Home Symmetries of Ricci flows
Article Open Access

Symmetries of Ricci flows

  • Enrique López EMAIL logo , Stylianos Dimas and Yuri Bozhkov
Published/Copyright: October 4, 2023

Abstract

In the present work, we find the Lie point symmetries of the Ricci flow on an n-dimensional manifold, and we introduce a method in order to reutilize these symmetries to obtain the Lie point symmetries of particular metrics. We apply this method to retrieve the Lie point symmetries of the Einstein equations (seen as a “static” Ricci flow) and of some particular types of metrics of interest, such as, on warped products of manifolds. Finally, we use the symmetries found to obtain invariant solutions of the Ricci flow for the particular families of metrics considered.

MSC 2010: 35B06; 68W30; 53E20

1 Introduction

Hamilton, in 1982 [10], introduced the Ricci flow: a nonlinear geometric evolution equation in which one starting with a smooth Riemannian manifold ( M n , g 0 ) deforms its metric using the following equation:

t g = 2 Ric g ( t ) ,

where Ric g ( t ) denotes the Ricci tensor of the metric g ( t ) and g ( 0 ) = g 0 . Similar to the heat flow of temperature distributions and other diffusion processes, the Ricci flow deforms the geometry toward more uniform ones whose limit allows us to draw topological conclusions about M n . For this reason, the Ricci flow has been studied and successfully applied to solve important manifold classification problems, such as the famous Poincaré’s conjecture and the Thurston’s geometrization conjecture. Perelman completely proved these two conjectures in 2003 [25]. There are numerous works dedicated to the study of various aspects of the Ricci flow, including the study of the black holes theory (see, e.g., the studies of Xing and Gao [28] and Bakas [2]).

It is well known that in contrast to the theory of linear differential equations – although the great progress in the past seven decades – there is still no unitary theory for nonlinear differential equations and systems. Nevertheless, various effective approaches have been developed: compact operators and operators of monotone type; topological degree theories; fixed point theory and modern variational principles; critical point theory; Morse theory and critical groups, treated from a general qualitative point of view [15,24]; Rellich and Pohozaev identities via a Noetherian approach [57]; analytical methods for constructing exact solutions of differential equations [18], etc.

We believe that the symmetry approach based on the application of Lie symmetry methods for differential equations may be fruitful in the nonlinear analysis. We note that Sophus Lie originally introduced the general theory of Lie transformation groups in the last part of the nineteenth century, specifically to study differential equations. After a period of dormancy, Lie’s ideas resurfaced, beginning with the pioneering works of L. Ovsyannikov and his collaborators. It is no exaggeration to state that the great array of applications of Lie group methods to nonlinear differential equations has been one of the great successes of the last 60 years. There is currently a vast literature in this area, and many of the basic methods and applications can be found in the studies of Olver [20,21] and Ovsiannikov [22] and the three-volume handbook by Ibragimov [13]. Some of the directions being developed include the construction of explicit symmetric solutions through symmetry reduction, the connection between symmetries and conservation laws via the Noether’s theorem for problems admitting variational structures, construction, and classification of differential invariants, invariant differential equations and variational principles, separation of variables and integrability of both linear and nonlinear partial differential equations, asymptotic behavior of solutions for long times and near blow-up using symmetric solutions, and the design of symmetry-preserving numerical algorithms.

We recall that the symmetries of a differential equation transform solutions of the equation to other solutions. One of the main benefits of this theory is that by following a completely algorithmic procedure, one is able to determine the symmetries of a differential equation or systems of differential equations. They comprise a structural property of the equation – in the essence they are equation’s DNA. The knowledge of the symmetries of an equation enables one to use them for a variety of purposes, from obtaining analytical solutions and reducing its order to finding of integrating factors and conservation laws as mentioned earlier. In fact, many, if not all, of the different empirical methods for solving ordinary differential equations (ODE) we have learned from standard courses at the undergraduate level emerge from symmetry. For instance, given a Lie point symmetry of a first-order ODE, we can immediately and explicitly obtain an integrating factor by a formula obtained by S. Lie. In this regard, we remind the words of Nail Ibragimov that

One of the most remarkable achievements of Lie was the discovery that the majority of known methods of integration of ordinary differential equations, which seemed up to that time artificial and internally disconnected, could be derived in a unified manner using the theory of groups. Moreover, Lie provided a classification of all ordinary differential equations of arbitrary order in terms of the symmetry groups they admit and thus described the whole collection of equations for which integration or lowering of the order could be effected by group-theoretical methods. However, these and other very valuable results he obtained could not for a long time be widely disseminated and remained known to only a few. It could be said that this is the state of affairs today with methods of solution of problems of mathematical physics: many of these are of a group-theoretical nature, but are presented as a result of a lucky guess [12].

All in all, symmetries play a far-reaching role in the analysis of differential equations: they allow us to recover the full picture from partial information. For this reason, it is not an uncommon practice to employ symmetries for constructing new solutions from known ones or for reducing a differential equation. At this point, it is worth mentioning that for the main bulk of calculations, we used the symbolic package SYM [9].

So far, we only have a description of the Lie point symmetries of the Ricci flow for the two-dimensional case, see Cimpoiasu and Constantinescu [8] and Wang [29]. The goal of this article is to carry out a complete group classification of the Ricci flow on n -dimensional manifolds and its use in constructing analytical solutions for particular metrics. Our central result is the following:

Theorem 1

The Lie algebra of the classical symmetries of the Ricci flow on a Riemannian manifold ( M n , g ) with n 2 is spanned by the following equations:

X 1 = t , X 2 = t t + i = 1 n j i n g i j g i j , X k + 2 = ξ k x k i = 1 n j i n g k i ξ k x j + g k j ξ k x i g i j ,

where ξ 1 , , ξ n are arbitrary smooth functions of x 1 , , x n , k = 1 , , n , and g i j are the coefficients of the metric tensor g.

This article is organized as follows: in Section 2, we give a brief introduction of all the notions utilized in the rest of the article and illustrate the key method we employ for obtaining the Lie point symmetries of the Ricci flow for particular families of metrics from the Lie algebra of Theorem 1. Our main result resides in Section 3, where we determine the Lie point symmetries of the Ricci flow for the n -dimensional case and its optimal system for its finite-dimensional sub algebra. In the next two sections, we show how this general result can be utilized to obtain the Lie point symmetries for particular metrics: we begin by retrieving the Lie symmetries of the Einstein equations for the n -dimensional case. Next, we determine the Lie point symmetries of the Ricci flow for product manifolds expressed by warped and doubly warped metrics. Then, we build invariant solutions from these symmetries. Finally, Section 6 contains comments and concluding remarks.

2 Preliminaries

In this section, we will briefly introduce the general concept of Lie point symmetry of differential equations. For a detailed approach, we recommend reading [11,13,20,22]. We restrict our attention to connected local Lie groups of symmetries, leaving aside discrete symmetries.

Let assume a system of differential equations of n -th order with p independent and q dependent variables

Δ ν ( x , u ( n ) ) = 0 , ν = 1 , , l ,

involving the derivatives of u with respect to x up to order n . Moreover, we consider all involved functions, vector fields, and tensors sufficiently smooth in their arguments. Note that the aforementioned system can be viewed as a smooth map from the jet space V × U ( n ) ( V is an open set of R p , while U ( n ) is the Cartesian product space whose coordinates represent all the partial derivatives, up to order n , of the components of the vector function u ) to R l :

Δ : V × U ( n ) R l .

If 0 R l is a regular value of the previous mapping, then

S Δ = { ( x , u ( n ) ) : Δ ( x , u ( n ) ) = 0 } V × U ( n )

determines a sub-manifold of the jet space. Henceforth, we will consider zero to be a regular value.

A smooth solution of the given system of differential equations is a smooth vector function u = f ( x ) such that

Δ ν ( x , f ( n ) ) = 0 ν = 1 , , l .

This is just a restatement of the fact that the derivatives J f α ( x ) of f must satisfy the algebraic constraints imposed by the system of differential equations.

A Lie point symmetry group of this system is a local group of transformations, which maps solutions to solutions. That is, if G be a local group of transformations acting on V × U , then g ( x , u ( n ) ) S Δ whenever ( x , u ( n ) ) S Δ and g G .

The tangent space of a local group of transformations is a Lie algebra – its vectors are often called infinitesimal generators. With starting point a Lie algebra by employing its exponential mapping, we can retrieve the local group of transformations with each element of the Lie algebra giving rise to a local subgroup of transformations called (local) one-parameter group. Through this local isomorphism, it is common practice to identify a Lie point symmetry with an infinitesimal generator.

Theorem 2

[20, Theorem 2.31]. Suppose that

Δ ν ( x , u ( n ) ) = 0 , ν = 1 , , l ,

is a system of differential equations over M V × U with zero being a regular value of Δ . If G is a local group of transformations acting on M , G its corresponding Lie algebra, and

pr ( n ) X [ Δ ν ( x , u ( n ) ) ] = 0 , ν = 1 , , l , whenever Δ ( x , u ( n ) ) = 0 ,

for every infinitesimal generator X of G , then G is a Lie point symmetry group of the system.

From these infinitesimal generators, it is possible to reduce the number of independent variables of the given system of equations and even obtain solutions. The solutions thus obtained are called invariant solutions of the infinitesimal generator employed. Thus, the natural question that arises once the set of symmetries has been calculated is how to obtain all the invariant solutions. This leads to the concept of optimal system, see [20, Proposition 3.6].

The main objective of our study is the Lie point symmetries of the Ricci flow for a general metric. A difficult and laborious task indeed! However, from these symmetries – by a method we developed – we will be able to retrieve the symmetries for any particular family of metrics by restricting the generic Lie algebra to the metric at hand, thereby eliminating the need to repeat each time the same copious process for finding the Lie point symmetries. This method is based on a particular form of the infinitesimal generator known as the canonical form, or simply, the characteristic.

Recall that given an infinitesimal generator

X = i = 1 p ξ i ( x , u ) x i + j = 1 q η j ( x , u ) u j ,

its canonical form is the vector

X = j = 1 q Q j u j ,

where

Q j = η j ( x , u ) i = 1 p ξ i ( x , u ) u j x i .

In the following example, we illustrate the method.

Example 1

We wish to obtain the Ricci flow symmetries of the metric

g ( x 1 , x 2 , t ) = e u ( x 1 , x 2 , t ) ( d x 1 d x 1 + d x 2 d x 2 )

from the symmetries of Theorem 1. Let

X = c 1 X 1 + c 2 X 2 + X 3 + X 4 + X 5 ,

where c 1 , c 2 R be a symmetry for the generic metric as provided by Theorem 1. First, we write X in its canonical form as follows:

Q X = 1 i j 2 Q i j t , x , g p s , g p s t , g p s x r g i j ,

where

Q i j = c 2 g i j ( x ) 1 s 2 g s i ξ s x j + g s j ξ s x i 1 s 2 ξ s ( x ) g i j x s ( c 1 + c 2 t ) g i j t .

Now, looking at the metric, it is easy to see that we need to impose the following restrictions to the generic metric of dimension 2: S = { g 11 = g 22 = e u and g 12 = 0 } . In addition, for this particular metric, the infinitesimal generator in the canonical form will be like

Q X S = Q x 1 , x 2 , t , u , u t , u x 1 , u x 2 u .

We are ready to give the restrictions that we need to impose to the Lie algebra of Theorem 1 for n = 2 :

Q X ( g 11 ) = Q X S ( e u ) Q X ( g 12 ) = Q X S ( 0 ) Q X ( g 22 ) = Q X S ( e u ) Q 11 = e u Q Q 12 = 0 Q 22 = e u Q ξ 1 u x 1 + ξ 2 u x 2 + ( c 1 + c 2 t ) u t c 2 + 2 ξ 1 x 1 e u = e u Q ξ 1 x 2 + ξ 2 x 1 e u = 0 ξ 1 u x 1 + ξ 2 u x 2 + ( c 1 + c 2 t ) u t c 2 + 2 ξ 2 x 2 e u = e u Q Q x 1 , x 2 , t , u , u t , u x 1 , u x 2 = ξ 1 u x 1 + ξ 2 u x 2 + ( c 1 + c 2 t ) u t c 2 + 2 ξ 1 x 1 ξ 1 x 1 ξ 2 x 2 = 0 ξ 1 x 2 + ξ 2 x 1 = 0 .

Therefore, the Lie algebra for the restricted problem is:

X = ( c 1 + c 2 t ) t + ξ 1 x 1 + ξ 2 x 2 + c 2 2 ξ 1 x 1 u ,

where ξ 1 and ξ 2 are arbitrary functions of x 1 and x 2 , satisfying the Cauchy-Riemann equations. Note that in the study by Cimpoiasu and Constantinescu [8], the very same Lie algebra from the ground up was obtained.

3 The Lie point symmetries of the Ricci flow

In this section, we compute the Lie point symmetries of the Ricci flow for the n -dimensional case, and we classify its finite-dimensional sub-algebra. We begin by stating and proving our main result.

By direct calculation, using SYM, of the Lie point symmetries for the cases n = 2 , 3 , and 4, arrive at Conjecture 1.

Conjecture 1

The Lie algebra of the classical symmetries, i.e., Lie point symmetries, of the Ricci flow on a Riemannian manifold ( M n , g ) with n 2 is spanned by the following base of vectors:

(1) X 1 = t , X 2 = t t + i = 1 n j i n g i j g i j , X k + 2 = ξ k x k i = 1 n j i n g k i ξ k x j + g k j ξ k x i g i j ,

where ξ 1 , , ξ n 1 e ξ n are arbitrary smooth functions of x 1 , , x n , k { 1 , , n } and g = ( g i j ) is the metric tensor.

It is easy to verify that the Lie algebra spanned by (1) leaves invariant the Ricci flow for n 2 .

  • The symmetry X 1 is obvious since the Ricci flow equations do not involve terms containing explicitly the variable t .

  • As for X 2 , observe that the Ricci tensor is invariant under scalings, i.e., Ric c g = Ric g for all c > 0 . Hence, by the one-parameter group generated by X 2 ,

    Ψ ε ( x , t , g i j ) = ( x , e ε t , e ε g i j ) ,

    we obtain that

    g ˆ i j ( x ˆ , t ˆ ) = e ε g i j ( x ˆ , e ε t ˆ )

    and

    t ˆ g ˆ i j = t g i j ( x , t ) = 2 Ric g i j = 2 Ric e ε g ˆ i j = 2 Ric g ˆ i j .

  • Finally, let us turn to the infinite ideal X k + 2 . We will show that these symmetries arise from the covariance of the Ricci flow equations. Let x ˆ = Ψ ε ( x ) be the flow determined by the initial value problem as follows:

    d x ˆ k d ε ε = 0 = ξ k ( x ˆ ) , x ˆ k ε = 0 = x k .

    Let g i j be the metric in coordinates x . Since it is a tensor, it is covariant. So, in any other coordinate system, x ˆ , it has the fofollowing form:

    (2) g ˆ s l = g i j x i x ˆ s x j x ˆ l .

    We wish to analyze g ˆ s l around ε = 0 , hence obtaining the infinitesimal change of g ˆ s l . By the properties of the flow, we know that x = Ψ ε ( x ˆ ) . Thus,

    x i x ˆ s = x i x ˆ s ε = 0 + d d ε x i x ˆ k ε = 0 ε + O ( ε 2 ) = δ i s + x ˆ s d Ψ ε i d ε ε = 0 ε + O ( ε 2 ) = δ i s ξ i x s ε + O ( ε 2 ) .

    So, equation (2) becomes

    g ˆ s l = g i , j δ i s ξ i x s ε + O ( ε 2 ) δ j l ξ j x l ε + O ( ε 2 ) = g s l g s i ξ i x l + g i l ξ i x s ε + O ( ε 2 ) .

    Therefore, as we have claimed, the symmetries X k + 2 merely depict the covariance of the Ricci flow.

The conjecture functions as a necessary condition; indeed, the Ricci flow admits at least the Lie algebra of symmetries spanned by equation (1). To prove the conjecture, we need to show that it is also sufficient, i.e., the Ricci flow admits no more Lie symmetries than the ones already found.

Proof

Let us suppose that

X = ξ t ( t , x 1 , , x n , g ) t + ξ μ ( t , x 1 , , x n , g ) x μ + η ( μ ν ) ( t , x 1 , , x n , g ) g μ ν

is the infinitesimal generator of a Lie point symmetry of the Ricci flow. Observe that the second summation is restricted so that the mixed derivatives of g are not counted twice.

Using this generic form, we can arrive at the determining equations. At this point, we employ the idea of Marchildon [17] to solve them. First, we write the Ricci flow as

2 R α β + t g α β = 0 ,

where

(2) R α β = 1 2 g γ δ { γ δ g α β α β g γ δ + β δ g α γ + α γ g δ β } + g γ δ g τ ρ { Γ τ γ α Γ ρ δ β Γ τ γ δ Γ ρ α β } ,

and Γ γ α β = Γ τ γ α g τ β with

Γ τ γ α = 1 2 ( α g τ γ + γ g τ α τ g γ α ) .

At this point it is advantageous – for the sake of clarity – to define the tensor

X μ ν κ λ g μ ν g κ λ = g μ κ g ν λ + g μ λ g ν κ , if κ λ , g μ κ g ν λ , if κ = λ .

Since X μ ν k λ = X a ν k λ g μ a = X a b k λ g b ν g μ a , we have that

(3) X μ ν κ λ = δ μ κ g ν λ + δ μ λ g ν κ , if κ λ , δ μ κ g ν λ , if κ = λ ,

and

X μ ν κ λ = δ μ κ δ ν λ + δ μ λ δ ν κ , if κ λ , δ μ κ δ ν λ , if κ = λ .

In particular,

(4) g μ ν g k λ = X μ ν k λ , ( γ δ g α β ) ( k λ g μ ν ) = X γ δ k λ X α β μ ν , 2 Γ τ γ α ( k g μ ν ) = δ α k X τ γ μ ν + δ γ k X τ α μ ν δ τ k X γ α μ ν .

We need to express the partial derivatives of the Ricci tensor with respect to g μ ν and its partial derivatives. In what follows, we denote fixed indices with a hat symbol, ˆ . From equations (2) and (4), we obtain

R α β ( κ λ g μ ν ) = ( κ λ g μ ν ) g γ δ 2 { γ δ g α β α β g γ δ + β δ g α γ + α γ g δ β } = 1 2 g γ δ { X γ δ κ λ X α β μ ν X α β κ λ X γ δ μ ν + X δ β κ λ X γ α μ ν + X γ α κ λ X δ β μ ν } ,

R α β ( κ g μ ν ) = ( κ g μ ν ) ( g γ δ g τ ρ { Γ τ γ α Γ ρ δ β Γ τ γ δ Γ ρ α β } ) = 1 2 g γ δ g τ ρ { [ δ α κ X τ γ μ ν + δ γ κ X τ α μ ν δ τ κ X γ α μ ν ] Γ ρ δ β + [ δ β κ X ρ δ μ ν + δ δ κ X ρ β μ ν δ ρ κ X δ β μ ν ] Γ τ γ α [ δ δ κ X τ γ μ ν + δ γ κ X τ δ μ ν δ τ κ X γ δ μ ν ] Γ ρ α β [ δ β κ X ρ α μ ν + δ α κ X ρ β μ ν δ ρ κ X α β μ ν ] Γ τ γ δ } ,

R α β g μ ν = 1 2 { γ δ g α β + α β g γ δ δ β g γ α γ α g δ β } X γ δ μ ν { Γ τ γ α Γ ρ δ β Γ τ γ δ Γ ρ α β } { g γ δ X τ ρ μ ν + g τ ρ X γ δ μ ν } ,

see [17]. Note that for any symmetric tensor A , we have

(5) A ( μ ν ) X γ δ μ ν = A γ δ .

The second prolongation of X is

X ( 2 ) = ξ μ x μ + ξ t t + η ( μ ν ) g μ ν + η ( μ ν ) k ( k g μ ν ) + η ( μ ν ) t ( t g μ ν ) + η ( μ ν ) ( k λ ) ( k λ g μ ν ) .

Therefore, to obtain the linearization conditions, we first need to expand the equations

(6) X ( 2 ) R α β + 1 2 t g α β = 1 2 η α β t + η γ δ 2 { γ δ g α β + α β g γ δ δ β g γ α γ α g δ β } { Γ τ γ α Γ ρ δ β Γ τ γ δ Γ ρ α β } { g γ δ η τ ρ + g τ ρ η γ δ } + 1 2 g γ δ g τ ρ { [ η τ γ α + η τ α γ η γ α τ ] Γ ρ δ β + [ η ρ δ β + η ρ β δ η δ β ρ ] Γ τ γ α [ η τ γ δ + η τ δ γ η γ δ τ ] Γ ρ α β [ η ρ α β + η ρ β α η α β ρ ] Γ τ γ δ } + 1 2 g γ δ { η α β γ δ η γ δ α β + η γ α δ β + η δ β γ α } = 0 ,

where

η γ δ = a b η a b X γ δ a b = η ( a b ) X γ δ a b .

The functions η τ γ α , η τ γ t , and η α β γ δ are given by the following expressions:

η τ γ α = D α ( η τ γ ξ σ σ g τ γ ξ t t g τ γ ) + ξ σ α σ g τ γ + ξ t α t g τ γ , η τ γ t = D t ( η τ γ ξ σ σ g τ γ ξ t t g τ γ ) + ξ σ t σ g τ γ + ξ t t t g τ γ , η α β γ δ = D γ D δ ( η α β ξ σ σ g α β ξ t t g α β ) + ξ σ γ δ σ g α β + ξ t γ δ t g α β ,

see Olver [20, Theorem 2.36], where D α and D t denote the total derivative operator with respect to x α and t , respectively, i.e.,

D α = α + α g ( μ ν ) g μ ν + α κ g ( μ ν ) ( κ g μ ν ) + α κ λ g ( μ ν ) ( κ λ g μ ν ) , D t = t + t g ( μ ν ) g μ ν + t κ g ( μ ν ) ( κ g μ ν ) + t κ λ g ( μ ν ) ( κ λ g μ ν ) .

Hence,

(7) η τ γ α = α η τ γ [ σ g τ γ ] α ξ σ [ t g τ γ ] α ξ t + α g ( μ ν ) η τ γ g μ ν [ α g ( μ ν ) ] σ g τ γ ξ σ g μ ν [ α g ( μ ν ) ] t g τ γ ξ t g μ ν , η τ γ t = t η τ γ [ σ g τ γ ] t ξ σ [ t g τ γ ] t ξ t + t g ( μ ν ) η τ γ g μ ν [ t g ( μ ν ) ] σ g τ γ ξ σ g μ ν [ t g ( μ ν ) ] t g τ γ ξ t g μ ν ,

and

(8) η α β γ δ = γ δ η α β γ δ ξ σ ( σ g α β ) γ δ ξ t ( t g α β ) + ( δ g ( μ ν ) ) γ η α β g μ ν + γ g ( μ ν ) δ η α β g μ ν ( γ g ( μ ν ) ) ( σ g α β ) δ ξ σ g μ ν ( δ g ( μ ν ) ) ( σ g α β ) γ ξ σ g μ ν ( γ g ( μ ν ) ) ( t g α β ) δ ξ t g μ ν ( δ g ( μ ν ) ) ( t g α β ) γ ξ t g μ ν + ( δ g ( a b ) ) ( γ g ( μ ν ) ) 2 η α β g μ ν g a b ( δ g ( a b ) ) ( γ g ( μ ν ) ) ( σ g α β ) 2 ξ σ g μ ν g a b ( δ g ( a b ) ) ( γ g ( μ ν ) ) ( t g α β ) 2 ξ t g μ ν g a b ( δ σ g α β ) γ ξ σ ( γ σ g α β ) δ ξ σ ( δ t g α β ) γ ξ t ( γ t g α β ) δ ξ t + ( γ δ g ( μ ν ) ) η α β g μ ν ( γ g ( μ ν ) ) ( δ σ g α β ) ξ σ g μ ν ( δ g ( μ ν ) ) ( γ σ g α β ) ξ σ g μ ν ( γ g ( μ ν ) ) ( δ t g α β ) ξ t g μ ν ( δ g ( μ ν ) ) ( γ t g α β ) ξ t g μ ν ( σ g α β ) ( δ γ g ( μ ν ) ) ξ σ g μ ν ( t g α β ) ( δ γ g ( μ ν ) ) ξ t g μ ν .

Finally, we obtain the linearized symmetry conditions – having in mind that the metric must also satisfy the Ricci flow equation, 2 R α β + t g α β = 0 by substituting equations (7) and (8) into equation (6).

The following is the strategy that we will abide by: from the linearized symmetry conditions, we will extract the determining equations starting with the terms involving derivatives of the metric tensor that do not appear in the Ricci flow equation, then continue with terms that involve higher-order derivatives of the metric, and finally those involving the derivatives of the low-order metric tensor since the corresponding determining equations are usually much easier to solve in this order.

Using the expressions of X μ ν κ λ , X μ ν κ λ , and X κ λ μ ν , we can rewrite certain terms, for instance,

(9) g γ δ ξ t g ( μ ν ) t g γ δ α β g μ ν = t g ( c d ) ( a b g ( μ ν ) ) ξ t g μ ν δ a α δ b β X γ δ c d g γ δ = t g ( c d ) ( a b g ( μ ν ) ) ξ t g μ ν δ a α δ b β G c d ,

where

G ρ σ = g ρ σ , if ρ = σ 2 g ρ σ , if ρ σ .

Renaming the indexes on the right-hand side of equation (9) as c ρ , a δ , d κ , and b γ , we obtain

g γ δ ξ t g ( μ ν ) t g γ δ α β g μ ν = t g ( ρ κ ) ( δ γ g ( μ ν ) ) ξ t g μ ν δ δ α δ γ β G ρ κ .

We are now ready to extract some of the coefficients equations (3)–(5).

From the g g terms: Note that the only components in equation (6) that include the term g g are η α β γ δ , η γ δ α β , η γ α δ β , and η δ β γ α , and since the Ricci flow does not involve the terms t g κ σ g , we have our first family of determinant equations as follows:

(10) g γ δ ξ t g ( μ ν ) { t g α β δ γ g μ ν + t g γ δ α β g μ ν t g α γ δ β g μ ν t g δ β α γ g μ ν } = 0 .

Rearranging the previous equation until we have the same index in the derivatives of g , we obtain

(11) t g ( ρ κ ) ( δ γ g ( μ ν ) ) ξ t g μ ν ( g γ δ X α β ρ κ + δ δ α δ γ β G ρ κ δ γ β X α δ ρ κ δ δ α X β γ ρ κ ) = 0 .

Case κ = ρ = β , δ = γ = α , α β : From equation (11), we have

ξ t g ( μ ν ) ( g α α X α β β β X β α β β ) = 0 .

Thus,

ξ t g μ ν g α β = 0 ,

for all μ ν . Therefore, ξ t depends only on x and t .

Observe that, the terms i g t j g are multiplied by ξ t g ( μ ν ) , thus they vanish.

Since, in the Ricci flow, there are no terms of the form κ g σ ρ g , we obtain

(12) g γ δ ξ σ g ( μ ν ) { γ g μ ν δ σ g α β + δ g μ ν γ σ g α β + σ g α β δ γ g μ ν + α g μ ν β σ g γ δ + β g μ ν α σ g γ δ + σ g γ δ α β g μ ν δ g μ ν β σ g α γ β g μ ν δ σ g α γ σ g α γ δ β g μ ν γ g μ ν α σ g δ β α g μ ν γ σ g δ β σ g δ β α γ g μ ν } = 0 .

Rearranging equation (12), we obtain

(13) γ g ( μ ν ) δ σ g ( ρ κ ) 2 g γ δ X α β ρ κ ξ σ g μ ν + g σ δ X α β μ ν ξ γ g ρ κ + ( δ α γ δ β δ + δ α δ δ β γ ) G ρ κ ξ σ g μ ν + δ α δ δ β σ G μ ν ξ γ g ρ κ δ β δ X α γ ρ κ ξ σ g μ ν δ β γ X α δ ρ κ ξ σ g μ ν δ β σ X α δ μ ν ξ γ g ρ κ δ α δ X β γ κ ρ ξ σ g μ ν δ α γ X β δ κ ρ ξ σ g μ ν δ α σ X β δ ν μ ξ γ g ρ κ = 0 .

Case δ = σ = κ = κ ˆ , α κ ˆ , β κ ˆ , α γ , β γ : From equation (13), we have

0 = g κ ˆ κ ˆ X α β μ ν ξ γ g ρ κ ˆ ,

i.e.,

0 = ξ γ g ρ κ ˆ ,

for all γ and ρ κ ˆ . Then, ξ γ depends only on x and t .

From the g terms: Applying the same argument as in equations (10) and (13), we have the following set of determining equations:

g γ δ { δ t g α β γ ξ t + γ t g α β δ ξ t + β t g γ δ α ξ t + α t g γ δ β ξ t β t g γ α δ ξ t δ t g γ α β ξ t α t g δ β γ ξ t γ t g δ β α ξ t } = 0 .

Rearranging indices, one can show that the previous equations become

(14) γ ξ t δ t g ( ρ κ ) ( 2 X α β ρ κ g γ δ + G ρ κ ( δ β δ δ α γ + δ α δ δ β γ ) X α γ ρ κ δ β δ X α δ ρ κ δ β γ X β γ ρ κ δ α δ δ α γ X β δ ρ κ ) = 0 .

Case γ = δ = α = β , ρ = κ = κ ˆ , κ ˆ β : From equation (14), it is necessary that

( β ξ t ) g κ ˆ κ ˆ = 0 .

Thus,

β ξ t = 0

for all β . Therefore, ξ t depends only on t .

Now we group the terms that contain only second spatial derivatives of the metric tensor. However, since we want that 2 R α β + t g α β = 0 , we must also consider the terms that contain t g α β . We have

+ 1 2 g γ δ ( γ δ g α β α β g γ δ + β δ g α γ + α γ g δ β ) t ξ t 1 2 g γ δ ( γ δ g μ ν μ ν g γ δ + ν δ g μ γ + μ γ g ν δ ) η α β g ( μ ν ) + 1 2 η γ δ ( γ δ g α β + α β g γ δ δ β g γ α γ α g δ β ) + 1 2 g γ δ γ ξ σ δ σ g α β + δ ξ σ γ σ g α β γ δ g ( μ ν ) η α β g μ ν + 1 2 g γ δ β ξ σ α σ g γ δ + α ξ σ β σ g γ δ α β g ( μ ν ) η γ δ g μ ν 1 2 g γ δ β ξ σ δ σ g γ α + δ ξ σ β σ g γ α δ β g ( μ ν ) η γ α g μ ν 1 2 g γ δ γ ξ σ α σ g δ β + α ξ σ γ σ g δ β γ α g ( μ ν ) η δ β g μ ν .

Rearranging the indices, we obtain

(15) γ δ g ρ σ ( δ α ρ δ β σ η γ δ + δ γ α δ δ β η ρ σ δ α σ δ β δ η ρ γ δ β σ δ δ α η γ ρ + 2 δ α ρ δ β σ g κ γ κ ξ δ η α β g ( ρ σ ) g γ δ + δ γ β g ρ σ α ξ δ + δ γ α g ρ σ β ξ δ δ γ α δ δ β g μ ν η μ ν g ( ρ σ ) δ α σ δ γ β g ρ κ κ ξ δ δ α σ g ρ γ β ξ δ + δ δ β g κ γ η κ α g ( ρ σ ) δ β σ δ γ α g κ ρ κ ξ δ δ β σ g γ ρ α ξ δ + δ δ α g γ κ η κ β g ( ρ σ ) + { δ α ρ δ β σ g γ δ g ρ σ δ γ α δ δ β + δ ρ α δ γ β g σ δ + δ β σ δ α γ g δ ρ } t ξ t + g γ δ η α β g ( ρ σ ) + g ρ σ η α β g ( γ δ ) g δ σ η α β g ( γ ρ ) g σ δ η α β g ( ρ γ ) .

Case σ = γ = δ = ρ = ρ ˆ , ρ ˆ α β : From equation (15), we obtain

η α β g ρ ˆ ρ ˆ g ρ ˆ ρ ˆ = 0 ,

and in consequence

(16) η α β = F α β ( g α ρ , g β ρ , g α α , g β β , g κ ρ , x , t ) ,

with α ρ β κ .

Case ρ = σ = α , γ = δ = δ ˆ , α δ ˆ β : From equation (15), we have the following determining equation:

g α δ ˆ β ξ δ ˆ g δ ˆ α η α β g α δ ˆ = 0 .

Using equation (16), we obtain

(17) η α β = κ β , α g α κ β ξ κ + F α β ( g α β , g δ β , g β β , g α α , g κ δ , x , t ) ,

with α β δ κ .

Case ρ = σ = β , γ = δ = δ ˆ , β δ ˆ α : From equation (15), we obtain

(18) g β δ ˆ α ξ δ ˆ g δ ˆ β η α β g β δ ˆ = 0 .

Using equation (17), we obtain from equation (18) that

(19) η α β = κ β , α g α κ β ξ κ + g β κ α ξ κ + F α β ( g α β , g β β , g α α , g κ δ , x , t )

for α β δ κ .

Note that the only way for η α β to depend on g κ δ with α β κ δ is that n 4 , our next goal is to prove that η α β g κ δ = 0 . We consider for the next step n 4 .

Case δ = γ , and α , β δ : From equation (15), we obtain

δ δ g ρ σ δ α ρ δ β σ η δ δ + 2 δ α ρ δ β σ g κ δ κ ξ δ g δ δ η α β g ( ρ σ ) δ α σ g ρ δ β ξ δ δ β σ g δ ρ α ξ δ δ α ρ δ β σ g δ δ t ξ t + g δ δ η α β g ( ρ σ ) + g ρ σ η α β g δ δ g δ σ η α β g ( δ ρ ) g σ δ η α β g ( ρ δ ) = 0 .

Furthermore, if ρ σ , and due to the fact that δ δ g ρ σ = δ δ g σ ρ , we need to symmetrize the above expression for the indices ρ and σ , and the result is

(20) δ δ g ρ σ ( δ α ρ δ β σ + δ α σ δ β ρ ) η δ δ + 2 ( δ α ρ δ β σ + δ α σ δ β ρ ) g κ δ κ ξ δ g δ δ η α β g ρ σ ( δ α σ g ρ δ + δ α ρ g σ δ ) β ξ δ ( δ β σ g δ ρ + δ β ρ g δ σ ) α ξ δ ( δ α ρ δ β σ + δ α σ δ β ρ ) g δ δ t ξ t + g δ δ η α β g ρ σ + 2 g ρ σ η α β g δ δ g δ σ η α β g ( δ ρ ) + g δ ρ η α β g ( δ σ ) g σ δ η α β g ( ρ δ ) + g ρ δ η α β g ( σ δ ) .

Finally, if σ = δ = δ ˆ , ρ = ρ ˆ , and ρ ˆ β α δ ˆ , we have the following determining equation:

(21) η α β g ρ ˆ δ ˆ g δ ˆ δ ˆ = 0 .

From equations (19) and (21), we obtain

(22) η α β = κ β , α g α κ β ξ κ + g β κ α ξ κ + F α β ( g α β , g β β , g α α , x , t ) ,

with α β .

Case σ = γ = δ = ρ = ρ ˆ , α = β , and ρ ˆ β : From equation (15), we have the following determining equation:

g ρ ˆ ρ ˆ η α α g ρ ˆ ρ ˆ = 0 .

Therefore,

(23) η α α = F α α ( g α ρ , g α α , g ρ κ , x , t ) ,

with ρ α κ .

Case σ = ρ = ρ ˆ , α = β , γ = δ = δ ˆ , and ρ ˆ α δ ˆ : From equation (15), we have the following determining equation:

(24) g δ ˆ ρ ˆ η α α g δ ˆ ρ ˆ = 0 .

By equations (23) and (24), it follows that

(25) η α α = F α α ( g α ρ , g α α , x , t ) ,

with α ρ .

Case δ = γ = α , σ = ρ = ρ ˆ , and α ρ ˆ β : From equations (15) and (19), we have the following determining equation:

g ρ ˆ ρ ˆ β ξ α + η α β g α α + g α ρ ˆ η ρ ˆ β g ρ ˆ ρ ˆ + β ξ ρ ˆ = 0 ,

and by differentiating the above equation with respect to g α ρ ˆ ,

g ρ ˆ ρ ˆ g α ρ ˆ β ξ α + η α β g α α + g α ρ ˆ g α ρ ˆ η ρ ˆ β g ρ ˆ ρ ˆ + β ξ ρ ˆ = 0 .

Note that the expressions in parentheses do not depend on g α ρ , see equation (17). By manipulating the above equation, we can show that this equation is equivalent to

( g α α g ρ ˆ ρ ˆ g α ρ ˆ g α ρ ˆ ) η ρ ˆ β g ρ ˆ ρ ˆ + β ξ ρ ˆ = 0 .

On the other hand, since g 1 is a positive definite matrix, we have that ( g α α g ρ ˆ ρ ˆ g α ρ ˆ g α ρ ˆ ) > 0 . Hence,

η ρ ˆ β g ρ ˆ ρ ˆ + β ξ ρ ˆ = 0

for all ρ ˆ β . By equation (22), we have

(26) η α β = κ g α κ β ξ κ + g β κ α ξ κ + F α β ( g α β , x , t ) = g α κ β ξ κ g β κ α ξ κ + F α β ( g α β , x , t ) .

with α β .

Case ρ = β = α , γ = σ = δ = δ ˆ and α δ ˆ : From equation (20), we have the following determining equation:

(27) 2 g δ ˆ δ ˆ α ξ δ ˆ g δ ˆ δ ˆ η α α g α δ ˆ = 0 .

Putting (27) and (25), we obtain

η α α = 2 κ α g κ α α ξ κ + F ˆ α α ( g α α , x , t ) .

We can write

(28) η α α = 2 κ = 1 n g κ α α ξ κ + F α α ( g α α , x , t ) = 2 g κ α α ξ κ + F α α ( g α α , x , t ) .

Case ρ = σ = α = β , γ = δ = δ ˆ and δ ˆ α : From equation (15) we have the following determining equation:

η δ ˆ δ ˆ + 2 g κ δ ˆ κ ξ δ ˆ g δ ˆ δ ˆ t ξ t = 0 .

Using equations (26) and (28), we obtain

(29) 0 = η δ ˆ δ ˆ + 2 g κ δ ˆ κ ξ δ ˆ g δ ˆ δ ˆ t ξ t = η a b g δ ˆ a g δ ˆ b + 2 g κ δ ˆ κ ξ δ ˆ g δ ˆ δ ˆ t ξ t = F a b g a δ ˆ g b δ ˆ ( g a κ b ξ κ + g b κ a ξ κ ) g a δ ˆ g b δ ˆ + 2 g κ δ ˆ κ ξ δ ˆ g δ ˆ δ ˆ t ξ t = F a b g a δ ˆ g b δ ˆ g δ ˆ δ ˆ t ξ t .

Differentiating with respect to g δ ˆ δ ˆ and making elementary simplifications, it can be shown that

(30) g δ ˆ δ ˆ F δ ˆ δ ˆ g δ ˆ δ ˆ F a b g a δ ˆ g b δ ˆ = 0 .

Combining equations (29) and (30), we can conclude that

(31) F δ δ = t ξ t g δ δ + F ˆ δ δ ( x , t ) .

Case γ = δ = α , ρ = σ = β and α β : From equation (15), obtain the following determining equation:

η α β ( g κ β k ξ α + g κ α κ ξ β ) g α β ( α ξ α β ξ β ) + g α β η β β g β β η α β g α β + g α β t ξ t = 0 .

Using equations (26) and (28), this above equation becomes

0 = F a b g α a g β b + g α β F α β g α β g α β F β β g β β + g α β t ξ t .

By equation (31), we have

(32) 0 = F a b g α a g β b + g α β F α β g α β

for all α β .

Differentiating with respect to g κ ˆ κ ˆ with κ ˆ α β , we obtain that

0 = g α κ ˆ g β κ ˆ F κ ˆ κ ˆ g κ ˆ κ ˆ + F a b [ g α κ ˆ g a κ ˆ g b β + g a α g β κ ˆ g b κ ˆ ] g α κ ˆ g β κ ˆ F α β g α β = t ξ t g α κ ˆ g β κ ˆ + g α κ ˆ g β κ ˆ F κ ˆ β g κ ˆ β + g β κ ˆ g α κ ˆ F α κ ˆ g α κ ˆ g α κ ˆ g β κ ˆ F α β g α β = g α κ ˆ g β κ ˆ t ξ t + F κ ˆ β g κ ˆ β + F κ ˆ α g κ ˆ α F α β g α β ,

for all fixed κ ˆ α β . Hence,

0 = t ξ t + F κ ˆ β g κ ˆ β + F κ ˆ α g κ ˆ α F α β g α β .

We can conclude that

F α β = F ˆ α β 1 ( x , t ) g α β + F ˆ α β 2 ( x , t )

for all α β , see equation (26).

Now, we differentiate equation (32) with respect to g α β as follows:

0 = F α β g α β ( g α α g β β + g α β g α β ) + F a b ( ( g α α g a β + g α β g α a ) g β b + g α a ( g α β g β b + g β β g α b ) ) ( g α α g β β + g α β g β α ) F α β g α β = 2 F α β g α β ( g α α g β β + g α β g α β ) + F β β g β β g β β g α α + F α β g α β g β α g α β + F α β g α β g α β g α β + F β β g β β g β β g α α = 2 g α α g β β F α β g α β + t ξ t

for all fixed α β . Thus,

0 = F α β g α β + t ξ t .

Therefore,

(33) F α β = t ξ t g α β + F ˆ α β ( x , t ) ,

for all β α . Putting equations (31) and (33) together, we have that

(34) F α β = t ξ t g α β + F ˆ α β ( x , t )

for all α and β .

From the g terms: For brevity, we leave the partial derivatives of the metric in terms of the Christoffel symbols as follows:

(35) 1 2 ( Γ α β σ + Γ β α σ ) t ξ σ + 1 2 g γ δ g τ ρ { ( α η τ γ + γ η τ α τ η γ α ) Γ ρ δ β + ( β η ρ δ + δ η ρ β ρ η δ β ) Γ τ γ α ( δ η τ γ + γ η τ δ τ g γ δ ) Γ ρ α β ( β η ρ α + α η ρ β ρ η α β ) Γ τ γ δ } + 1 2 g γ δ [ γ δ ξ σ ( Γ α β σ + Γ β α σ ) + α β ξ σ ( Γ γ δ σ + Γ δ γ σ ) δ β ξ σ ( Γ γ α σ + Γ α γ σ ) γ α ξ σ ( Γ δ β σ + Γ β δ σ ) 2 γ η α β g ( μ ν ) ( Γ μ ν δ + Γ ν μ δ ) α η γ δ g ( μ ν ) ( Γ μ ν β + Γ ν μ β ) β η γ δ g ( μ ν ) ( Γ μ ν α + Γ ν μ α ) + δ η γ α g ( μ ν ) ( Γ μ ν β + Γ ν μ β ) + β η γ α g ( μ ν ) ( Γ μ ν δ + Γ ν μ δ ) + γ η δ β g ( μ ν ) ( Γ μ ν α + Γ ν μ α ) + α η δ β g ( μ ν ) ( Γ μ ν γ + Γ ν μ γ ) .

On the other hand, note that

γ η α β g ( μ ν ) ( Γ μ ν δ + Γ ν μ δ ) = γ F α β g α β ( Γ α β δ + Γ μ δ ) ,

see equations (26) and (28). Using equations (26) and (28), equation (35) becomes

(36) 1 2 ( Γ α β σ + Γ β α σ ) t ξ σ + 1 2 g γ δ g τ ρ { ( α F ˆ τ γ + γ F ˆ τ α τ F ˆ γ α ) Γ ρ δ β + ( β F ˆ ρ δ + δ F ˆ ρ β ρ F ˆ δ β ) Γ τ γ α ( δ F ˆ τ γ + γ F ˆ τ δ τ F ˆ γ δ ) Γ ρ α β ( β F ˆ ρ α + α F ˆ ρ β ρ F ˆ α β ) Γ τ γ δ } .

Rearranging indices and considering the symmetry Γ λ μ ν = Γ λ ν μ , equation (36) becomes

(37) { g τ λ ( δ ν β g γ μ + δ μ β g γ ν ) ( α F ˆ τ γ + γ F ˆ τ α τ F ˆ γ α ) + g λ ρ ( δ ν α g μ δ + δ μ α g ν δ ) ( β F ˆ ρ δ + δ F ˆ ρ β ρ F ˆ δ β ) g τ λ g γ δ ( δ μ α δ ν β + δ ν α δ μ β ) ( δ F ˆ τ γ + γ F ˆ τ δ τ F ˆ γ δ ) 2 g λ ρ g μ ν ( β F ˆ ρ α + α F ˆ ρ β ρ F ˆ α β ) δ λ α ( δ μ β t ξ ν + δ ν β t ξ μ ) λ β ( δ μ α t ξ ν + δ ν α t ξ μ ) } 1 2 Γ λ μ ν .

Case μ α , ν α , μ β , λ = λ ˆ and ν β : From equation (37), we have the following determining equation:

0 = g λ ˆ ρ ( β F ˆ ρ α + α F ˆ ρ β ρ F ˆ α β )

for α , β , λ ˆ . As F ˆ ρ α depend only on x and t , see (34), we can conclude that

0 = β F ρ ˆ α + α F ρ ˆ β ρ ˆ F α β

for all α , β , ρ ˆ . Permuting the indices and the fact that F ˆ α β = F ˆ β α , we can conclude that

β F ˆ ρ ˆ α = 0

for all β , ρ ˆ , α . In particular, F ˆ α β depends only on t

(38) F α β = t ξ t g α β + F ˆ α β ( t )

Since F α β depends only on t and g α β , the expression (37) becomes

{ δ λ α ( δ μ β t ξ ν + δ ν β t ξ μ ) λ β ( δ μ α t ξ ν + δ ν α t ξ μ ) } 1 2 Γ λ μ ν .

Case α = λ , μ = β , α β , and ν = ν ˆ : From the above equation, we obtain the following determining equation:

t ξ ν ˆ + δ ν ˆ β t ξ β = 0

for all β , ν ˆ . Then,

t ξ β = 0

for all β . This means that ξ β is only a function of x .

Terms that do not contain derivatives of the metric tensor: we have the following determining equation:

(39) 0 = g α β t t ξ t + t F ˆ α β .

Hence,

t t ξ t = 0 .

Thus, ξ t = a 1 t + a 2 , where a 1 and a 2 are constants. From equation (39), we can also conclude that

t F ˆ α β = 0 ,

Therefore, (38) becomes

F α β = a 1 g α β + C α β ,

where C α β is a constant.

From the g g terms: We have

0 = { Γ τ γ α Γ ρ δ β Γ τ γ δ Γ ρ α β } { g γ δ g d ρ g c τ + g τ ρ g d δ g c γ } C c d

for all fixed α and β . Rearranging indices, we obtain

Γ τ γ a Γ ρ δ β ( δ α a ( g γ δ g d ρ g c τ + g τ ρ g d δ g c γ ) δ α δ ( g γ a g d ρ g c τ + g τ ρ g d a g c γ ) ) C c d .

Case γ = a = τ = α and ρ = δ = β : From the above equation, we have the following determining equation:

Γ α α α Γ β β β ( g α β g d β g c α ) C c d = 0 ,

thus

( g α β g d β g c α ) C c d = 0 .

By differentiating it for g β β g α β , we obtain

g c α C c β + g α β C β β = 0 .

From this equation, we have

C β β = 0 , C α β = 0 .

Gathering everything together, we can conclude that a sufficient condition for X to be Lie point symmetry of the Ricci flow is to be of the form

X = ( a 0 + a 1 t ) t + ξ k ( x ) x k g k i ξ k x j + g k j ξ k x i a 1 g i j g ( i j ) .

Hence, we have proved Theorem 1.□

Now, we classify the finite-dimensional algebra Theorem 1

Proposition 1

The one-dimensional optimal system associated with the Lie algebra span { X 1 , X 2 } is: { X 1 , X 2 } . While the two dimensional optimal system is trivially the whole Lie algebra span { X 1 , X 2 } .

Proof

See example 10.1 in [11].□

Observation 1

The Ricci flow viewed as a mapping of a suitable jet space is analytic. Thus, we have the guarantee that the symmetries obtained through the infinitesimal criterion are the maximum continuous symmetry group, see Corollary 2.74 in [20].

4 The Ricci flow for particular families of metrics

4.1 The Einstein equations

In this section, we obtain the Lie point symmetries of the vacuum Einstein equations retrieving them from the ones of the Ricci flow.

Proposition 2

The Lie algebra of the Lie point symmetries of the vacuum Einstein equations in n -dimensional manifold is spanned by the vectors:

X 1 = 1 i j n g i j g i j , X k + 1 = ξ k x k 1 i j n g k i ξ k x j + g k j ξ k x i g i j ,

where ξ 1 , , ξ n are arbitrary smooth functions of { x 1 , , x n } , k { 1 , , n } .

Proof

Let

X = c 1 X 1 + c 2 X 2 + X 3 + + X n + 2 ,

where c 1 , c 2 R , be a symmetry of Theorem 1. We write X in its canonical form:

Q X = 1 k l n Q k l t , x , g p s , g p s t , g p s x r g k l

and employ the method illustrated in Example 1. That being so, we need to consider the following restriction: g i j t = 0 , i.e., g i j = g i j ( x 1 , , x n ) . So, the characteristic takes the form

Q ˆ k l = c 2 g k l ( x ) 1 s n ξ s ( x ) g k l x s 1 s n g s k ξ s x l + g s l ξ s x k ,

or equivalently,

X = ξ k x k 1 i j n c 2 g i j + g k i ξ k x j + g k j ξ k x i g i j ,

from which we arrive at the desired proposition.□

Observation 2

These symmetries in n dimensions were obtained first by Marchildon in [17].

In the following sections, we apply this method to find the symmetries of the Ricci flow for particular metric families.

4.2 Ricci flow on warped product manifolds

Given two Riemannian manifolds ( B n , g B ) , denoted as the base, and ( F m , g F ) , denoted as the fiber, and a positive smooth warping function φ on B n , we consider on the product manifold B n × F m the warped metric g = π 1 * g B + ( φ π 1 ) 2 π 2 * g F , where π 1 and π 2 are the natural projections on B n and F m , respectively. Under these conditions, the product manifold is called the warped product of B n and F m .

We will study the Ricci flow on the warped product M = B n × F m with

( B n , g B ) = R n , 1 ψ 2 ( x ) d x i d x i ,

and by abuse of notation, we write

g = 1 ψ 2 ( x ) d x i d x i + φ 2 ( x ) g F ( y ) ,

where ψ and φ being positive smooth functions. These kinds of products have been extensively studied, see for example, Angenent and Knopf [1], Ivey [14], Oliynyk and Woolgar [19], and Ma and Xu [16].

We know that the Ricci flow preserves the isometry group of the initial Riemannian manifold. For warped metrics, this means that the Ricci flow preserves the warped structure (see [19]). As a consequence, we can write

g = 1 ψ 2 ( x , t ) d x i d x i + φ 2 ( x , t ) g F ( y , t ) .

Consider now the smooth fields W 1 and W 2 on F m , it follows that (see [4,23])

Ric M ( W ˜ 1 , W ˜ 2 ) = Ric g F ( W 1 , W 2 ) Δ B φ φ + B φ 2 φ 2 ( m 1 ) g ( W ˜ 1 , W ˜ 2 ) ,

and by the Ricci flow, we have

Ric M ( W ˜ 1 , W ˜ 2 ) = φ t φ g F ( W 1 , W 2 ) .

Thus,

Ric g F ( W 1 , W 2 ) = φ φ t + φ 2 Δ B φ φ + B φ 2 φ 2 ( m 1 ) g F ( W 1 , W 2 ) .

Finally, we have

φ φ t + φ 2 Δ B φ φ + B φ 2 φ 2 ( m 1 ) = μ ,

with μ depending only on t. In particular, ( F m , g F ) must be an Einstein manifold for each t . This motivates us to study the case in which the metric over time is of the following form:

(40) g = 1 ψ 2 ( x , t ) d x i d x i + φ 2 ( x , t ) g can .

where ( F m , g can ) is an Einstein manifold.

Theorem 3

The Lie algebra of the Lie point symmetries of the Ricci flow of equation (40), with ( F m , g can ) being an Einstein manifold non-homothetic to the Euclidean metric, is spanned by the following equations:

X 1 = t , X 2 = t t ψ 2 ψ + φ 2 φ , X k + 2 = ξ k ( x ) x + ψ ξ k x k ψ ,

where ξ k ( x ) are smooth functions of R n with k { 1 , , n } , such that

ξ i x j + ξ j x i = 0 and ξ i x i ξ j x j = 0 ,

with i j .

Proof

We begin by taking a symmetry of the Ricci flow

X = c 1 X 1 + c 2 X 2 + X 3 + + X n + m + 2 ,

where c 1 , c 2 R . We write X in its canonical form:

Q X = 1 i j n Q i j t , x , g p s , g p s t , g p s x r g i j .

Having in mind (40), we see that we need to apply the method on the restrictions g i i = 1 ψ 2 ( x , t ) , g i , j = 0 for i , j { 1 , , n } with i j and g k + n , l + n = φ ( x , t ) 2 ( g can ) i j with 1 k l m . Thus, we look for functions Q ψ and Q φ depending on x , t , φ , ψ , φ x , ψ x , ϕ t , ψ t such that

Q X = Q φ φ + Q ψ ψ ,

under the restrictions described above, we obtain the conditions

2 ψ 3 Q ψ = Q X ( g i i ) ( x , t , φ , ψ ) = c 2 2 ξ i x i 1 ψ 2 ( c 1 + c 2 t ) 2 ψ t ψ 3 ξ k 2 ψ x k ψ 3 ,

for all i { 1 , , n } . Then

(41) Q ψ = c 2 2 ξ i x i ψ ( c 1 + c 2 t ) ψ t ξ k ψ x i .

In addition, we have that

(42) Q ψ = ψ 3 2 Q X ( g i i ) ( x , t , φ , ψ ) = ψ 3 2 Q X ( g j j ) ( x , t , φ , ψ )

for i , j { 1 , , n } . Looking (41), we can see that (42) is valid if and only if

ξ i x i = ξ j x j ,

with i , j { 1 , , n } . For i , j { 1 , , n } with i j , we have

0 = Q X ( g i j ) ( x , t , φ , ψ ) = 1 ψ 2 ξ i x j + ξ j x i ,

so

ξ i x j + ξ j x i = 0 ,

for all i j .

For i { 1 , , n } and l { 1 , , m } . We obtain

0 = Q X ( g i , n + l ) ( x , t , φ , ψ ) = 1 ψ 2 ξ i x l + n + φ 2 ξ l + n x i .

In this case, ξ i x l + n represents the partial derivative of the function ξ i with respect to the coordinate l + n of the fiber. On the other hand, as the functions ξ i depend on the independent variables, we can conclude that

(43) ξ i ( x 1 , , x n ) and ξ l + n ( x n + 1 , , x n + m ) ,

for all i { 1 , , n } and l { 1 , , m } . For l , p { 1 , , m } , we obtain

(44) 2 φ Q φ ( g can ) l , p = Q X ( g n + l , n + p ) ( x , t , φ , ψ ) = φ 2 c 2 ( g can ) l p ( g can ) k l ξ n + k x n + p ( g can ) k p ξ n + k x n + l 2 ( c 1 + c 2 t ) φ φ t ( g can ) l p ξ n + k φ 2 ( g can ) l p x n + k 2 φ φ x s ξ s ( g can ) l , p ,

in this last equation k { 1 , , m } and s { 1 , , n } . Note that Q φ doesn’t depend of g can (the coefficients of the canonical metric are not constant), therefore

ξ n + k ( g can ) l p x n + k = 0 , for k { 1 , , m } .

Thus, ξ n + k = 0 , for , k { 1 , , m } so

Q φ = c 2 2 φ ( c 1 + c 2 t ) φ t ξ s φ x s .

Putting all of the above together

X = ( c 1 + 2 c 2 t ) t + ξ k x k + c 2 φ + ξ k x k ψ c 2 ψ ψ ,

with k { 1 , , n } , ξ i x i ξ j x j = 0 , and ξ i x j + ξ j x i = 0 when i j .□

Now we turn our attention to the special case where the fiber is the Euclidean space endowed with the canonical metric.

Theorem 4

The Lie algebra of the Lie point symmetries of the Ricci Flow of (40), with F m being R m endowed with the Euclidean metric, is spanned by the following equations:

X 1 = t , X 2 = t t ψ 2 ψ + φ 2 φ , X 3 = φ φ , X k + 3 = ξ k ( x ) x + ψ ξ k x k ψ ,

where ξ k ( x ) are smooth functions of R n , with k { 1 , , n } , such that

ξ i x j + ξ j x i = 0 and ξ i x i ξ j x j = 0 ,

with i j .

Proof

Following the previous proof, we have that Q ψ is still determined by the expression given by equation (41), ξ i with i { 1 , , n } , continues to satisfy the Cauchy Riemann equations, equation (43) is still valid, and equation (44) is transformed to

2 φ Q φ δ l , p = Q X ( g n + l , n + p ) ( x , t , φ , ψ ) = φ 2 c 2 δ l p ξ n + l x n + p ξ n + p x n + l 2 ( c 1 + c 2 t ) φ φ t δ l p 2 φ φ x s ξ s δ l , p

for l , p { 1 , , m } , from which it is possible to conclude that

Q φ = c 2 2 φ ( c 1 + c 2 t ) φ t ξ s φ x s + c 3 φ

where c 3 R . By combining all the above statements, we have

X = ( c 1 + 2 c 2 t ) t + ξ k x k + ( c 2 + c 3 φ ) φ + ξ k x k ψ c 2 ψ ψ ,

with k { 1 , , n } , ξ i x i ξ j x j = 0 , and ξ i x j + ξ j x i = 0 when i j .□

Corollary 1 is an immediate consequence of Theorem 4.

Corollary 1

The Lie algebra of the Lie point symmetries of the Ricci flow on R n with the conformal flow, i.e.,

g = 1 ψ 2 ( x , t ) d x i d x i ,

is spanned by the following set of vectors:

X 1 = t , X 2 = t t ψ 2 ψ , X 3 = ξ k ( x ) x + ψ ξ k x k ψ ,

where ξ k ( x ) are smooth functions of R n , with k { 1 , , n } , such that

ξ i x j + ξ j x i = 0 and ξ i x i ξ j x j = 0 ,

with i j .

Proof

The result follows immediately through the first part of the proof of Theorem 3.□

4.3 Ricci flow on doubly warped product manifolds

We now restrict our attention to flows evolving from a fixed initial metric that is cohomogeneous with respect to the natural S O ( p + 1 ) × S O ( q + 1 ) action. That is, we consider the doubly warped product metrics on S p + 1 × S q with p 2 and q 2 , i.e.,

g = χ ( x , t ) 2 d x d x + φ ( x , t ) 2 g S can p + ψ ( x , t ) 2 g S can q ,

where g S can n denotes the canonical metric of the sphere of dimension n . This kind of metric was extensively studied by Stolarski [27].

Theorem 5

The Lie algebra of the Lie point symmetries of the Ricci flow on S p + 1 × S q for metrics of the form

(45) g = χ 2 ( x , t ) d x d x + φ 2 ( x , t ) g S can p + ψ 2 ( x , t ) g S can q

is spanned by the following infinitesimal generators:

X 1 = t , X 2 = 2 t t + χ χ + φ φ + ψ ψ , X 3 = ξ ( x ) x χ ξ x χ ,

with ξ ( x ) being an arbitrary smooth function.

Proof

Similar to the proof of Theorem 3, we start by taking a symmetry from Theorem 1, i.e.,

X = c 1 X 1 + c 2 X 2 + 1 k p + q + 1 X 2 + k ,

where c 1 , , c 3 + p + i q R , and expressing it in canonical form as follows:

Q X = 1 i j 1 + p + q Q i j g i j .

Using the metric form (45), we see that we need to apply the method on the restrictions g 11 = χ 2 ( x , t ) , g 1 + i 1 , 1 + j 1 = φ 2 ( g S can p ) i 1 , j 1 for i 1 , j 1 { 1 , , p } and g 1 + p + i 2 , 1 + p + j 2 = ψ 2 ( g S can q ) i 2 , j 2 for i 2 , j 2 { 1 , , q } , while the rest of the coefficients of the metric are identically zero. Thus, we look for functions Q χ , Q φ , and Q ψ depending on x and t and of the dependent variables χ , φ , and ψ , as well as their first derivatives, such that

Q = Q φ φ + Q ψ ψ + Q χ χ ,

and under the restrictions described above and proceeding analogously to the proof of Theorem 3, we obtain

Q χ = χ t ( c 1 + c 2 t ) χ x ξ 1 ( x ) + c 2 χ 2 χ 2 ξ 1 x , Q φ = ( c 1 + c 2 t ) φ t ξ 1 φ x + c 2 2 φ , Q ψ = ( c 1 + c 2 t ) ψ t ξ 1 ψ x + c 2 2 ψ .

From this, we can conclude the statement of the theorem.□

5 Invariant solutions

5.1 Invariant solutions for warped product manifolds

First, we need to explicitly express the Ricci flow equations for the metric tensor (40). We know that

Ric g ( Y ˜ , Z ˜ ) = Ric g B ( Y , Z ) m φ B 2 φ ( Y , Z ) , Ric g ( Y ˜ , V ˜ ) = 0 , Ric g ( V ˜ , W ˜ ) = Ric g F ( V , W ) Δ B φ φ + B φ 2 φ 2 ( m 1 ) g ( V ˜ , W ˜ )

for all Y , Z L ( B ) and V , W L ( F ) , see Proposition 9.106 in [3] or [4]. On the other hand, consider ( p , q ) B n × F m , where { 1 B , , n B } and { 1 F , , m F } are coordinate bases of T p B and T q F , respectively. Then,

Ric g ( ˜ i B , ˜ j B ) = Ric g B ( i B , j B ) m φ B 2 φ ( i B , j B ) .

From conformal theory, we know that

Ric B ( i B , j B ) = ( n 2 ) ψ i j ψ + Δ ψ ψ ( n 1 ) ψ 2 ψ 2 δ i j

and

B 2 φ ( i B , j B ) = φ i j + ψ i φ j + ψ j φ i ψ ψ , φ ψ δ i j ,

see Theorem 1.159 in [3]. Thus,

Ric g ( ˜ i B , ˜ j B ) = ( n 2 ) ψ i j ψ + Δ ψ ψ ( n 1 ) ψ 2 ψ 2 δ i j m φ φ i j + ψ i φ j + ψ j φ i ψ ψ , φ ψ δ i j .

On the other hand,

Ric g ( ˜ i F , ˜ j F ) = Ric g F ( i F , j F ) Δ B φ φ + B φ 2 φ 2 ( m 1 ) g ( ˜ i F , ˜ j F ) = μ ( g can ) i j Δ B φ φ + B φ 2 φ 2 ( m 1 ) φ 2 ( g can ) i j .

Again, from conformal theory, we have

Ric g ( ˜ i F , ˜ j F ) = μ ψ 2 φ 2 Δ φ φ ( n 2 ) φ , ψ ψ φ + ( m 1 ) φ 2 φ 2 ( g can ) i j .

Therefore,

Ric g = ( n 2 ) ψ i j ψ + Δ ψ ψ ( n 1 ) ψ 2 ψ 2 δ i j m φ φ i j + ψ i φ j + ψ j φ i ψ ψ , φ ψ δ i j d x i d x j + μ φ ψ 2 Δ φ ( n 2 ) φ , ψ ψ ( m 1 ) ψ 2 φ 2 g can .

That being done, the family of metrics (40) satisfies the Ricci flow equation if and only if the following system is satisfied:

(46) ψ t ψ 3 δ i j = ( n 2 ) ψ i j ψ + Δ ψ ψ ( n 1 ) ψ 2 ψ 2 δ i j m φ φ i j + ψ i φ j + ψ j φ i ψ ψ , φ ψ δ i j , φ φ t = μ φ ψ 2 Δ φ ( n 2 ) φ , ψ ψ ( m 1 ) ψ 2 φ 2

for all i , j 1 , , n .

Then, we use the symmetries found in Section 4.2 to reduce the system of equation (46).

Note that { X 1 , X 2 } is the one-dimensional optimal system associated with the finite-dimensional sub algebra of Theorem 3, see Proposition 1. So, we only need to consider, for instance, the symmetry X 2 applying to it the one-parameter group of transformations generated by X 1 , in this case, t t + ε . That is, we shall employ the symmetry

X = ( 1 + 2 k t ) t k ψ ψ + k φ φ ,

where we have chosen ε = 1 2 k , k 0 . Therefore, the invariant surface conditions are as follows:

(47) k ψ + ( 1 + 2 k t ) ψ t = 0 , k φ ( 1 + 2 k t ) φ t = 0 .

Solving the system (47) by the method of characteristics, we obtain

ψ = 1 1 + 2 k t F ( x ) , φ = 1 + 2 k t G ( x ) ,

where F and G are smooth positive functions of x . Substituting these expressions of ψ and ϕ into the system (46), we obtain

(48) k F 2 δ i j = ( n 2 ) F i j F + Δ F F ( n 1 ) F 2 F 2 δ i j m G G i j + F i G j + F j G i F F , G F δ i j , k G 2 = μ G F 2 Δ G ( n 2 ) G , F F ( m 1 ) F 2 G 2

for all i , j { 1 , , n } . In view of the difficulty in solving the previous system, we are going to further reduce it by considering the case when the base has dimension 1 and the fiber is the canonical sphere, i.e., the product manifold M 1 × S m . By this reduction, system (48) becomes

(49) k F 2 = m G G x x + F x G x F , k G 2 = μ + k m G 2 + ( m 1 ) F 2 G x 2 ,

where μ = m 1 . Therefore, from the second one, we have

( F G x ) 2 = k m G 2 + 1 G x = ± 1 F k m G 2 + 1 .

Integrating it, we obtain

G ( x ) = ± m k sinh k m x 0 x d τ F ( τ ) , k > 0 , ± m k sin k m x 0 x d τ F ( τ ) , k < 0 .

Observe that this solution satisfies identically the first equation of the system (49). Furthermore, the solution suggests the following parameterization:

s ( x ) = x 0 x d τ F ( τ ) .

Therefore,

( 1 + 2 k t ) F 2 ( x ) d x d x = ( 1 + 2 k t ) d s d s .

So, having all the solutions of system (49) at hand, when k 0 , we arrive at the following solutions of the Ricci flow for the warped space M 1 × S n :

g = ( 1 + 2 k 2 t ) d s d s + m k 2 sinh 2 k 2 m s g can , g = ( 1 2 k 2 t ) d s d s + m k 2 sin 2 k 2 m s g can ,

when k 0 . Note that the first metric yields the standard hyperbolic m + 1 -space, see [16]. Both of them are well-known solutions in the bibliography, found independently of each other. This is an example of how symmetries can unify solutions found with different ad hoc methods.

5.2 Invariant solutions for doubly warped product manifolds

First, we need to express explicitly the Ricci flow equations. In this case, see [26],

(50) χ χ t = p φ φ x x φ x χ x χ + q ψ ψ x x ψ x χ x χ , φ φ t = ( p 1 ) + φ 2 χ 2 1 φ φ x x φ x χ x χ + ( p 1 ) φ x φ 2 + q φ x ψ x φ ψ , ψ ψ t = ( q 1 ) + ψ 2 χ 2 1 ψ ψ x x ψ x χ x χ + ( q 1 ) ψ x ψ 2 + p φ x ψ x ψ φ .

By the same argument as in Section 5.1, we will make use of the following symmetry:

X = ( 1 + 2 k t ) t + k χ χ + k φ φ + k ψ ψ ,

where k 0 . Therefore, the corresponding invariant surface conditions are

(51) Q χ = k χ ( 1 + 2 t k ) χ t = 0 , Q φ = k φ ( 1 + 2 k t ) φ t = 0 , Q ψ = k ψ ( 1 + 2 k t ) ψ t = 0 .

Solving system (51), we obtain

χ ( x , t ) = 1 + 2 t k F ( x ) , φ ( x , t ) = 1 + 2 t k G ( x ) , ψ ( x , t ) = 1 + 2 t k H ( x ) .

Inserting these expressions of χ , ψ , and φ into system (50), we arrive at the reduced system

(52) k F 2 = p G G G F F + q H H H F F , k G 2 = ( p 1 ) + G F 2 G G G F F G + ( p 1 ) G G 2 + q G H G H , k H 2 = ( q 1 ) + H F 2 H H H F F H + ( q 1 ) H H 2 + p G H G H .

Taking inspiration from the procedure to obtain the solutions for the case of single warped product manifolds, we start by parametrizing G and H by the arc length,

s ( x ) = x 0 x F ( τ ) d τ .

The system (52) becomes

(53) k = p G ( s ) G ( s ) + q H ( s ) H ( s ) , k = G ( s ) G ( s ) + ( p 1 ) G ( s ) 2 1 G ( s ) 2 + q G ( s ) H ( s ) G ( s ) H ( s ) , k = H ( s ) H ( s ) + ( q 1 ) H ( s ) 2 1 H ( s ) 2 + p G ( s ) H ( s ) G ( s ) H ( s ) .

Once more, by studying the form of the solutions found in Section 5.1, we start looking for particular solutions of the same form. This hypothesis was fruitful, the following functions

G = ± p + q k sin k p + q s and H = ± p + q k cos k p + q s , G = ± ( p 1 ) ( p + q ) k ( p + q 1 ) sin k p + q s and H = ± ( q 1 ) ( p + q ) k ( p + q 1 ) sin k p + q s ,

for k < 0 , and

G = ± ( p 1 ) ( p + q ) k ( p + q 1 ) sinh k p + q s and H = ± ( q 1 ) ( p + q ) k ( p + q 1 ) sinh k p + q s ,

for k > 0 , are indeed solutions of system (53).

Hence, our study yields the following special solutions:

g ( s , t ) = ( 1 2 k 2 t ) d s d s + ( 1 2 k 2 t ) p + q k 2 sin 2 k 2 p + q s g S can p + ( 1 2 k 2 t ) p + q k 2 cos 2 k 2 p + q s g S can q , g ( s , t ) = ( 1 2 k 2 t ) d s d s + ( 1 2 k 2 t ) ( p 1 ) ( p + q ) k 2 ( p + q 1 ) sin 2 k 2 p + q s g S can p + ( 1 2 k 2 t ) ( q 1 ) ( p + q ) k 2 ( p + q 1 ) sin 2 k 2 p + q s g S can q , g ( s , t ) = ( 1 + 2 k 2 t ) d s d s + ( 1 + 2 k 2 t ) ( p 1 ) ( p + q ) k 2 ( p + q 1 ) sinh 2 k 2 p + q s g S can p + ( 1 + 2 k 2 t ) ( q 1 ) ( p + q ) k 2 ( p + q 1 ) sinh 2 k 2 p + q s g S can q ,

where k 0 . To the best of our knowledge, these solutions are new in the literature.

6 Conclusion and discussion

In the present work, we determine the Lie point symmetries of the Ricci flow in arbitrary dimensions. By using their algebraic properties, we are able to “recycle” them in order to expeditiously obtain the Lie point symmetries of the Ricci flow for a particular family of metrics. A task that can be very challenging if one starts each time from scratch. Indeed, even in the reasonable case of studying a four-dimensional metric to obtain its Lie point symmetries one had to solve a system of determining equations that involves about two million PDE!

Unfortunately, classical symmetries, being a very broad notion that can be applied virtually to any kind of system of differential equations, have a very serious drawback; they usually will not help us to obtain the kinds of solutions that we are looking for. This is apparent in our case: the infinite-dimensional sub algebra merely points to the tensorial nature of our system, while the finite one says, on the one hand, that the system is autonomous and, on the other hand, ushers us to consider separable solutions. Solutions that are known to hold if and only if the initial manifold is an Einstein one. A fact that limits our choices regarding the initial condition to be considered.

One possible way out is by considering more “exotic” kinds of symmetries, like the nonclassical ones, see Section 9.3 in [11]. In a word, nonclassical symmetries are symmetries that are admitted only by specific families of solutions of a differential equation and not by the differential equation itself, as is the case with the classical symmetries. Although our initial studies were promising, none of the nonclassical symmetries found yielded nontrivial solutions. This do not come as a surprise since the systems of differential equations involved are now nonlinear. Therefore, a more exhausting classification must be carried over to obtain nonclassical symmetries that can yield solutions of the Ricci flow problem of some interest.

Acknowledgements

Enrique López thanks the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) - Brazil, for the financial support (Finance Code 001).

  1. Conflict of interest: The authors declare that they have no conflict of interest.

References

[1] S. Angenent and D. Knopf, An example of neckpinching for Ricci flow on Sn+1, Math. Res. Lett. 11 (2004), 493–518. 10.4310/MRL.2004.v11.n4.a8Search in Google Scholar

[2] I. Bakas, Ricci flows and infinite dimensional algebras, Fortschr. Phys. 52 (2004), 464–471. 10.1002/prop.200410131Search in Google Scholar

[3] A. Besse, Einstein Manifolds, Class. Math., Springer, Berlin, 2007. Search in Google Scholar

[4] R.L. Bishop and B. O’Neill, Manifolds of negative curvature, Trans. Amer. Math. Soc. 145 (1969), 1–49. 10.1090/S0002-9947-1969-0251664-4Search in Google Scholar

[5] Y. Bozhkov and E. Mitidieri, The Noether approach to Pokhozhaev’s identities, Mediterr. J. Math. 4 (2007), 383–405. 10.1007/s00009-007-0125-ySearch in Google Scholar

[6] Y. Bozhkov, E. Mitidieri, Lie symmetries and criticality of semilinear differential systems, Symmetry Integrability Geom Meth Appl. 3 (2007), 053, http://www.emis.de/journals/SIGMA/2007/053. 10.3842/SIGMA.2007.053Search in Google Scholar

[7] Y. Bozhkov and E. Mitidieri, Conformal killing vector fields and Rellich type identities on Riemannian manifolds, II, Mediterr. J. Math. 9 (2021), 1–20. 10.1007/s00009-011-0126-8Search in Google Scholar

[8] R. Cimpoiasu and R. Constantinescu, Symmetries and invariants for the 2D-Ricci flow model, J. Nonlinear Math. Phys. 13 (2006), 285–292. 10.2991/jnmp.2006.13.2.10Search in Google Scholar

[9] S. Dimas and D. Tsoubelis, SYM: A new symmetry-finding package for mathematica, in: The 10th International Conference on Modern Group Analysis, Nicosia, 2005, p. 64–70. Search in Google Scholar

[10] R. Hamilton, Three-manifolds with positive Ricci curvature, J. Differential Geom. 17 (1982), 255–306. 10.4310/jdg/1214436922Search in Google Scholar

[11] P. Hydon, Symmetry Methods for Differential Equations: A Beginner’s Guide, Cambridge University Press, Cambridge, 2000. 10.1017/CBO9780511623967Search in Google Scholar

[12] N. Ibragimov, Group analysis of ordinary differential equations and the invariance principle in mathematical physics (for the 150th anniversary of Sophus Lie), Russian Math. Surveys. 47 (1992), 89–156. 10.1070/RM1992v047n04ABEH000916Search in Google Scholar

[13] N. Ibragimov (Ed.), CRC Handbook of Lie Group Analysis of Differential Equations, vol. 1–3, CRC Press, Boca Raton, 1994–1996. Search in Google Scholar

[14] T. Ivey, The Ricci flow on radially symmetric R3, Comm. Partial Differential Equations. 19 (1994), 1481–1500. 10.1080/03605309408821061Search in Google Scholar

[15] A. Kristály, V. Rădulescu, and C. Varga, Variational Principles in Mathematical Physics, Geometry, and Economics: Qualitative Analysis of Nonlinear Equations and Unilateral Problems, Cambridge University Press, Cambridge, 2010. 10.1017/CBO9780511760631Search in Google Scholar

[16] L. Ma and X. Xu, Ricci flow with hyperbolic warped product metrics, Math. Nachr. 284 (2011), 739–746. 10.1002/mana.200710226Search in Google Scholar

[17] L. Marchildon, Lie symmetries of Einstein’s vacuum equations in N dimensions, J. Nonlinear Math. Phys. 5 (1998), 68–81. 10.2991/jnmp.1998.5.1.7Search in Google Scholar

[18] S. Meleshko, Methods for Constructing Exact Solutions of Partial Differential Equations: Mathematical and Analytical Techniques with Applications to Engineering, Springer Science, New York, 2005. Search in Google Scholar

[19] T. Oliynyk and E. Woolgar, Rotationally symmetric Ricci flow on asymptotically flat manifolds, Comm. Anal. Geom. 15 (2007), 535–568. 10.4310/CAG.2007.v15.n3.a4Search in Google Scholar

[20] P. Olver, Applications of Lie Groups to Differential Equations, 2nd ed., Springer-Verlag, New York, 1993. 10.1007/978-1-4612-4350-2Search in Google Scholar

[21] P. Olver, Equivalence, Invariants, and Symmetry, Cambridge University Press, Cambridge, 1995. 10.1017/CBO9780511609565Search in Google Scholar

[22] L. Ovsiannikov, Group Analysis of Differential Equations, Academic Press, New York, 2014. Search in Google Scholar

[23] B. O’neill, Semi-Riemannian Geometry With Applications to Relativity, Academic Press, Cambridge, Massachusetts, 1983. Search in Google Scholar

[24] N. Papageorgiou, V. Rădulescu, and D. Repovš, Nonlinear Analysis - Theory and Methods, Springer Monographs in Mathematics, Springer, Cham, 2019. 10.1007/978-3-030-03430-6Search in Google Scholar

[25] G. Perelman, Finite extinction time for the solutions to the Ricci flow on certain three-manifolds, 2008. https://arxiv.org/abs/math/0307245. 10.3731/topologica.1.005Search in Google Scholar

[26] P. Petersen, Riemannian Geometry, 3rd ed., Graduate Texts in Mathematics, vol. 171, Springer, Cham, 201610.1007/978-3-319-26654-1Search in Google Scholar

[27] M. Stolarski, Curvature Blow-up in Doubly-warped Product Metrics Evolving by Ricci Flow, Ph.D. Thesis, University of Texas at Austin, Austin, 2019. Search in Google Scholar

[28] Q. Xing and X. Gao, A mathematical interpretation of Hawking’s Black Hole Theory by Ricci flow, Z. Angew. Math. Phys. 5 (2017), 321–328. 10.4236/jamp.2017.52029Search in Google Scholar

[29] J. Wang, Symmetries and solutions to geometrical flows, Sci. China Math. 56 (2013), 1689–1704. 10.1007/s11425-013-4635-8Search in Google Scholar

Received: 2023-01-07
Accepted: 2023-06-13
Published Online: 2023-10-04

© 2023 the author(s), published by De Gruyter

This work is licensed under the Creative Commons Attribution 4.0 International License.

Articles in the same Issue

  1. Regular Articles
  2. On sufficient “local” conditions for existence results to generalized p(.)-Laplace equations involving critical growth
  3. On the critical Choquard-Kirchhoff problem on the Heisenberg group
  4. On the local behavior of local weak solutions to some singular anisotropic elliptic equations
  5. Viscosity method for random homogenization of fully nonlinear elliptic equations with highly oscillating obstacles
  6. Double-phase parabolic equations with variable growth and nonlinear sources
  7. Logistic damping effect in chemotaxis models with density-suppressed motility
  8. Bifurcation diagrams of one-dimensional Kirchhoff-type equations
  9. Standing wave solution for the generalized Jackiw-Pi model
  10. Weak-strong uniqueness and energy-variational solutions for a class of viscoelastoplastic fluid models
  11. Eigenvalue inequalities for the buckling problem of the drifting Laplacian of arbitrary order
  12. Homoclinic solutions for a differential inclusion system involving the p(t)-Laplacian
  13. Equivalence between a time-fractional and an integer-order gradient flow: The memory effect reflected in the energy
  14. Bautin bifurcation with additive noise
  15. Small solitons and multisolitons in the generalized Davey-Stewartson system
  16. Nonstationary Poiseuille flow of a non-Newtonian fluid with the shear rate-dependent viscosity
  17. A global compactness result with applications to a Hardy-Sobolev critical elliptic system involving coupled perturbation terms
  18. On a strongly damped semilinear wave equation with time-varying source and singular dissipation
  19. Singularity for a nonlinear degenerate hyperbolic-parabolic coupled system arising from nematic liquid crystals
  20. Stability of stationary solutions to the three-dimensional Navier-Stokes equations with surface tension
  21. Uniform decay estimates for the semi-linear wave equation with locally distributed mixed-type damping via arbitrary local viscoelastic versus frictional dissipative effects
  22. Symmetry and nonsymmetry of minimal action sign-changing solutions for the Choquard system
  23. Periodic and quasi-periodic solutions of a four-dimensional singular differential system describing the motion of vortices
  24. Multiple nontrivial solutions of superlinear fractional Laplace equations without (AR) condition
  25. Existence and blow-up of solutions in Hénon-type heat equation with exponential nonlinearity
  26. Existence and concentration behavior of positive solutions to Schrödinger-Poisson-Slater equations
  27. On the fractional Korn inequality in bounded domains: Counterexamples to the case ps < 1
  28. Gradient estimates for nonlinear elliptic equations involving the Witten Laplacian on smooth metric measure spaces and implications
  29. Dynamical analysis of a reaction–diffusion mosquito-borne model in a spatially heterogeneous environment
  30. Existence and multiplicity of solutions for a quasilinear system with locally superlinear condition
  31. Nontrivial solution for Klein-Gordon equation coupled with Born-Infeld theory with critical growth
  32. Modeling Wolbachia infection frequency in mosquito populations via a continuous periodic switching model
  33. Infinitely many localized semiclassical states for nonlinear Kirchhoff-type equation
  34. Fujita-type theorems for a quasilinear parabolic differential inequality with weighted nonlocal source term
  35. Approximations of center manifolds for delay stochastic differential equations with additive noise
  36. Periodic solutions to a class of distributed delay differential equations via variational methods
  37. Groundstate for the Schrödinger-Poisson-Slater equation involving the Coulomb-Sobolev critical exponent
  38. Existence of nontrivial solutions for the Klein-Gordon-Maxwell system with Berestycki-Lions conditions
  39. Global Sobolev regular solution for Boussinesq system
  40. Normalized solutions for the p-Laplacian equation with a trapping potential
  41. Nonlinear elliptic–parabolic problem involving p-Dirichlet-to-Neumann operator with critical exponent
  42. Blow-up for compressible Euler system with space-dependent damping in 1-D
  43. High energy solutions of general Kirchhoff type equations without the Ambrosetti-Rabinowitz type condition
  44. On the dynamics of grounded shallow ice sheets: Modeling and analysis
  45. A survey on some vanishing viscosity limit results
  46. Blow-up for logarithmic viscoelastic equations with delay and acoustic boundary conditions
  47. Generalized Liouville theorem for viscosity solutions to a singular Monge-Ampère equation
  48. Front propagation in a double degenerate equation with delay
  49. Positive solutions for a class of singular (pq)-equations
  50. Higher integrability for anisotropic parabolic systems of p-Laplace type
  51. The Poincaré map of degenerate monodromic singularities with Puiseux inverse integrating factor
  52. On a system of multi-component Ginzburg-Landau vortices
  53. Existence and concentration of solutions to Kirchhoff-type equations in ℝ2 with steep potential well vanishing at infinity and exponential critical nonlinearities
  54. Multiplicity results for fractional Schrödinger-Kirchhoff systems involving critical nonlinearities
  55. On double phase Kirchhoff problems with singular nonlinearity
  56. Estimates for eigenvalues of the Neumann and Steklov problems
  57. Nodal solutions with a prescribed number of nodes for the Kirchhoff-type problem with an asymptotically cubic term
  58. Dirichlet problems involving the Hardy-Leray operators with multiple polars
  59. Incompressible limit for compressible viscoelastic flows with large velocity
  60. Characterizations for the genuine Calderón-Zygmund operators and commutators on generalized Orlicz-Morrey spaces
  61. Non-local gradients in bounded domains motivated by continuum mechanics: Fundamental theorem of calculus and embeddings
  62. Noncoercive parabolic obstacle problems
  63. Touchdown solutions in general MEMS models
  64. Existence and nonexistence of nontrivial solutions for the Schrödinger-Poisson system with zero mass potential
  65. Short-time existence of a quasi-stationary fluid–structure interaction problem for plaque growth
  66. Fine bounds for best constants of fractional subcritical Sobolev embeddings and applications to nonlocal PDEs
  67. Symmetries of Ricci flows
  68. Asymptotic behavior of solutions of a second-order nonlinear discrete equation of Emden-Fowler type
  69. On the topological gradient method for an inverse problem resolution
  70. Supersolutions to nonautonomous Choquard equations in general domains
  71. Uniform complex time heat Kernel estimates without Gaussian bounds
  72. Global existence for time-dependent damped wave equations with nonlinear memory
  73. Normalized solutions for a critical fractional Choquard equation with a nonlocal perturbation
  74. Reversed Hardy-Littlewood-Sobolev inequalities with weights on the Heisenberg group
  75. Lamé system with weak damping and nonlinear time-varying delay
  76. Global well posedness for the semilinear edge-degenerate parabolic equations on singular manifolds
  77. Blowup in L1(Ω)-norm and global existence for time-fractional diffusion equations with polynomial semilinear terms
  78. Boundary regularity results for minimisers of convex functionals with (p, q)-growth
  79. Parametric singular double phase Dirichlet problems
  80. Special Issue on Nonlinear analysis: Perspectives and synergies
  81. Editorial to Special issue “Nonlinear analysis: Perspectives and synergies”
  82. Identification of discontinuous parameters in double phase obstacle problems
  83. Global boundedness to a 3D chemotaxis-Stokes system with porous medium cell diffusion and general sensitivity
  84. On regular solutions to compressible radiation hydrodynamic equations with far field vacuum
  85. On Cauchy problem for fractional parabolic-elliptic Keller-Segel model
  86. The evolution of immersed locally convex plane curves driven by anisotropic curvature flow
  87. Stability of combination of rarefaction waves with viscous contact wave for compressible Navier-Stokes equations with temperature-dependent transport coefficients and large data
  88. On concave perturbations of a periodic elliptic problem in R2 involving critical exponential growth
  89. Existence of solutions for a double-phase variable exponent equation without the Ambrosetti-Rabinowitz condition
Downloaded on 11.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/anona-2023-0106/html
Scroll to top button