Startseite A new monotone quantity in mean curvature flow implying sharp homotopic criteria
Artikel Open Access

A new monotone quantity in mean curvature flow implying sharp homotopic criteria

  • Chung-Jun Tsai , Mao-Pei Tsui und Mu-Tao Wang EMAIL logo
Veröffentlicht/Copyright: 30. September 2025

Abstract

A new monotone quantity in graphical mean curvature flows of higher codimensions is identified in this work. The submanifold deformed by the mean curvature flow is the graph of a map between Riemannian manifolds, and the quantity is monotone increasing under the area-decreasing condition of the map. The flow provides a natural homotopy of the corresponding map and leads to sharp criteria regarding the homotopic class of maps between complex projective spaces, and maps from spheres to complex projective spaces, among others.

1 Introduction

We start with the following theorem and its corollary proved in [17].

Theorem 1.1

Let f : ( S n , g std ) ( S m , g std ) be an area-decreasing smooth map between standard unit spheres of dimensions n , m 2 . Let Γ f be the graph of 𝑓 as a submanifold of S n × S m ; then the mean curvature flow of Γ f in S n × S m exists smoothly and remains graphical for all time, and converges smoothly to the graph of a constant map.

Corollary 1.2

Any area-decreasing map from ( S n , g std ) to ( S m , g std ) with dimensions n , m 2 must be homotopically trivial.

Explicit examples of non-trivial homotopy classes in π n ( S m ) when n m inspire interest to find local geometric conditions that determine the homotopy class of a map from S n to S m . The “stretch” or “1-dilatation” of such a map has been considered by Lawson [8], Hsu [7], Gromov [3], and DeTurck–Gluck–Storm [2]. In [4], Gromov proposed to study how the 𝑘-dilation of a map S n S m is related to its homotopy class. In particular, he showed in [4, page 179 and corollary on page 230] (see also [5, Section 2]) that there exists a constant ε ( n , m ) > 0 such that a map f : ( S n , g std ) ( S m , g std ) with 2-dilation (see below) less than ε ( n , m ) everywhere must be homotopically trivial. It was conjectured that ε ( n , m ) can be chosen to be one, as a map with 1-dilation less than one is clearly homotopically trivial. Unbeknown to the authors of [17], the conjecture was affirmed by Corollary 1.2: a map f : ( S n , g std ) ( S m , g std ) with 2-dilation less than one is exactly an area-decreasing map. By contrast, such a result does not hold for the 3-dilation by the work of Guth [6]; see stronger results about 𝑘-dilations therein.

Area-decreasing maps in the mean curvature flow arise in a rather different context. It was shown in [20] that the Gauss maps of a mean curvature flow form a harmonic map heat flow, and area-decreasing maps correspond to a convex region in the Grassmannian. It was proved in [17] that the area-decreasing condition is preserved by the mean curvature flow under curvature assumptions of the domain and the target manifolds (see [18, 21] for the case when the target space is 2-dimensional). There were several later extensions [9, 16, 1] of the results in [17] to allow more general curvature conditions, all based on the study of the evolution equation of the symmetric 2-tensor S [ 2 ] (see [17, Section 5], or Section 3 of this article), whose positivity corresponds to the area-decreasing condition. In this article, we consider the evolution equation (3.6) of the logarithmic determinant of S [ 2 ] . This novel formulation significantly improves our understanding of the ambient curvature terms involved in the evolution equation, which leads to the sharp homotopic criteria mentioned in the title. In particular, unlike previous works whose assumptions rely on the sectional curvatures of the domain and target manifolds, Theorem 1.4 relies on the Ricci curvatures. This enables us to prove Corollaries 1.5 and 1.6 regarding the complex projective spaces.

It is best to describe the area-decreasing condition and the new monotone quantity in terms of singular values. Recall, for a C 1 map f : ( Σ 1 , g 1 ) ( Σ 2 , g 2 ) between Riemannian manifolds, the singular values of d f at a point p Σ 1 are the non-negative square roots of eigenvalues of ( d f | p ) T ( d f | p ) . Suppose Σ 1 is of dimension 𝑛 and the singular values are λ i , i = 1 , , n ; then the 2-dilation of 𝑓 is defined to be max i < j λ i λ j . A map is said to be area-decreasing if the 2-dilation is less than one everywhere, or λ i λ j < 1 for any i < j .

This article considers the mean curvature flow of the graph of 𝑓, Γ f , as a submanifold of Σ 1 × Σ 2 . Under suitable curvature assumptions on Σ 1 and Σ 2 and the area-decreasing condition, the flow remains the graph of a family of maps f t . The new monotone quantity is

log ( 1 i < j n 2 ( 1 λ i 2 λ j 2 ) ( 1 + λ i 2 ) ( 1 + λ j 2 ) ) ,

where λ i are the singular values of f t .

We state the main theorems in the following.

Theorem 1.3

Let ( Σ 1 , g 1 ) and ( Σ 2 , g 2 ) be two compact, Riemannian manifolds. Let n = dim Σ 1 , m = dim Σ 2 , and assume n m 2 . Suppose that their sectional curvatures satisfy

κ 1 1 and κ 2 < 2 n m 1 m 1 .

Then, for any area-decreasing map f : Σ 1 Σ 2 , the mean curvature flow in Σ 1 × Σ 2 starting from Γ f exists for all time, and remains graphical. Moreover, it converges smoothly to the graph of a constant map as t .

The condition κ 1 1 means that κ 1 ( σ ) 1 for any σ Λ 2 ( T Σ 1 ) .

Theorem 1.4

Let ( Σ 1 , g 1 ) and ( Σ 2 , g 2 ) be two compact Riemannian manifolds. Assume dim Σ 1 dim Σ 2 2 . Suppose that their Ricci and sectional curvatures satisfy

Ric 1 g 1 Ric 2 g 2 and κ 1 + κ 2 > 0 .

When dim Σ 1 > dim Σ 2 , we assume additionally that κ 1 0 . Then the same conclusion as Theorem 1.3 holds true.

The first curvature condition means that Ric 1 ( u , u ) Ric 2 ( v , v ) for any unit vectors u T Σ 1 and v T Σ 2 . The second curvature condition means that κ 1 ( σ 1 ) + κ 2 ( σ 2 ) > 0 for any σ 1 Λ 2 ( T Σ 1 ) and σ 2 Λ 2 ( T Σ 2 ) .

We state some corollaries of homotopic criteria which follow from the above theorems. Consider the complex projective spaces endowed with the Fubini–Study metrics g FS . Our theorems imply the following results.

Corollary 1.5

Suppose that n m 1 . Any area-decreasing map from ( C P n , g FS ) to ( C P m , g FS ) must be homotopically trivial.

It is known that the homotopy class of a map in [ C P n , C P m ] is determined by the induced map on the cohomology groups [13] for n < m , and by the degree for n = m (see [11]). Since the identity map is not homotopically trivial, the area-decreasing condition in Corollary 1.5 is sharp.

Another corollary concerns π 2 n + 1 ( C P n ) , which is isomorphic to π 2 n + 1 ( S 2 n + 1 ) Z by the long exact sequence induced by the Hopf fibration S 1 S 2 n + 1 C P n .

Corollary 1.6

Any smooth map f : ( S 2 n + 1 , g std ) ( C P n , g FS ) with 2-dilation less than 2 n 2 n + 1 everywhere must be homotopically trivial.

The Hopf fibration S 2 n + 1 C P n represents the generator of π 2 n + 1 ( C P n ) Z . It is a straightforward computation to show that the Hopf fibration ( S 2 n + 1 , g std ) ( C P n , g FS ) has singular values 0 (of multiplicity 1) and 1 (of multiplicity 2 n ). With this understood, Corollary 1.6 is sharp as n .

The paper is organized as follows. In Section 2, we review some properties of graphical mean curvature flow. In Section 3, the evolution equation of the logarithmic determinant of S [ 2 ] is derived. Section 4 is devoted to prove the main theorems, Theorems 1.3 and 1.4. In Section 5, we explain the implications of the main theorems.

2 Preliminaries

For two Riemannian manifolds ( Σ 1 , g 1 ) and ( Σ 2 , g 2 ) , consider the symmetric 2-tensor 𝑆 on Σ 1 × Σ 2 defined by

S ( X , Y ) = π 1 ( X ) , π 1 ( Y ) π 2 ( X ) , π 2 ( Y )

for any X , Y T ( Σ 1 × Σ 2 ) . It is not hard to see that 𝑆 is parallel with respect to the Levi-Civita connection of the product metric. Let n = dim Σ 1 and m = dim Σ 2 .

For a map f : Σ 1 Σ 2 , let Γ f be the graph of 𝑓. As in [17, Section 4], one can apply the singular value decomposition at any p Σ 1 to d f | p to find orthonormal bases { e i } i = 1 n for T ( p , f ( p ) ) Γ f and { e α } α = n + 1 n + m for ( T ( p , f ( p ) ) Γ f ) such that

(2.1) S ( e i , e j ) = 1 λ i 2 1 + λ i 2 δ i j , S ( e i , e α ) = 2 λ i 1 + λ i 2 δ i ( α n ) , S ( e α , e β ) = ( 1 λ α n 2 ) 1 + λ α n 2 δ α β ,

where { λ i } i = 1 n are the singular values of d f | p . Indeed, there exist orthonormal bases { u i } i = 1 n for T p Σ 1 and { v j } j = 1 m for T f ( p ) Σ 2 such that

(2.2) e i = 1 1 + λ i 2 ( u i + λ i v i ) for i { 1 , , n } , e α = 1 1 + λ α n 2 ( λ α n u α n + v α n ) for α { n + 1 , , n + m } .

It is convenient to set λ i to be 0 when i > min { n , m } .

Suppose that { Γ f t } is a mean curvature flow. Let F t be the embedding

x Σ 1 ( x , f t ( x ) ) Σ 1 × Σ 2 .

Consider the tensor F t S on Σ 1 , and denote its components by S i j . From [17, (3.7)], F t S obeys the following equation along the mean curvature flow:

(2.3) ( d d t Δ ) S i j = R k i k α S α j + R k j k α S α i + h α k h α k i S j + h α k h α k j S i 2 h α k i h β k j S α β ,

which is in terms of an evolving orthonormal frame. Here, the Laplacian Δ is the rough Laplacian for 2-tensors of Γ f ( t ) , and R k i k α = R ( e k , e α ) e i , e k is the coefficient of the Riemann curvature tensor of Σ 1 × Σ 2 .

The (spatial) gradient of F t S is computed on [17, p. 1121] as follows:

(2.4) S i j ; k = h α k i S α j + h α k j S α i ,

where h α k i = ̄ e k e i , e α is the coefficient of the second fundamental form of Γ f ( t ) in Σ 1 × Σ 2 .

In terms of the frame { e i } i = 1 n (2.1), F t S is diagonal, and

( S i i + S j j ) = 2 ( 1 λ i 2 λ j 2 ) ( 1 + λ i 2 ) ( 1 + λ j 2 ) .

It follows that the 2-positivity[1] of F t S is equivalent to f ( t ) being area-decreasing.

3 The evolution equation of log ( det S [ 2 ] )

In [17, Section 5], the authors introduce a tensor S [ 2 ] from F t S , which can be viewed as a symmetric endomorphism on Λ 2 ( T Γ f ) . With respect to an orthonormal frame,

(3.1) S ( i j ) ( k ) [ 2 ] = S i k δ j + S j δ i k S i δ j k S j k δ i

for any i < j and k < .

Suppose that, at a space-time point 𝑝, 𝑆 is diagonal, S i j | p = S i i δ i j . As in (2.1), at this point,

(3.2) S i i = 1 λ i 2 1 + λ i 2 , and set C i i = 2 λ i 1 + λ i 2 .

Note that S i i 2 + C i i 2 = 1 . The fact that S [ 2 ] is positive is equivalent to S i i + S j j > 0 for i < j . We need the following calculation lemma for the evolution of S i i + S j j , i < j .

Lemma 3.1

Suppose that S [ 2 ] is positive definite. For any i , j with 1 i < j n ,

( S i i + S j j ) 1 ( d d t Δ ) ( S i i + S j j ) + 1 2 ( S i i + S j j ) 2 | ( S i i + S j j ) | 2 2 k [ h ( n + j ) k i 2 + h ( n + i ) k j 2 + h ( n + i ) k i 2 + h ( n + j ) k j 2 + > n ( h ( n + ) k i 2 + h ( n + ) k j 2 ) ] 2 ( S i i + S j j ) 1 k [ R k i k ( n + i ) C i i + R k j k ( n + j ) C j j ] + 2 ( S i i + S j j ) 2 k ( h ( n + i ) k i C j j + h ( n + j ) k j C i i ) 2 .

In the above expression, the summation index 𝑘 is from 1 to 𝑛.

Proof

By (2.3) and (2.1),

( d d t Δ ) S i i = 2 k , α R k i k α S α i + 2 α , k h α k i 2 S i i 2 α , k h α k i 2 S α α = 2 k R k i k ( n + i ) C i i + 2 k , h ( n + ) k i 2 S i i + 2 k , h ( n + ) k i 2 S ,

where 𝑘 is from 1 to 𝑛, and ℓ is from 1 to 𝑚. It follows that

( d d t Δ ) ( S i i + S j j ) = 2 k [ R k i k ( n + i ) C i i + R k j k ( n + j ) C j j ] + 2 k , h ( n + ) k i 2 ( S i i + S ) + 2 k , h ( n + ) k j 2 ( S j j + S ) .

On the other hand, by (2.4) and (2.1),

(3.3) | ( S i i + S j j ) | 2 = 4 k ( h ( n + i ) k i C i i + h ( n + j ) k j C j j ) 2 .

Therefore,

( S i i + S j j ) 1 ( d d t Δ ) ( S i i + S j j ) + 1 2 ( S i i + S j j ) 2 | ( S i i + S j j ) | 2 = 2 ( S i i + S j j ) 1 k [ R k i k ( n + i ) C i i + R k j k ( n + j ) C j j ] + 2 ( S i i + S j j ) 1 [ I + II ] ,

where

I = k , h ( n + ) k i 2 ( S i i + S ) + k , h ( n + ) k j 2 ( S j j + S ) , II = ( S i i + S j j ) 1 k [ h ( n + i ) k i C i i + h ( n + j ) k j C j j ] 2 .

We claim that

(3.4) I + II ( S i i + S j j ) k [ h ( n + j ) k i 2 + h ( n + i ) k j 2 + h ( n + i ) k i 2 + h ( n + j ) k j 2 + > n ( h ( n + ) k i 2 + h ( n + ) k j 2 ) ] + ( S i i + S j j ) 1 k ( h ( n + i ) k i C j j + h ( n + j ) k j C i i ) 2 ,

which proves the lemma.

By taking out = i , j , the term I can be rewritten as

(3.5) I = I + k [ h ( n + j ) k i 2 + h ( n + i ) k j 2 ] ( S i i + S j j ) + k { i , j } h ( n + ) k i 2 ( S i i + S ) + k { i , j } h ( n + ) k j 2 ( S j j + S ) ,

where

I = 2 k [ h ( n + i ) k i 2 S i i + h ( n + j ) k j 2 S j j ] .

If m > n , consider the > n terms in the last two summands (both are non-negative due to our assumption) in (3.5). When > n , S = 1 . It follows that

S i i + S S i i + S j j and S j j + S S i i + S j j .

Then

k > n h ( n + ) k i 2 ( S i i + S ) + k > n h ( n + ) k j 2 ( S j j + S ) ( S i i + S j j ) k > n [ h ( n + ) k i 2 + h ( n + ) k j 2 ] .

Therefore,

I I + ( S i i + S j j ) k [ h ( n + j ) k i 2 + h ( n + i ) k j 2 + > n ( h ( n + ) k i 2 + h ( n + ) k j 2 ) ] .

Grouping I and II yields

I + II = 2 k [ h ( n + i ) k i 2 S i i + h ( n + j ) k j 2 S j j ] + ( S i i + S j j ) 1 k [ h ( n + i ) k i C i i + h ( n + j ) k j C j j ] 2 = ( S i i + S j j ) k [ h ( n + i ) k i 2 + h ( n + j ) k j 2 ] + ( S i i + S j j ) 1 k [ h ( n + i ) k i C j j + h ( n + j ) k j C i i ] 2 ,

where the last equality follows from the identity

2 S i i + ( S i i + S j j ) 1 C i i 2 = ( S i i + S j j ) + ( S i i + S j j ) 1 C j j 2 .

Putting these together proves claim (3.4). ∎

We are now ready to derive the evolution equation of log ( det S [ 2 ] ) .

Theorem 3.2

Suppose that S [ 2 ] is positive definite. The function log ( det S [ 2 ] ) satisfies

(3.6) ( d d t Δ ) log ( det S [ 2 ] ) 2 | A | 2 + 2 ( n 2 ) 1 i n 1 k n h ( n + i ) k i 2 + 2 R S + 2 Q S ,

where

(3.7) R S = 1 i < j n ( S i i + S j j ) 1 1 k n [ R k i k ( n + i ) C i i + R k j k ( n + j ) C j j ] ,
Q S = 1 i < j n ( S i i + S j j ) 2 1 k n ( h ( n + i ) k i C i i + h ( n + j ) k j C j j ) 2 + 1 i < j n ( S i i + S j j ) 2 1 k n ( h ( n + i ) k i C j j + h ( n + j ) k j C i i ) 2 .
In particular, Q S 0 .

Proof

Suppose that S [ 2 ] is positive definite, and denote its inverse by 𝑄. Note that Q ( i j ) ( k ) | p = ( S i i + S j j ) 1 δ i k δ j . Consider the function log det ( S [ 2 ] ) . At 𝑝, this quantity is

1 i < j n log ( S i i + S j j ) .

For brevity, write A , B for the indices of S [ 2 ] . We compute

(3.8) ( d d t Δ ) log ( det S [ 2 ] ) = Q A B [ ( d d t Δ ) S A B [ 2 ] ] ( Q A B ) ( S A B [ 2 ] ) = Q A B [ ( d d t Δ ) S A B [ 2 ] ] + Q A C ( S C D [ 2 ] ) Q D B ( S A B [ 2 ] ) = at p A Q A A [ ( d d t Δ ) S A A [ 2 ] ] + A , B Q A A Q B B | S A B [ 2 ] | 2 .

The first term on the last line of (3.8) is

(3.9) 1 i < j n ( S i i + S j j ) 1 ( d d t Δ ) ( S i i + S j j ) .

For the second term, it follows from definition (3.1) that S ( i j ) ( k ) [ 2 ] 0 when i , j , k , are all distinct. If i , j , k are all distinct, S A B [ 2 ] = ± S k for A = ( i j ) or ( j i ) , and B = ( i k ) or ( k i ) . With this understood, the second term of (3.8) becomes

(3.10) 1 i < j n ( S i i + S j j ) 2 | ( S i i + S j j ) | 2 + 2 1 i n 1 j < k n j i , k i ( S i i + S j j ) 1 ( S i i + S k k ) 1 | S j k | 2 .

Since the last summation is taken over j < k , there is a coefficient 2.

With (3.8), (3.9) and (3.10), one finds that

( d d t Δ ) log ( det S [ 2 ] ) 1 i < j n [ ( S i i + S j j ) 1 ( d d t Δ ) ( S i i + S j j ) + 1 2 ( S i i + S j j ) 2 | ( S i i + S j j ) | 2 ] + 1 2 1 i < j n ( S i i + S j j ) 2 | ( S i i + S j j ) | 2 .

By invoking Lemma 3.1 and (3.3),

( d d t Δ ) log ( det ( S [ 2 ] ) ) 2 R S 2 Q S 2 1 i < j n 1 k n [ h ( n + j ) k i 2 + h ( n + i ) k j 2 + h ( n + i ) k i 2 + h ( n + j ) k j 2 + 2 > n ( h ( n + ) k i 2 + h ( n + ) k j 2 ) ] 2 | A | 2 + 2 ( n 2 ) 1 i n 1 k n h ( n + i ) k i 2 .

This finishes the proof of this theorem. ∎

3.1 Some estimates

Consider the function

(3.11) φ ( λ 1 , , λ n ) = log ( 1 i < j n 1 λ i 2 λ j 2 ( 1 + λ i 2 ) ( 1 + λ j 2 ) )

defined where λ i 2 λ j 2 < 1 for any i j . Since

0 < 1 λ i 2 λ j 2 ( 1 + λ i 2 ) ( 1 + λ j 2 ) 1 ,

𝜑 always takes value within ( , 0 ] . It turns out that one can conclude the upper bound of λ i 2 from the lower bound of 𝜑.

Lemma 3.3

Consider the function φ ( λ 1 , , λ n ) defined by (3.11) on

{ ( λ 1 , , λ n ) R n : λ i 2 λ j 2 < 1 for any i j } .

Suppose that φ ( λ 1 , , λ n ) δ for some δ > 0 . Then

  1. λ i 2 e δ 1 for all 𝑖;

  2. λ i 2 λ j 2 ( e δ 1 ) / ( e δ + 1 ) for any i j ;

  3. there exists c 1 = c 1 ( n , δ ) > 0 such that | φ ( λ 1 , , λ n ) | c 1 i = 1 n λ i 2 .

Proof

It follows from φ δ that

1 i < j n 1 λ i 2 λ j 2 ( 1 + λ i 2 ) ( 1 + λ j 2 ) e δ .

Since 0 < 1 λ i 2 λ j 2 1 and 1 + λ i 2 1 for every i j , one finds that

e δ ( 1 λ i 2 λ j 2 ) ( 1 + λ i 2 ) ( 1 + λ j 2 ) for every i j .

Hence

(3.12) e δ 1 λ i 2 + λ j 2 and e δ 1 e δ + 1 λ i 2 λ j 2 .

Since log ( 1 x ) x 1 x for 0 x < 1 and log ( 1 + x ) x for x > 1 ,

φ ( λ 1 , , λ n ) = 1 i < j n [ log ( 1 λ i 2 λ j 2 ) + log ( 1 + λ i 2 ) + log ( 1 + λ j 2 ) ] 1 i < j n [ λ i 2 λ j 2 1 λ i 2 λ j 2 + λ i 2 + λ j 2 ] c 1 ( n , δ ) i = 1 n λ i 2 ,

where the last inequality uses (3.12). This finishes the proof of the lemma. ∎

For an area-decreasing map 𝑓, S [ 2 ] on Γ f is positive definite. In terms of the singular values (2.1),

log ( det S [ 2 ] ) = n ( n 1 ) 2 log 2 + log ( 1 i < j n 2 ( 1 λ i 2 λ j 2 ) ( 1 + λ i 2 ) ( 1 + λ j 2 ) ) .

The above discussion can be used to bound the gradient of log ( det S [ 2 ] ) .

Lemma 3.4

There exists a constant c 2 = c 2 ( n ) > 0 with the following significance. For an area-decreasing map f : Σ 1 Σ 2 , suppose that, at a point p Σ 1 , there exists a δ > 0 such that

log ( det S [ 2 ] ) n ( n 1 ) 2 log 2 δ at p .

Then | log ( det S [ 2 ] ) | 2 c 2 e 4 δ ( e δ 1 ) | A | 2 at 𝑝.

Proof

Since

det S [ 2 ] = ( det S [ 2 ] ) ( Q A B S A B [ 2 ] ) and Q A B | p = ( S i i + S j j ) 1

for A = B = ( i j ) , we have

k log ( det S [ 2 ] ) = at p 1 i < j n ( S i i + S j j ) 1 ( S i i ; k + S j j ; k ) = 2 1 i < j n ( 1 + λ i 2 ) ( 1 + λ j 2 ) ( 1 λ i 2 λ j 2 ) ( h ( n + i ) k i λ i 1 + λ i 2 + h ( n + j ) k j λ j 1 + λ j 2 ) .

Together with Lemma 3.3 (i) and (ii), this implies that

| k log ( det S [ 2 ] ) | n ( n 1 ) e δ ( 1 + e δ ) e δ 1 | A | .

The lemma follows from this estimate. ∎

4 Proof of main theorems

This section is devoted to prove Theorems 1.3 and 1.4. If f t is area-decreasing, then Theorem 3.2 implies that

(4.1) ( d d t Δ ) log ( det S [ 2 ] ) 2 | A | 2 + 2 R S .

We are going to express R S (3.7) in terms of the curvatures of g 1 and g 2 , and the singular values of f t . Due to (2.2), the ambient curvature R k i k ( n + i ) is

(4.2) R ( e k , e n + i ) e i , e k = R g 1 ( π 1 ( e k ) , π 1 ( e n + k ) ) ( π 1 ( e i ) ) , π 1 ( e k ) + R g 2 ( π 2 ( e k ) , π 2 ( e n + k ) ) ( π 2 ( e i ) ) , π 2 ( e k ) = λ i 1 + λ i 2 ( 1 1 + λ k 2 κ 1 ( i , k ) λ k 2 1 + λ k 2 κ 2 ( i , k ) ) ,

where κ 1 ( i , k ) is the sectional curvature of u i u k , and κ 2 ( i , k ) is the sectional curvature of v i v k . With (3.2),

(4.3) R S = 1 i < j n ( S i i + S j j ) 1 × [ 2 λ i 2 ( 1 + λ i 2 ) 2 1 k n ( 1 1 + λ k 2 κ 1 ( i , k ) λ k 2 1 + λ k 2 κ 2 ( i , k ) ) + 2 λ j 2 ( 1 + λ j 2 ) 2 1 k n ( 1 1 + λ k 2 κ 1 ( j , k ) λ k 2 1 + λ k 2 κ 2 ( j , k ) ) ] .

4.1 Proof of Theorem 1.3

Re-group the curvature term

Suppose that n m 2 , and κ 1 1 and κ 2 τ for some τ > 0 . It follows that the curvature term (4.3) in (4.1) obeys

R S i < j ( S i i + S j j ) 1 [ λ i 2 ( 1 + λ i 2 ) 2 P i + λ j 2 ( 1 + λ j 2 ) 2 P j ] ,

where

P i = 2 1 k n , k i ( 1 1 + λ k 2 λ k 2 1 + λ k 2 τ ) .

Note that P i can be rewritten as

P i = 2 1 k n , k i ( ( 1 + S k k ) ( 1 S k k ) τ ) .

Since S k k = 1 for k min { n , m } = m , one finds that

P i = ( 2 n m 1 ) ( m 1 ) τ + ( 1 + τ ) ( S i i + 1 k m S k k ) .

When m > 2 , we have S i i + 1 k m S k k > 0 , and P i ( 2 n m 1 ) ( m 1 ) τ . When m = 2 , we claim that

i < j ( S i i + S j j ) 1 [ λ i 2 ( 1 + λ i 2 ) 2 ( S i i + 1 k m S k k ) + λ j 2 ( 1 + λ j 2 ) 2 ( S j j + 1 k m S k k ) ] > 0 .

In either case, we have

(4.4) R S 1 i < j n ( S i i + S j j ) 1 [ λ i 2 ( 1 + λ i 2 ) 2 + λ j 2 ( 1 + λ j 2 ) 2 ] × ( ( 2 n m 1 ) ( m 1 ) τ ) .

The proof of the claim is based on direct computations. When m = 2 ,

1 i < j 2 ( S i i + S j j ) 1 [ λ i 2 ( 1 + λ i 2 ) 2 ( S i i + 1 k 2 S k k ) + λ j 2 ( 1 + λ j 2 ) 2 ( S j j + 1 k 2 S k k ) ] = ( S 11 + S 22 ) 1 [ λ 1 2 ( 1 + λ 1 2 ) 2 1 λ 2 2 1 + λ 2 2 + λ 2 2 ( 1 + λ 2 2 ) 2 1 λ 1 2 1 + λ 1 2 ] = ( S 11 + S 22 ) 1 ( λ 1 2 + λ 2 2 ) ( 1 λ 1 2 λ 2 2 ) ( 1 + λ 1 2 ) 2 ( 1 + λ 2 2 ) 2 0 ,
1 i 2 ( S i i + 1 ) 1 λ i 2 ( 1 + λ i 2 ) 2 ( S i i + 1 k 2 S k k ) = 1 2 ( λ 1 2 1 + λ 1 2 1 λ 2 2 1 + λ 2 2 + λ 2 2 1 + λ 2 2 1 λ 1 2 1 + λ 1 2 ) = ( λ 1 λ 2 ) 2 + 2 λ 1 λ 2 ( 1 λ 1 λ 2 ) 2 ( 1 + λ 1 2 ) ( 1 + λ 2 2 ) 0 .
Let

Φ = log det ( S [ 2 ] ) n ( n 1 ) 2 log 2 ,

which is the log det of ( 1 2 F t S ) [ 2 ] . In terms of singular values, Φ is equal to (3.11). According to (4.1) and (4.4),

(4.5) ( d d t Δ ) Φ 2 | A | 2 1 i < j n ( S i i + S j j ) 1 [ λ i 2 ( 1 + λ i 2 ) 2 + λ j 2 ( 1 + λ j 2 ) 2 ] × ( ( 2 n m 1 ) ( m 1 ) τ )

as long as f t is area-decreasing.

Long-time existence

The assumption of Theorem 1.3 implies that 𝜏 can be chosen to be less than 2 n m 1 m 1 , and Φ δ 0 for some δ 0 > 0 at t = 0 .

As long as f t is area-decreasing, it follows from (4.5) that

(4.6) ( d d t Δ ) Φ 2 | A | 2 0 .

Let

T 0 = sup { T 0 : Γ f t exists on [ 0 , T ) and f t is area-decreasing on [ 0 , T ) } .

By the maximum principle on (4.6), min Γ f t Φ is non-decreasing in 𝑡, and thus Φ δ 0 for any t [ 0 , T 0 ) ; it follows from Lemma 3.3 (i) and (ii) that being graphical and being area-decreasing is preserved for t [ 0 , T 0 ) . Moreover, the quantitative bounds, Lemma 3.3 (i) and (ii), imply that T 0 must be the maximal existence time of the mean curvature flow.

Thus (4.6) holds as long as the flow exists. With such an inequality, one can perform the blow-up argument on Huisken’s backward heat kernel, and apply White’s regularity theorem to conclude that there is no finite time singularity. This argument is the same as [19, Theorem A]; see also [9, p. 5751]. It will be omitted here.

Smooth convergence to a constant map

The first task is to study lim t Φ . Since Φ δ 0 for all 𝑡, Lemma 3.3 (i) and (ii) imply that there exists c 3 = c 3 ( n , m , τ , δ 0 ) > 0 such that the right-hand side of (4.5) is no less than c 3 i = 1 n λ i 2 . Due to Lemma 3.3 (iii), there exists c 1 = c 1 ( n , δ 0 ) such that (4.5) becomes

(4.7) ( d d t Δ ) Φ 2 | A | 2 c 3 c 1 Φ .

By the maximum principle, 0 Φ δ 0 exp ( c 3 c 1 t ) on Γ f t for any t 0 . Therefore, Φ 0 as t . With Lemma 3.3 (i), one finds that λ i 0 as t . In other words, f t converges to a constant map in the level of first order derivative.

The next step is to derive the equation for exp ( Φ ) , which can be used to bound the second fundamental form. Recall that Φ and log ( det S [ 2 ] ) differ by a constant, and hence

Φ = log ( det S [ 2 ] ) .

Let δ ( t ) = δ 0 exp ( c 3 c 1 t ) . According to Lemma 3.4,

| log ( det S [ 2 ] ) | 2 c 2 e 4 δ ( t ) ( e δ ( t ) 1 ) | A | 2

on Γ f t for any t 0 . Therefore, there exists a T 1 1 such that | log ( det S [ 2 ] ) | 2 | A | 2 whenever t T 1 . With (4.6),

( d d t Δ ) exp ( Φ ) = exp ( Φ ) [ ( d d t Δ ) Φ | log ( det S [ 2 ] ) | 2 ] exp ( Φ ) | A | 2

whenever t T 1 . By the same argument as in [9, pp. 5753–5754], this inequality can be used to prove that max Γ t | A | 2 0 as t . With λ i 0 as t , f t constant map as t . This finishes the proof of Theorem 1.3.

4.2 Proof of Theorem 1.4

The proof of Theorem 1.4 is completely parallel to that of Theorem 1.3. We only emphasize how to derive equations analogous to (4.5), (4.6) and (4.7).

Similar to (4.2), write Ric 1 ( i , j ) for the Ricci curvature of ( u i , u j ) , and Ric 2 ( i , j ) for the Ricci curvature of ( v i , v j ) , where { u i } and { v i } are the bases given by (2.2). When k > m , set κ 2 ( k , i ) = 0 and Ric 2 ( k , i ) = 0 for any 𝑖.

Rewrite (4.3) as follows:

2 λ i 2 ( 1 + λ i 2 ) 2 1 k n ( 1 1 + λ k 2 κ 1 ( i , k ) λ k 2 1 + λ k 2 κ 2 ( i , k ) ) = λ i 2 ( 1 + λ i 2 ) 2 1 k n [ ( 1 + S k k ) κ 1 ( i , k ) + ( 1 + S k k ) κ 2 ( i , k ) ] = λ i 2 ( 1 + λ i 2 ) 2 [ ( Ric 1 ( i , i ) Ric 2 ( i , i ) ) + 1 k n S k k ( κ 1 ( i , k ) + κ 2 ( i , k ) ) ] .

Hence

R S = 1 i < j n ( S i i + S j j ) 1 [ λ i 2 ( 1 + λ i 2 ) 2 ( Ric 1 ( i , i ) Ric 2 ( i , i ) )
+ λ j 2 ( 1 + λ j 2 ) 2 ( Ric 1 ( j , j ) Ric 2 ( j , j ) ) ]
(4.8) + 1 i < j n ( S i i + S j j ) 1 [ λ i 2 ( 1 + λ i 2 ) 2 1 k n S k k ( κ 1 ( i , k ) + κ 2 ( i , k ) )
+ λ j 2 ( 1 + λ j 2 ) 2 1 k n S k k ( κ 1 ( j , k ) + κ 2 ( j , k ) ) ] .
Re-group (4.8) as

1 i < j n 1 k n W i j k = 1 i < j n ( W i j i + W i j j ) + 1 i < j n 1 k n k { i , j } W i j k = 1 i < j n ( W i j i + W i j j ) + 1 i < j < k n ( W i j k + W j k i + W i k j ) ,

where W i j k is

( S i i + S j j ) 1 [ λ i 2 ( 1 + λ i 2 ) 2 S k k ( κ 1 ( i , k ) + κ 2 ( i , k ) ) + λ j 2 ( 1 + λ j 2 ) 2 S k k ( κ 1 ( j , k ) + κ 2 ( j , k ) ) ] .

We compute

W i j i + W i j j = ( S i i + S j j ) 1 [ λ j 2 ( 1 + λ j 2 ) 2 S i i ( κ 1 ( j , i ) + κ 2 ( j , i ) ) + λ i 2 ( 1 + λ i 2 ) 2 S j j ( κ 1 ( i , j ) + κ 2 ( i , j ) ) ] = λ i 2 + λ j 2 2 ( 1 + λ i 2 ) ( 1 + λ j 2 ) ( κ 1 ( i , j ) + κ 2 ( i , j ) )

and

W i j k + W j k i + W i k j = ( S i i + S j j ) 1 [ λ i 2 ( 1 + λ i 2 ) 2 S k k ( κ 1 ( i , k ) + κ 2 ( i , k ) ) + λ j 2 ( 1 + λ j 2 ) 2 S k k ( κ 1 ( j , k ) + κ 2 ( j , k ) ) ] + ( S j j + S k k ) 1 [ λ j 2 ( 1 + λ j 2 ) 2 S i i ( κ 1 ( j , i ) + κ 2 ( j , i ) ) + λ k 2 ( 1 + λ k 2 ) 2 S i i ( κ 1 ( k , i ) + κ 2 ( k , i ) ) ] + ( S i i + S k k ) 1 [ λ i 2 ( 1 + λ i 2 ) 2 S j j ( κ 1 ( i , j ) + κ 2 ( i , j ) ) + λ k 2 ( 1 + λ k 2 ) 2 S j j ( κ 1 ( k , j ) + κ 2 ( k , j ) ) ] = V i j k ( κ 1 ( i , j ) + κ 2 ( i , j ) ) + V j k i ( κ 1 ( j , k ) + κ 2 ( j , k ) ) + V i k j ( κ 1 ( i , k ) + κ 2 ( i , k ) )

where

(4.9) V i j k = ( 1 + λ k 2 ) ( λ i 2 + λ j 2 2 λ i 2 λ j 2 2 λ i 2 λ j 2 λ k 2 + λ i 4 λ j 2 λ k 2 + λ i 2 λ j 4 λ k 2 ) 2 ( 1 + λ i 2 ) ( 1 + λ j 2 ) ( 1 λ i 2 λ k 2 ) ( 1 λ j 2 λ k 2 ) = ( 1 + λ k 2 ) 2 ( 1 + λ i 2 ) ( 1 + λ j 2 ) ( 1 λ i 2 λ k 2 ) ( 1 λ j 2 λ k 2 ) × [ ( λ i λ j ) 2 ( 1 + λ i 2 λ j 2 λ k 2 ) + 2 λ i λ j ( 1 λ i λ j λ k 2 ) ( 1 λ i λ j ) ] .

In particular, V i j k 0 provided that f t is area-decreasing.

It follows that, under the assumption of Theorem 1.4,

R S = 1 i < j n [ λ i 2 ( 1 + λ j 2 ) 2 ( 1 + λ i 2 ) ( 1 λ i 2 λ j 2 ) ( Ric 1 ( i , i ) Ric 2 ( i , i ) )
+ λ j 2 ( 1 + λ i 2 ) 2 ( 1 + λ j 2 ) ( 1 λ i 2 λ j 2 ) ( Ric 1 ( j , j ) Ric 2 ( j , j ) ) ]
+ 1 i < j n λ i 2 + λ j 2 2 ( 1 + λ i 2 ) ( 1 + λ j 2 ) ( κ 1 ( i , j ) + κ 2 ( i , j ) )
+ 1 i < j < k n [ V i j k ( κ 1 ( i , j ) + κ 2 ( i , j ) ) + V j k i ( κ 1 ( j , k ) + κ 2 ( j , k ) )
+ V i k j ( κ 1 ( i , k ) + κ 2 ( i , k ) ) ]
τ 1 i < j m λ i 2 + λ j 2 2 ( 1 + λ i 2 ) ( 1 + λ j 2 )
for some τ > 0 . In the last inequality, we recall the convention that κ 2 ( i , k ) = 0 when k > m (see the end of the second paragraph in Section 4.2), and the assumption that κ 1 0 when n > m . By (4.1),

( d d t Δ ) log ( det S [ 2 ] ) 2 | A | 2 + τ 1 i < j m λ i 2 + λ j 2 ( 1 + λ i 2 ) ( 1 + λ j 2 )

provided that f t is area-decreasing. This immediate leads to (4.6). With the same argument, by using Lemma 3.3, one can obtain (4.7).

5 Implications and examples

5.1 Dilations

For a map f : ( Σ 1 , g 1 ) ( Σ 2 , g 2 ) , pick any point 𝑝, and let λ i be the singular value of d f | p . For any ρ > 0 , denote by λ i ( ρ ) the singular value of

d f | p : ( T p Σ 1 , g 1 ) ( T f ( p ) Σ 2 , ρ 2 g 2 ) .

It is easy to see that

(5.1) λ i ( ρ ) = ρ λ i .

The effect of the dilation on the sectional curvature is

(5.2) κ ρ 2 g 2 = ρ 2 κ g 2 .

It is worth noting that λ i ( ρ ) λ j ( ρ ) κ ρ 2 g 2 is independent of 𝜌.

5.2 Round spheres

Denote by g std the round metric of radius 1 on the spheres. Theorem 1.3 can be used to derive some quantitative criterion on the homotopy class of maps between spheres.

In [4, Section 2.5], Gromov proved that there exists ε ( n , m ) > 0 such that a map

f : ( S n , g std ) ( S m , g std )

with λ i λ j < ε ( n , m ) (for any i j ) everywhere must be homotopically trivial. One may also consult [6, p. 1882]. In [17, Corollary 1.2], the bound is improved to be 1. Theorem 1.3 with the dilation trick can further improve such a bound.

Corollary 5.1

Suppose that n m 2 . Any smooth map f : ( S n , g std ) ( S m , g std ) whose singular value obeys

λ i λ j < 2 n m 1 m 1

(for any i j ) everywhere must be homotopically trivial.

Proof

By the compactness of S n , there exists

0 < ρ 1 < 2 n m 1 m 1

such that λ i λ j < ρ 2 everywhere (and for any i j ). Consider f : ( S n , g std ) ( S m , ρ 2 g std ) . With the help of (5.1), 𝑓 is area-decreasing with respect to this rescaled metric. Due to (5.2), the sectional curvature of ( S m , ρ 2 g std ) is ρ 2 , which is less than 2 n m 1 m 1 . Thus Theorem 1.3 applies, and 𝑓 is homotopically trivial. ∎

To the best our knowledge, the bound given by Corollary 5.1 is sharpest so far. What follows are some known results and remarks about the bound.

  1. The bound obtained in [17] is 1. If one applies the results in [9, 16] to this setting, the bound has not been improved, and is still 1.

  2. Recently, Assimos, Savas-Halilaj and Smoczyk [1] investigated the case when m = 2 . In [1, Corollary C], the bound is improved to n 1 . Note that, when m = 2 , Corollary 5.1 gives 2 n 3 .

  3. For classical Hopf fibrations, S 3 S 2 , S 7 S 4 and S 15 S 8 , the singular values are 2 and 0, with respect to g std . Interestingly, for these dimensions, the bounds for λ i λ j given by Corollary 5.1 are all 3. The result of [1] only applies to S 3 S 2 , and the bound is 2. A natural question is whether the bound can be pushed to 4 in these dimensions.

A map from Σ 1 R n + 1 to Σ 2 R m + 1 is called a polynomial map if it is the restriction of a polynomial map from R n + 1 to R m + 1 . An interesting question [15] in algebraic topology is whether an element of π n ( S m ) can be represented by a polynomial map between unit spheres. The above corollary can be used to prove a bound on the degree of a polynomial map of non-trivial homotopy class.

Corollary 5.2

Suppose that n m 2 . A polynomial map f : ( S n , g std ) ( S m , g std ) which is not homotopically trivial must have degree no less than

2 n m 1 m 1 .

Proof

According to [14, Theorem 1.1], a degree 𝑘 polynomial map between spheres has λ i k for all 𝑖 everywhere. This corollary follows from Corollary 5.1. ∎

5.3 Complex projective spaces

Consider the complex projective space endowed with the Fubini–Study metric g FS . In terms of the homogeneous coordinate 𝐙,

g FS = | Z | 2 | d Z | 2 ( Z ̄ d Z ) ( Z d Z ̄ ) | Z | 4 .

For orthonormal vectors 𝑋 and 𝑌, the sectional curvature along X Y is

κ g FS ( X Y ) = 1 + 3 ( J X , Y ) 2 ,

where 𝐽 is the complex structure. The Ricci curvature is Ric g FS = 2 ( + 1 ) g FS , where ℓ is the complex dimension of the projective space. Note that, when = 1 , we have C P 1 S 2 , and g FS = 1 4 g std .

It is not hard to see that Theorem 1.4 applies to complex projective spaces.

Corollary 5.3

Corollary 5.3 (Corollary 1.5)

Suppose that n m 1 . Any area-decreasing map from ( C P n , g FS ) to ( C P m , g FS ) must be homotopically trivial.

As an area-decreasing map f : C P n C P m with n m 1 induces a trivial map on the cohomology groups f : H 2 ( C P m ) H 2 ( C P n ) , it is natural to raise the following question.

Question

Suppose a map f : C P n C P m with n > m 1 induces a trivial map on the cohomology groups f : H 2 ( C P m ) H 2 ( C P n ) . Is it necessarily homotopically trivial?

We remark that, in [12], it was proved that a pinched symplectomorphism of ( C P n , g FS ) is homotopic to a biholomorphic isometry through the mean curvature flow. See also [22].

The following corollary concerns π 2 n + 1 ( C P n ) , which is isomorphic to

π 2 n + 1 ( S 2 n + 1 ) Z

by the long exact sequence induced by the Hopf fibration S 1 S 2 n + 1 C P n .

Corollary 5.4

Corollary 5.4 (Corollary 1.6)

Any smooth map f : ( S 2 n + 1 , g std ) ( C P n , g FS ) whose singular value satisfies

λ i λ j < 2 n 2 n + 1

(for any i j ) everywhere must be homotopically trivial.

Proof

Let

ρ = 2 n + 1 2 n .

Consider f : ( S 2 n + 1 , g std ) ( C P n , ρ 2 g FS ) . According to (5.1), 𝑓 is area-decreasing with respect to this rescaled metric. By (5.2), ( C P n , ρ 2 g FS ) is an Einstein metric with Einstein constant ρ 2 ( 2 n + 1 ) = 2 n . Since the Einstein constant of ( S 2 n + 1 , g std ) is 2 n , Theorem 1.4 applies, and 𝑓 is homotopically trivial. ∎

Remark 5.5

These discussions apply to the quaternionic projective spaces as well. The quaternionic projective space H P n also carries a naturally defined Fubini–Study metric g FS . For orthonormal vectors 𝑋 and 𝑌, the sectional curvature along X Y is

sec g FS ( X Y ) = 1 + 3 μ = 1 3 ( J μ X , Y ) 2 .

The Ricci curvature is Ric g FS = 4 ( n + 2 ) g FS . Note that, when = 1 , we have H P 1 S 4 , and g FS = 1 4 g std . It follows that the analogous statement to Corollary 1.6 holds true for ( H P n , g FS ) . For Corollary 5.3, the bound becomes

4 n + 2 4 ( n + 2 ) .

Again, by considering the Hopf fibration S 3 S 4 n + 3 H P n , the bound is sharp as n .

6 Comments on other approaches and earlier results

Other than the mean curvature flow and methods that are of topological nature or rely on the ℎ-principle [4, 6], there is another approach of Llarull [10] to study area-decreasing maps by Dirac operators.

Theorem 6.1

Suppose 𝑀 is an 𝑛-dimensional compact spin Riemannian manifold with scalar curvature no less than n ( n 1 ) . Any area-non-increasing map f : M S n with non-zero degree is an isometry.

In view of Corollary 1.5, it is natural to speculate that the mean curvature flow will give a classification of the homotopy class of area-non-increasing maps from ( C P n , g FS ) to ( C P n , g FS ) .

In the rest of this section, we explain how Theorem 3.2 also implies [16, Theorem A]. In particular, we show that the term R S in Theorem 3.2 is non-negative under the assumptions of [16, Theorem A]: n 2 , and there exists a constant σ > 0 such that

(6.1) κ 1 > σ and Ric 1 ( n 1 ) σ ( n 1 ) κ 2 .

As before, start with (4.1), and rewrite (4.3). Since κ 2 σ ,

R S 1 i < j n ( S i i + S j j ) 1 [ 2 λ i 2 ( 1 + λ i 2 ) 2 1 k n k i ( 1 1 + λ k 2 κ 1 ( i , k ) λ k 2 1 + λ k 2 σ ) + 2 λ j 2 ( 1 + λ j 2 ) 2 1 k n k j ( 1 1 + λ k 2 κ 1 ( j , k ) λ k 2 1 + λ k 2 σ ) ] .

With 2 1 + λ k 2 = 1 + S k k and 2 λ k 1 + λ k 2 = 1 S k k , R S is no less than

1 i < j n ( S i i + S j j ) 1 [ λ i 2 ( 1 + λ i 2 ) 2 1 k n k i ( ( κ 1 ( i , k ) σ ) + S k k ( κ 1 ( i , k ) + σ ) ) + λ j 2 ( 1 + λ j 2 ) 2 1 k n k j ( ( κ 1 ( j , k ) σ ) + S k k ( κ 1 ( j , k ) + σ ) ) ] = 1 i < j n ( S i i + S j j ) 1 [ λ i 2 ( 1 + λ i 2 ) 2 ( Ric 1 ( i , i ) ( n 1 ) σ ) + λ j 2 ( 1 + λ j 2 ) 2 ( Ric 1 ( j , j ) ( n 1 ) σ ) ] + 1 i < j n ( S i i + S j j ) 1 [ λ i 2 ( 1 + λ i 2 ) 2 1 k n k i S k k ( κ 1 ( i , k ) + σ ) + λ j 2 ( 1 + λ j 2 ) 2 1 k n k j S k k ( κ 1 ( j , k ) + σ ) ] .

For the latest summation, one can apply the exact same argument as that for (4.8). It follows that

R S 1 i < j n [ λ i 2 ( 1 + λ j 2 ) 2 ( 1 + λ i 2 ) ( 1 λ i 2 λ j 2 ) ( Ric 1 ( i , i ) ( n 1 ) σ ) + λ j 2 ( 1 + λ i 2 ) 2 ( 1 + λ j 2 ) ( 1 λ i 2 λ j 2 ) ( Ric 1 ( j , j ) ( n 1 ) σ ) ] + 1 i < j n λ i 2 + λ j 2 2 ( 1 + λ i 2 ) ( 1 + λ j 2 ) ( κ 1 ( i , j ) + σ ) + 1 i < j < k n [ V i j k ( κ 1 ( i , j ) + σ ) + V j k i ( κ 1 ( j , k ) + σ ) + V i k j ( κ 1 ( i , k ) + σ ) ] ,

where V i j k is given by (4.9). Under assumption (6.1), the rest of the argument is the same as that for the proof of Theorem 1.4.

Award Identifier / Grant number: 112-2636-M-002-003

Award Identifier / Grant number: 112-2628-M-002-004-MY4

Award Identifier / Grant number: 109-2115-M-002-006

Award Identifier / Grant number: 112-2115-M-002-015-MY3

Award Identifier / Grant number: DMS-1810856

Award Identifier / Grant number: DMS-2104212

Funding statement: C.-J. Tsai is supported in part by the National Science and Technology Council grants 112-2636-M-002-003 and 112-2628-M-002-004-MY4. M.-P. Tsui is supported in part by the National Science and Technology Council grants 109-2115-M-002-006 and 112-2115-M-002-015-MY3. M.-T. Wang is supported in part by the National Science Foundation under Grants DMS-1810856 and DMS-2104212. Part of this work was carried out when M.-T. Wang was visiting the National Center of Theoretical Sciences.

Acknowledgements

The authors thank Man-Chun Lee and Luen-Fai Tam for suggestions on the earlier draft of this paper.

References

[1] R. Assimos, A. Savas-Halilaj and K. Smoczyk, Graphical mean curvature flow with bounded bi-Ricci curvature, Calc. Var. Partial Differential Equations 62 (2023), no. 1, Paper No. 12. 10.1007/s00526-022-02369-3Suche in Google Scholar

[2] D. DeTurck, H. Gluck and P. Storm, Lipschitz minimality of Hopf fibrations and Hopf vector fields, Algebr. Geom. Topol. 13 (2013), no. 3, 1369–1412. 10.2140/agt.2013.13.1369Suche in Google Scholar

[3] M. Gromov, Homotopical effects of dilatation, J. Differential Geometry 13 (1978), no. 3, 303–310. 10.4310/jdg/1214434601Suche in Google Scholar

[4] M. Gromov, Carnot–Carathéodory spaces seen from within, Sub-Riemannian geometry, Progr. Math. 144, Birkhäuser, Basel (1996), 79–323. 10.1007/978-3-0348-9210-0_2Suche in Google Scholar

[5] L. Guth, Homotopically non-trivial maps with small k-dilation, preprint (2007), https://arxiv.org/abs/0709.1241. Suche in Google Scholar

[6] L. Guth, Contraction of areas vs. topology of mappings, Geom. Funct. Anal. 23 (2013), no. 6, 1804–1902. 10.1007/s00039-013-0246-3Suche in Google Scholar

[7] A. C. L. Hsu, A characterization of the Hopf map by stretch, Math. Z. 129 (1972), 195–206. 10.1007/BF01187348Suche in Google Scholar

[8] H. B. Lawson, Jr., Stretching phenomena in mappings of spheres, Proc. Amer. Math. Soc. 19 (1968), 433–435. 10.1090/S0002-9939-1968-0230324-4Suche in Google Scholar

[9] K.-W. Lee and Y.-I. Lee, Mean curvature flow of the graphs of maps between compact manifolds, Trans. Amer. Math. Soc. 363 (2011), no. 11, 5745–5759. 10.1090/S0002-9947-2011-05204-9Suche in Google Scholar

[10] M. Llarull, Sharp estimates and the Dirac operator, Math. Ann. 310 (1998), no. 1, 55–71. 10.1007/s002080050136Suche in Google Scholar

[11] C. A. McGibbon, Self-maps of projective spaces, Trans. Amer. Math. Soc. 271 (1982), no. 1, 325–346. 10.1090/S0002-9947-1982-0648096-XSuche in Google Scholar

[12] I. Medoš and M.-T. Wang, Deforming symplectomorphisms of complex projective spaces by the mean curvature flow, J. Differential Geom. 87 (2011), no. 2, 309–341. 10.4310/jdg/1304514975Suche in Google Scholar

[13] J. M. Møller, On spaces of maps between complex projective spaces, Proc. Amer. Math. Soc. 91 (1984), no. 3, 471–476. 10.1090/S0002-9939-1984-0744651-4Suche in Google Scholar

[14] C. Peng and Z. Tang, Dilatation of maps between spheres, Pacific J. Math. 204 (2002), no. 1, 209–222. 10.2140/pjm.2002.204.209Suche in Google Scholar

[15] J. Peng and Z. Tang, Algebraic maps from spheres to spheres, Sci. China Ser. A 42 (1999), no. 11, 1147–1154. 10.1007/BF02875982Suche in Google Scholar

[16] A. Savas-Halilaj and K. Smoczyk, Homotopy of area decreasing maps by mean curvature flow, Adv. Math. 255 (2014), 455–473. 10.1016/j.aim.2014.01.014Suche in Google Scholar

[17] M.-P. Tsui and M.-T. Wang, Mean curvature flows and isotopy of maps between spheres, Comm. Pure Appl. Math. 57 (2004), no. 8, 1110–1126. 10.1002/cpa.20022Suche in Google Scholar

[18] M.-T. Wang, Mean curvature flow of surfaces in Einstein four-manifolds, J. Differential Geom. 57 (2001), no. 2, 301–338. 10.4310/jdg/1090348113Suche in Google Scholar

[19] M.-T. Wang, Long-time existence and convergence of graphic mean curvature flow in arbitrary codimension, Invent. Math. 148 (2002), no. 3, 525–543. 10.1007/s002220100201Suche in Google Scholar

[20] M.-T. Wang, Gauss maps of the mean curvature flow, Math. Res. Lett. 10 (2003), no. 2–3, 287–299. 10.4310/MRL.2003.v10.n3.a2Suche in Google Scholar

[21] M.-T. Wang, Subsets of Grassmannians preserved by mean curvature flows, Comm. Anal. Geom. 13 (2005), no. 5, 981–998. 10.4310/CAG.2005.v13.n5.a7Suche in Google Scholar

[22] M.-T. Wang, Mean curvature flows and isotopy problems, Surveys in differential geometry. Geometry and topology, Surv. Differ. Geom. 18, International Press, Somerville (2013), 227–235. 10.4310/SDG.2013.v18.n1.a6Suche in Google Scholar

Received: 2024-02-20
Revised: 2025-09-05
Published Online: 2025-09-30

© 2025 the author(s), published by De Gruyter

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

Heruntergeladen am 20.10.2025 von https://www.degruyterbrill.com/document/doi/10.1515/crelle-2025-0066/html
Button zum nach oben scrollen