton constant.
In [2], S. Pigola introduced the concept of an almost Ricci soliton by allowing the Ricci soliton constant κ to be a function. More specifically, a quadruple (N,g,η,κ) is defined as an almost Ricci soliton (ARS) if Eq (1.1) holds, where κ is a smooth function on N, referred to as the potential function of the ARS.
If the soliton potential vector field η is Killing, then the ARS (N,g,η,κ) is said to be trivial. We also say that the ARS (N,g,η,κ) is parallel if its potential vector field η is a parallel vector field.
An almost Ricci soliton (N,g,η,κ) is categorized as steady under the condition κ=0, as shrinking under the condition κ>0, and as expanding under the condition κ<0. When the potential vector field η is gradient, i.e, η=Df for some smooth function f on N, the almost Ricci soliton (N,g,f,κ) is called a gradient. Equation (1.1) becomes
Ric+Hf=κg, | (1.2) |
where Hf is the Hessian of the function f.
The study of almost Ricci solitons (ARS) provides a natural generalization of Ricci solitons by allowing the soliton constant κ to vary as a smooth function rather than remaining constant. These structures naturally appear in the study of Ricci flow, where they describe shapes of spaces that evolve in a self-similar way over time. Recently, researchers have paid more attention to compact ARS spaces because they often show strong and rigid geometric behavior under certain curvature conditions.
In [3], Corollary 1 and Theorem 2, it was shown that in dimension ≥3, if a compact ARS has either constant scalar curvature or a nontrivial conformal vector field, then the space must be isometric to a standard Euclidean sphere. This extends earlier results about Ricci solitons to the more general ARS case, showing that symmetry and compactness strongly influence the geometry.
In [4], it has been shown that if a nontrivial ARS (N,g,η,κ) has constant scalar curvature, then the vector field η is gradient, and the space (N,g) must again be a sphere. This connects ARS spaces with classical gradient Ricci solitons, suggesting that many of the same rigid properties still apply even in this more general setting.
It is also known (see [5]) that all compact Ricci solitons are gradient solitons. While this does not always hold for ARS (because κ is not constant), adding the condition of constant scalar curvature forces the ARS to behave like a gradient soliton.
In two dimensions, Hamilton showed in [6] that all compact Ricci solitons are trivial. However, [3] gave an example of a nontrivial compact two-dimensional almost Ricci soliton, proving that ARS allow for more complex geometry than Ricci solitons, especially in low dimensions.
For a more detailed study of these topics, including classification results and more examples, readers can refer to the works (see [7,8,9,10,11,12,13]).
Finally, it is worth noting that many researchers have effectively applied soliton theory across various scientific and engineering disciplines. They have also devised several powerful techniques for obtaining analytic solutions, including the inverse scattering transform [14] and Hirota's bilinear method [15], among others.
The present work presents several sufficient conditions for a compact almost Ricci soliton to be isometric to a sphere and other conditions under which it becomes trivial or parallel.
The organization of this paper is as follows: Section 2 contains the fundamental concepts, presenting an overview of the essential principles and key equations within almost Ricci soliton theory. Section 3 determines the sufficient conditions for an almost Ricci soliton on a compact Riemannian manifold to be trivial or parallel. Theorem 3.6 states that given a non-parallel almost Ricci soliton (N,g,η,κ) on an n-dimensional compact and connected Riemannian manifold (N,g), where n≥3, then, (N,g) is isometric to a Euclidean sphere Sn(c) if and only if the following inequality holds:
∫N(Ric(η,η)+(n−1)g(η,Dκ))dv≤0. | (1.3) |
Theorem 3.8 demonstrates that for an almost Ricci soliton on an n-dimensional compact Riemannian manifold (N,g), with n≥2, the condition
∫N(Ric(η,η)+(n−2)g(η,Dκ))dv≤0 | (1.4) |
leads to the potential vector field being parallel. This extends Theorem 3 in [3], which shows that an almost Ricci soliton on a compact Riemannian manifold, with n≥3, is trivial (i.e, the potential vector field is Killing) if condition (1.4) is satisfied. This also generalizes Theorem 1.1 in [16], stating that a compact Ricci soliton is trivial whenever
∫N(Ric(η,η))dv≤0. |
On an n-dimensional Riemannian manifold (N,g), where n≥2, the Riemannian curvature tensor R is given by:
R(U,V)Z=D[U,V]Z−DUDVZ+DVDUZ, | (2.1) |
for all U,V,Z∈X(N), with D being the Levi-Civita connection on (N,g) and X(N) representing the collection of smooth vector fields on (N,g). For more details, see [17].
If {e1,…,en} is a local orthonormal frame on the manifold (N,g), the divergence of a vector field U∈X(N) is described by the equation:
div(U)=n∑i=1g(DDeieiU,ei). |
Similarly, the Laplacian of U is given by:
ΔU=n∑i=1DeiDeiU−DDeieiU. |
The divergence of the (1,1)-tensor ϕ on (N,g) is expressed as:
div(ϕ)=n∑i=1(Deiϕ)(ei). |
The gradient of a smooth function f on N is the vector field Df characterized by:
g(Df,U)=U(f), | (2.2) |
for all U∈X(N), and the Hessian of f is the symmetric (0,2) tensor field defined by"
Hf(U,V)=g(DUDf,V), | (2.3) |
for all U,V∈X(N).
The Laplacian of a smooth function f on N is defined as:
Δf=div(Df). |
Taking the trace of (1.1) yields
S+div(η)=nκ, | (2.4) |
where S is the scalar curvature of (N,g). Also, we have
div(κη)=g(η,Dκ)+nκ2−κS, | (2.5) |
div(Sη)=g(η,DS)+nκS−S2, | (2.6) |
and
div((nκ−S)η)=(nκ−S)2+η(nκ−S). | (2.7) |
For a gradient Ricci soliton (N,g,f,κ), Equation (2.4) becomes
S+Δf=nκ. | (2.8) |
Take a Riemannian manifold (N,g), and η∈X(N). Let θη be the 1-form dual to η (i.e, θη(U)=g(U,η),U∈X(N)), then the Lie derivative Lη can be written as
Lηg(U,V)+dθη(U,V)=2g(DUη,V), | (2.9) |
for all U,V∈X(N).
Now, consider B a symmetric tensor field, and ϕ a skew-symmetric tensor field on (N,g) with the following equations:
Lηg(U,V)=2g(B(U),V), | (2.10) |
and
dθη(U,V)=2g(ϕ(U),V), | (2.11) |
where U,V∈X(N). We get from (2.9) that
DUη=B(U)+ϕ(U), | (2.12) |
for each U∈X(N).
A vector field η is called closed conformal if
DUη=ΨU, | (2.13) |
for each U∈X(N), where Ψ is a smooth function on (N,g).
In this subsection, we prove several key lemmas that will be crucial for the rest of the paper.
Assume (N,g,η,κ) is an almost Ricci soliton on a compact Riemannian manifold (N,g). By substituting (2.10) in (1.1), we deduce that
B=κI−Q, | (3.1) |
where Q is the Ricci operator, that is, the operator satisfying
Ric(U,V)=g(Q(U),V), |
for all U,V∈X(N). Also, by substituting (3.1) in (2.12), it follows that
DUη=κU−Q(U)+ϕ(U), | (3.2) |
for all U∈X(N).
The following lemma gives the value of the Ricci curvature in the direction of η.
Lemma 3.1. Let (N,g,η,κ) be an almost Ricci soliton on an n-dimensional Riemannian manifold (N,g). Then
Ric(η,V)=(1−n)V(κ)+12V(S)−g(div(ϕ),V), | (3.3) |
for all V∈X(N).
Proof. From (2.1), we obtain
R(U,V)η=U(κ)V−V(κ)U−(DUQ)(V)+(DVQ)(U)+(DUϕ)(V)−(DVϕ)(U), |
for all U,V∈X(N).
Assume {e1,…,en} is a local orthonormal frame on (N,g) such that it is parallel. Considering equation (1.1) and Corollary 54 in [17], p 88, we obtain
Ric(V,η)=n∑i=1g(R(ei,V)η,ei)=n∑i=1g(ei(κ)V−V(κ)ei−(DeiQ)(V)+(DVQ)(ei)+(Deiϕ)(V)−(DVϕ)(ei),ei)=V(κ)−nV(κ)+g(div(Q),V)−g(div(ϕ),V)=(1−n)V(κ)+12V(S)−g(div(ϕ),V). |
Remark 1. From the above lemma, we see that
Q(η)=(1−n)Dκ+12DS−div(ϕ). | (3.4) |
In the following lemma, we compute the divergence of the image of η by the Ricci tensor Q.
Lemma 3.2. Let (N,g,η,κ) be an almost Ricci soliton on an n-dimensional Riemannian manifold (N,g). Then
div(Q(η))=12g(DS,η)+κS−||Q||2. | (3.5) |
Proof. Consider {e1,…,en} as a local orthonormal frame on (N,g). By considering (3.4), we deduce
div(Q(η))=n∑i=1g(DeiQ(η),ei)=n∑i=1g((DeiQ)(η),ei)+n∑i=1g(Q(Deiη),ei)=n∑i=1g(η,(DeiQ)ei)+n∑i=1g(Q(Deiη),ei)=g(η,n∑i=1(DeiQ)ei)+n∑i=1g(Deiη,Q(ei))=g(η,12DS)+n∑i=1g(κei−Q(ei)+ϕ(ei),Q(ei))=g(η,12DS)+κn∑i=1g(ei,Q(ei))−n∑i=1g(Q(ei),Q(ei))+n∑i=1g(Q(ei),ϕ(ei))=g(η,12DS)+κS−||Q||2. |
The following formula is well known in the case of Ricci solitons. We extend it here to the case of almost Ricci solitons.
Lemma 3.3. Let (N,g) be an n-Riemannian manifold. If (N,g,η,κ) is an almost Ricci soliton on (N,g), then
||Ric−Sng||2=||Q||2−S2n. | (3.6) |
Proof. Let λ1,λ2,...,λn be the eigenvalues of the Ricci operator Q. Then
||Ric−Sng||2=n∑i=1(λi−Sn)2=n∑i=1(λ2i+S2n2−2Snλi)=n∑i=1λ2i+n∑i=1S2n2−2Snn∑i=1λi=||Q||2+S2n−2S2n=||Q||2−S2n. |
The following lemma provides a useful formula for the Laplacian of the potential vector field for an almost Ricci soliton.
Lemma 3.4. Let (N,g,η,κ) be an almost Ricci soliton on an n-Riemannian manifold (N,g). Then
Δη=(2−n)Dκ−Q(η). | (3.7) |
Proof. Consider {e1,…,en} as a local orthonormal frame on (N,g). Considering Eq (1.1) and Corollary 54 in [17], p 88, we obtain
Δη=n∑i=1DeiDeiη−DDeieiη=n∑i=1Dei(κei−Q(ei)+ϕ(ei))=n∑i=1(g(κ,ei)ei−DeiQ(ei)+Deiϕ(ei))=n∑i=1(g(κ,ei)ei−(DeiQ)(ei)+(Deiϕ)(ei))=Dκ−12DS+div(ϕ). |
Inserting (3.4) in the preceding equation yields (3.7).
Theorem 3.5. Let (N,g,η,κ) be an almost Ricci soliton on an n-dimensional compact Riemannian manifold (N,g), where n≥2. If the inequality
∫N(Ric(η,η)+n(κ2+g(η,Dκ)))dv≤0 | (3.8) |
is satisfied, then the manifold (N,g) is Ricci flat, and η is a closed conformal vector field.
Proof. By (3.3) and (3.5), we have
Ric(η,η)=(1−n)g(η,Dκ)+12g(η,DS)−g(div(ϕ),η)=(1−n)g(η,Dκ)+12g(η,DS)+||ϕ||2+div(ϕ(η))=(1−n)g(η,Dκ)+div(Q(η))−κS+||Q||2+||ϕ||2+div(ϕ(η)). |
Consequently, it follows that
Ric(η,η)+(n−1)g(η,Dκ)=div(Q(η))−κS+||Q||2+||ϕ||2+div(ϕ(η)). | (3.9) |
By substituting the value of κS from (2.5) in (3.9), we deduce that
Ric(η,η)=−ng(η,Dκ)+div(Q(η))−div(κη)−nκ2+||Q||2+||ϕ||2+div(ϕ(η)). | (3.10) |
By integrating (3.10), one obtains
∫N(Ric(η,η)+n(κ2+g(η,Dκ)))dv=∫N(||Q||2+||ϕ||2)dv. | (3.11) |
Now, it is clear that condition (3.8) in (3.11) yields Q=ϕ=0. We deduce from (2.12) that Dη=κI, which means that η is a closed conformal vector field.
Also, since Q=0 we get Ric=0, meaning that (N,g) is Ricci flat.
Theorem 3.6. Let (N,g) be an n-dimensional connected and compact Riemannian manifold, where n≥3. If (N,g,η,κ) is a non-parallel almost Ricci soliton, then (N,g) is isometric to a Euclidean sphere Sn(c) if and only if the subsequent inequality is satisfied
∫N(Ric(η,η)+(n−1)g(η,Dκ))dv≤0. | (3.12) |
Proof. From (2.6), we have
κS=1n(div(Sη)+S2−g(η,DS)). | (3.13) |
By substituting the value of κS in (3.9) and then utilizing (3.6), we obtain
Ric(η,η)+(n−1)g(η,Dκ)=||Ric−Sng||2+div(Q(η)−1ndiv(Sη)+ϕ(η))+1ng(η,DS)+||ϕ||2. | (3.14) |
On the other hand, Theorem 1 in [4] states that
∫N||Ric−Sng||2dv=n−22n∫Ng(η,DS)dv. | (3.15) |
By integrating Eq (3.14), followed by the substitution of equation (3.15) in the result, we derive
∫N(Ric(η,η)+(n−1)g(η,Dκ))dv=12∫N||Ric−Sng||2dv+∫N||ϕ||2dv. |
Taking into account the hypothesis (1.3), we deduce that Ric=Sng and ϕ=0.
Given that n≥3 and Ric=Sng, the Schur Lemma (refer to [17]) can be applied, implying that S must be constant. Also, because ϕ=0, η is a closed conformal vector field with conformal function κ−Sn, that is
DUη=(κ−Sn)U, | (3.16) |
for all U∈X(N).
Given that η is not parallel, it follows that κ≠Sn meaning that η is a non-Killing conformal vector field. From the fact that Ric=Sng and equation (3.16), we obtain
LηRic=2Sn(κ−Sn)g. | (3.17) |
This relation allows us to apply Theorem 4.2 from [18], p. 54, leading to the conclusion that (N,g) is isometric to the standard Euclidean sphere Sn(c) with radius 1√c, where c=Sn(n−1).
Conversely, assume that (N,g,η,κ) is a non-parallel almost Ricci soliton on a connected, compact Riemannian manifold (N,g), where (N,g) is isometric to a Euclidean sphere Sn(c), with c>0. In other words, the Ricci curvature is expressed as Ric=(n−1)cg, whereas the scalar curvature is defined by S=n(n−1)c. From (1.1), it follows that
12Lηg=(κ−(n−1)c)g. | (3.18) |
But, by Corollary 2 in [4], η is gradient, i.e, η=Df for some smooth function f on N. Then (3.18) becomes
Hf=(κ−(n−1)c)g. |
It necessarily follows that κ≠(n−1)c, as otherwise D2f=0, implying that Df=η is a constant vector field, indicating that η is parallel.
As demonstrated by Lemma 2.3 in [18] on page 52, Eq (3.18) shows that
Δκ=−nc(κ−(n−1)). |
Additionally, as indicated by (2.4), it follows that
Δf=nκ−S=n(κ−(n−1)). |
We deduce that Δ(κ+cf)=0. According to the well-known Hopf Lemma, this implies that κ+cf is a constant.
Consequently, Dκ=−cDf=−cη, and therefore
Ric(η,η)+(n−1)g(η,Dκ)=c(n−1)|η|2−c(n−1)|η|2=0. |
This demonstrates that the inequality (1.3) holds true as an equality.
Next, we shall give an example of an almost Ricci soliton on the Euclidean sphere Sn(c), where c>0 satisfies (1.3).
Example 1. Consider (Sn(c),⟨,⟩), the Euclidean sphere in the Euclidean space (Rn+1,⟨,⟩) of radius 1√c, where c>0.
Let ∇ and ¯∇ be the Levi-Civita connections of Sn(c) and Rn+1, respectively.
Let ˉZ∈X(Rn+1) be a constant vector field, with Z as its restriction to Sn(c).
If ψ:Sn(c)⟶Rn+1 represents the position vector field, then consider the unit normal vector field on Sn(c), defined by N=√cψ.
Write
Z=ZT+θN, | (3.19) |
where ZT is the tangential component of Z, and θ=⟨Z,N⟩.
Employing the Gauss and Weingarten formulas, we obtain
∇UZT=−√cθU, | (3.20) |
and
∇θ=√cZT. | (3.21) |
Consequently, ZT serves as a closed conformal vector field associated with the conformal function −√cθ.
Assume that (Sn(c),⟨,⟩,ZT,κ) is an almost Ricci soliton on (Sn(c),⟨,⟩). Given that for the Euclidean sphere (Sn(c), we have Ric=(n−1)cg, and from (3.20) we know that 12LZTg=−√cθg, by substituting these expressions into equation (1.1), we deduce that
−√cθI+(n−1)cI=κI, |
indicating that the function κ is given explicitely by
κ=(n−1)c−θ√c. |
Since Ric(ZT,ZT)=(n−1)c⟨ZT,ZT⟩ and ⟨ZT,Dκ⟩=−c⟨ZT,ZT⟩, it follows that
∫Sn(c)(Ric(ZT,ZT)+(n−1)⟨ZT,Dκ⟩)dv=0, |
meaning that the inequality (1.3) is satisfied as an equality.
We close this subsection with the following result, which characterizes Euclidean spheres in terms of the constancy of the scalar curvature along the integral curves of the potential vector field.
Theorem 3.7. Let (N,g) be an n-dimensional compact oriented Riemannian manifold, n≥3. If (N,g,η,κ) is a non-trivial almost Ricci soliton on (N,g), then (N,g) is isometric to a Euclidean sphere if and only if η leaves the scalar curvature S constant (i.e, S is constant along the integral curves of η).
Proof. Assume that η leaves the scalar curvature S constant. This means that g(η,DS)=0.
Substituting this in Eq (3.15), we obtain
∫N||Ric−Sng||2dv=0. |
This implies that Ric=Sng. With n≥3, Schur's Lemma implies that S must be constant. According to (1.1), we have
12Lηg=(κ−Sn)g=1n(nκ−S)g=div(η)ng. |
Given that (N,g,η,κ) is non-trivial, it follows that div(η)≠0, indicating that η is a non-trivial conformal vector field. It follows that η is a non-homothetic conformal vector field on a compact Einstein Riemannian manifold. Following a similar argument as in the proof of Theorem 3.6 above, and applying Theorem 4.2 from [18], p. 54, we conclude that (N,g) is isometric to a Euclidean sphere. The converse is trivial, since if (N,g) is isometric to a Euclidean sphere, then the scalar curvature is constant.
The next result extends Theorem 3 from [3], shown when n≥3 and with the conclusion that η is a Killing vector field. It also generalizes Theorem 1.1 from [16], which proved for a Ricci soliton on a compact Riemannian manifold that satisfies ∫N(Ric(η,η))dv≤0.
Theorem 3.8. Let (N,g) be an n-dimensional compact Riemannian manifold, n≥2. If (N,g,η,κ) is an almost Ricci soliton on (N,g) satisfying the following inequality
∫N(Ric(η,η)+(n−2)g(η,Dκ))dv≤0, | (3.22) |
then η is a parallel vector field.
Proof. By Lemma 2 in [3], we have
12Δ|η|2=|Dη|2−Ric(η,η)−(n−2)g(η,Dκ). | (3.23) |
By integrating the above equation, we obtain
∫N(Ric(η,η)+(n−2)g(η,Dκ))dv=∫N|Dη|2dv, | (3.24) |
and from the hypothesis (3.22), it follows that Dη=0, which means that η is parallel.
Theorem 3.9. Let (N,g,η,κ) be an almost Ricci soliton on an n-dimensional compact Riemannian manifold (N,g), where n≥2. If (N,g,η,κ) satisfies the following inequality
∫N(Ric(Q(η),η)+(n−1)g(DS,Dκ))dv≤0, | (3.25) |
then η is a parallel vector field, and (N,g) is Einstein.
Proof. By (3.3) and (3.4), we have
Ric(Q(η),η)+Ric(div(ϕ),η)=12Ric(η,DS)−(n−1)Ric(η,Dκ)=12(−(n−1)g(DS,Dκ)+12|DS|2−g(div(ϕ,DS))−(n−1)(−(n−1)g(Dκ,η)+12g(DS,Sκ)−g(div(ϕ),Dκ))=−(n−1)g(DS,Dκ)+12div(ϕ(DS))+14|DS|2+(n−1)2|Dκ|2−(n−1)div(ϕ(Dκ)). |
So, we have
Ric(Q(η),η)+Ric(div(ϕ),η)=−(n−1)g(DS,Dκ)+12div(ϕ(DS))+14|DS|2+(n−1)2|Dκ|2−(n−1)div(ϕ(Dκ)). | (3.26) |
Also, from (3.3), one obtains
Ric(div(ϕ),η)=(n−1)div(ϕ(Dκ)−12div(ϕ(DS))−|div(ϕ)|2. | (3.27) |
Substituting (3.27) in (3.26) and integrating the resulting equation yields
∫N(Ric(Q(η),η)+(n−1)g(DS,Dκ))dv=∫N(14|DS|2+(n−1)2|Dκ|2+|div(ϕ)|2)dv. | (3.28) |
Assuming the hypothesis (3.25), it follows that the scalar curvature S and the potential function κ are constant, and div(ϕ)=0. From equation (2.4), it is deduced that nκ=S.
From (3.5), it follows
||Q||2=κS. |
It follows that Eq (3.6) becomes
||Ric−Sng||2=0. |
This implies that (N,g) is an Einstein manifold, and that η is a Killing vector field.
Furthermore, given that ∫NRic(η,η)dv=0 from (3.28), then (3.24) implies that
∫N|Dη|2dv=0, |
which means that Dη=0, and so η is a parallel vector field.
Remark According to [19], it is proved that an n-dimensional compact almost Ricci soliton (N,g,η,κ), where n≥3, becomes a trivial Ricci soliton precisely when nκ−S remains constant along the integral curve of the potential field η. This condition, essentially, is equivalent to η.div(η)=0. We aim to extend this result as follows.
Theorem 3.10. An almost Ricci soliton (N,g,η,κ) on an n-dimensional compact Riemannian manifold with n≥3 is trivial if and only if the vector field η is incompressible, meaning that div(η)=0.
Proof. Assume that (N,g,η,κ) is trivial. Then, η is a Killing vector field, meaning that Lηg=0. Consequently, div(η)=0.
Conversely, by assuming that div(η)=0, we deduce that nκ=S. By integrating Eq (2.6), we obtain
∫Ng(η,DS)dv=0. |
Substituting this in Eq (3.15), it follows that
∫N||Ric−Sng||2dv=0. |
This implies that Ric=Sng. With n≥3, Schur's Lemma indicates that S must be constant. Since nκ=S, it follows that κ is also constant. Therefore, (N,g,η,κ) is trivial.
Upon reviewing the previous outcome, we derive the following result:
Corollary 1. A compact almost Ricci soliton of dimension n≥3 with an affine potential vector field is trivial.
Proof. By [20], any affine vector field on a compact Riemannian manifold is incompressible.
Theorem 3.11. Let (N,g,η,κ) be an almost Ricci soliton on an n-dimensional compact Riemannian manifold (N,g), where n≥2. If (N,g,η,κ) satisfies the following inequality
∫N(g(η,DS)−2g(η,Dκ))dv≤0, | (3.29) |
then, η is a Killing vector field, and (N,g,η,κ) is trivial.
Proof. By taking the inner product of (3.7) with η and subsequently integrating, it follows
∫N(Ric(η,η)+g(η,Δη)+(n−2)g(η,Dκ))dv=0. | (3.30) |
According to Proposition 5.10 in [21], it follows that
∫N(Ric(η,η)+g(η,Δη)+12|Lηg|2−(div(η))2))dv=0. | (3.31) |
Using Eq (2.12), we obtain
|Dη|2=|B|2+|ϕ|2, |
and
tr(Dη)2=|B|2−|ϕ|2, |
where tr(Dη)2 represents the trace of (Dη)2.
Referencing Lemma 5.9 in [21], we have
|Lηg|2=4|B|2. | (3.32) |
Utilizing (3.31) and considering Eqs (2.7) and (3.32), we obtain
∫N(g(η,DS)−2g(η,Dκ))dv=2∫N|B|2dv. | (3.33) |
Assuming that ∫N(g(η,S)−2g(η,Dκ))dv≤0, we deduce that B=0, implying that Ric=κg. According to (1.1), η becomes a Killing vector field, and the almost Ricci soliton is trivial.
In this paper, we extend and refine recent works about Ricci solitons and almost Ricci solitons by using methods based on integral inequalities that involve Ricci curvature and the potential vector field. Unlike previous studies that often focused on specific curvature assumptions or geometric structures, we use integral conditions to set clear criteria for when a compact Riemannian manifold with an almost Ricci soliton is isometric to a sphere. We also explore the potential vector field in more detail, identifying conditions where it becomes either Killing or parallel, making the almost Ricci soliton trivial. By focusing on both the geometry and the properties of the vector field, our work not only broadens the scope of earlier results but also provides a more unified approach for studying solitons with fewer assumptions. Our methods and findings show how effective integral techniques are in analyzing Ricci-type structures and contribute to the ongoing development of Ricci soliton theory. We expect that future research will focus on exploring the properties of almost Ricci solitons in important contexts, like spacelike submanifolds of Lorentzian manifolds, especially within generalized Robertson-Walker (GRW) spacetimes and general Lorentzian warped products.
The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.
The authors would like to extend their sincere appreciation to Ongoing Research Funding Project, (ORF-2025-824), King Saud University, Riyadh, Saudi Arabia.
The authors declare that they have no conflicts of interest.
Conceptualization, M.G.; Investigation, N.A. and M.G.; Methodology, N.A. and M.G.; Resources, N.A.; Validation, M.G.; Writing-original draft, N.A.; Writing-review and editing, N.A. and M.G.
[1] |
J. Barzilai, J. M. Borwein, Two-point step size gradient methods, IMA J. Numer. Anal., 8 (1988), 141-148. doi: 10.1093/imanum/8.1.141
![]() |
[2] | E. Bisong, Batch vs. online larning, Building Machine Learning and Deep Learning Models on Google Cloud Platform, 2019. |
[3] |
L. Bottou, F. E. Curtis, J. Nocedal, Optimization methods for large-scale machine learning, SIAM Rev., 60 (2018), 223-311. doi: 10.1137/16M1080173
![]() |
[4] | Y. H. Dai, Y. Yuan, Nonlinear conjugate gradient methods, Shanghai: Shanghai Scientific Technical Publishers, 2000. |
[5] |
D. Davis, B. Grimmer, Proximally guided stochastic subgradient method for nonsmooth, nonconvex problems, SIAM J. Optim., 29 (2019), 1908-1930. doi: 10.1137/17M1151031
![]() |
[6] |
R. Dehghani, N. Bidabadi, H. Fahs, M. M. Hosseini, A conjugate gradient method based on a modified secant relation for unconstrained optimization, Numer. Funct. Anal. Optim., 41 (2020), 621-634. doi: 10.1080/01630563.2019.1669641
![]() |
[7] |
P. Faramarzi, K. Amini, A modified spectral conjugate gradient method with global convergence, J. Optim. Theory Appl., 182 (2019), 667-690. doi: 10.1007/s10957-019-01527-6
![]() |
[8] |
R. Fletcher, C. M. Reeves, Function minimization by conjugate gradients, Comput. J., 7 (1964), 149-154. doi: 10.1093/comjnl/7.2.149
![]() |
[9] |
J. C. Gilbert, J. Nocedal, Global convergence properties of conjugate gradient methods for optimization, SIAM J. Optim., 2 (1992), 21-42. doi: 10.1137/0802003
![]() |
[10] |
W. W. Hager, H. Zhang, Algorithm 851: CG DESCENT, a conjugate gradient method with guaranteed descent, ACM Trans. Math. Software, 32 (2006), 113-137. doi: 10.1145/1132973.1132979
![]() |
[11] |
A. S. Halilu, M. Y. Waziri, Y. B. Musa, Inexact double step length method for solving systems of nonlinear equations, Stat. Optim. Inf. Comput., 8 (2020), 165-174. doi: 10.19139/soic-2310-5070-532
![]() |
[12] |
H. Jiang, P. Wilford, A stochastic conjugate gradient method for the approximation of functions, J. Comput. Appl. Math., 236 (2012), 2529-2544. doi: 10.1016/j.cam.2011.12.012
![]() |
[13] |
X. Jiang, J. Jian, Improved Fletcher-Reeves and Dai-Yuan conjugate gradient methods with the strong Wolfe line search, J. Comput. Appl. Math., 348 (2019), 525-534. doi: 10.1016/j.cam.2018.09.012
![]() |
[14] | X. B. Jin, X. Y. Zhang, K. Huang, G. G. Geng, Stochastic conjugate gradient algorithm with variance reduction, IEEE Trans. Neural Networks Learn. Syst., 30 (2018), 1360-1369. |
[15] | R. Johnson, T. Zhang, Accelerating stochastic gradient descent using predictive variance reduction, Advances in Neural Information Processing Systems, 2013. |
[16] |
X. L. Li, Preconditioned stochastic gradient descent, IEEE Trans. Neural Networks Learn. Syst., 29 (2018), 1454-1466. doi: 10.1109/TNNLS.2017.2672978
![]() |
[17] |
Y. Liu, X. Wang, T. Guo, A linearly convergent stochastic recursive gradient method for convex optimization, Optim. Lett., 2020. Doi: 10.1007/s11590-020-01550-x. doi: 10.1007/s11590-020-01550-x
![]() |
[18] |
M. Lotfi, S. M. Hosseini, An efficient Dai-Liao type conjugate gradient method by reformulating the CG parameter in the search direction equation, J. Comput. Appl. Math., 371 (2020), 112708. doi: 10.1016/j.cam.2019.112708
![]() |
[19] | S. Mandt, M. D. Hoffman, D. M. Blei, Stochastic gradient descent as approximate Bayesian inference, J. Mach. Learn. Res., 18 (2017), 4873-4907. |
[20] | P. Moritz, R. Nishihara, M. I. Jordan, A linearly-convergent stochastic L-BFGS algorithm, Proceedings of the 19th International Conference on Artificial Intelligence and Statistics, 2016. |
[21] | L. M. Nguyen, J. Liu, K. Scheinberg, M. Takáč, SARAH: A novel method for machine learning problems using stochastic recursive gradient, Proceedings of the 34th International Conference on Machine Learning, 2017. |
[22] | A. Nitanda, Accelerated stochastic gradient descent for minimizing finite sums, Proceedings of the 19th International Conference on Artificial Intelligence and Statistics, 2016. |
[23] |
H. Robbins, S. Monro, A stochastic approximation method, Ann. Math. Statist., 22 (1951), 400-407. doi: 10.1214/aoms/1177729586
![]() |
[24] | N. N. Schraudolph, T. Graepel, Combining conjugate direction methods with stochastic approximation of gradients, Proceedings of the 9th International Workshop on Artificial Intelligence and Statistics, 2003. |
[25] | G. Shao, W. Xue, G. Yu, X. Zheng, Improved SVRG for finite sum structure optimization with application to binary classification, J. Ind. Manage. Optim., 16 (2020), 2253-2266. |
[26] | C. Tan, S. Ma, Y. H. Dai, Y. Qian, Barzilai-Borwein step size for stochastic gradient descent, Advances in Neural Information Processing Systems, 2016. |
[27] | P. Toulis, E. Airoldi, J. Rennie, Statistical analysis of stochastic gradient methods for generalized linear models, Proceedings of the 31th International Conference on Machine Learning, 2014. |
[28] | V. Vapnik, The nature of statistical learning theory, New York: Springer, 1995. |
[29] |
L. Xiao, T. Zhang, A proximal stochastic gradient method with progressive variance reduction, SIAM J. Optim., 24 (2014), 2057-2075. doi: 10.1137/140961791
![]() |
[30] | Z. Xu, Y. H. Dai, A stochastic approximation frame algorithm with adaptive directions, Numer. Math. Theory Methods Appl., 1 (2008), 460-474. |
[31] | W. Xue, J. Ren, X. Zheng, Z. Liu, Y. Ling, A new DY conjugate gradient method and applications to image denoising, IEICE Trans. Inf. Syst., 101 (2018), 2984-2990. |
[32] |
Q. Zheng, X. Tian, N. Jiang, M. Yang, Layer-wise learning based stochastic gradient descent method for the optimization of deep convolutional neural network, J. Intell. Fuzzy Syst., 37 (2019), 5641-5654. doi: 10.3233/JIFS-190861
![]() |