Skip to content
BY 4.0 license Open Access Published by De Gruyter October 2, 2020

Deforming a Convex Hypersurface by Anisotropic Curvature Flows

  • HongJie Ju , BoYa Li and YanNan Liu EMAIL logo

Abstract

In this paper, we consider a fully nonlinear curvature flow of a convex hypersurface in the Euclidean 𝑛-space. This flow involves 𝑘-th elementary symmetric function for principal curvature radii and a function of support function. Under some appropriate assumptions, we prove the long-time existence and convergence of this flow. As an application, we give the existence of smooth solutions to the Orlicz–Christoffel–Minkowski problem.

MSC 2010: 35J96; 35J75; 53A15; 53A07

1 Introduction

Let M0 be a smooth, closed, strictly convex hypersurface in the Euclidean space Rn, which encloses the origin and is given by a smooth embedding X0:Sn-1Rn. Consider a family of closed hypersurfaces {Mt} with Mt=X(Sn-1,t), where X:Sn-1×[0,T)Rn is a smooth map satisfying the initial value problem(1.1)

Xt(x,t)=1f(ν)σk(x,t)φ(X,ν)X,νη(t)ν-X,
X(x,0)=X0(x).
Here 𝑓 is a given positive and smooth function on the unit sphere Sn-1, 𝜈 is the unit outer normal vector of Mt at the point X(x,t). Further, , is the standard inner product in Rn, 𝜑 is a positive smooth function defined in (0,+), 𝜂 is a scalar function to be specified later, and 𝑇 is the maximal time for which the solution exists. We use {eij}, 1i,jn-1, and for the standard metric and the Levi–Civita connection of Sn-1 respectively. Principal radii of curvature are the eigenvalues of the matrix bij:=ijh+eijh with respect to {eij}. Moreover, σk(x,t) is the 𝑘-th elementary symmetric function for principal curvature radii of Mt at X(x,t) and 𝑘 is an integer with 1kn-1. In this paper, σk is normalized so that σk(1,,1)=1.

Geometric flows with speed of symmetric polynomial of the principal curvature radii of the hypersurface have been extensively studied; see e.g. [34, 8, 11, 36].

On the other hand, anisotropic curvature flows provide alternative methods to prove the existences of elliptic PDEs arising from convex geometry; see e.g. [3, 4, 5, 6, 21, 26, 27, 28, 33].

A positive homothetic self-similar solution of (1.1), if it exists, is a solution to the fully nonlinear equation

(1.2)cφ(h)σk(x)=f(x)onSn-1

for some positive constant 𝑐. Here ℎ is the support function defined on Sn-1. We are concerned with the existence of smooth solutions for equation (1.2).

When k=n-1, equation (1.2) is just the smooth case of the Orlicz–Minkowski problem. The Orlicz–Minkowski problem is a basic problem in the Orlicz–Brunn–Minkowski theory in convex geometry. It is a generalization of the classical Minkowski problem which asks what are the necessary and sufficient conditions for a Borel measure on the unit sphere Sn-1 to be a multiple of the Orlicz surface area measure of a convex body in Rn. In [14], Haberl, Lutwak, Yang and Zhang studied the even case of the Orlicz–Minkowski problem. After that, the Orlicz–Minkowski problem attracted great attention from many scholars; see for example [9, 10, 19, 22, 37, 35, 39].

When φ(s)=s1-p, k=n-1, equation (1.2) reduces to the Lp-Minkowski problem, which has been extensively studied; see e.g. [2, 7, 16, 18, 20, 23, 24, 29, 30, 31, 38].

When 1k<n-1, equation (1.2) is the so-called Orlicz–Christoffel–Minkowski problem. For φ(s)=s1-p, 1k<n-1, (1.2) is known as the Lp-Christoffel–Minkowski problem and is the classical Christoffel–Minkowski problem for p=1. Under a sufficient condition on the prescribed function, existence of the solution for the classical Christoffel–Minkowski problem was given in [12].

The Lp-Christoffel–Minkowski problem is related to the problem of prescribing the 𝑘-th 𝑝-area measures. Hu, Ma and Shen [17] proved the existence of convex solutions to the Lp-Christoffel–Minkowski problem for pk+1 under appropriate conditions. Using the methods of geometric flows, Ivaki [21] and then Sheng and Yi [33] also gave the existence of smooth convex solutions to the Lp-Christoffel–Minkowski problem for pk+1. In case 1<p<k+1, Guan and Xia [13] established the existence of convex body with the prescribed 𝑘-th even 𝑝-area measures.

In this paper, we study the long-time existence and convergence of flow (1.1) for strictly convex hypersurfaces and the existence of smooth solutions to the Orlicz–Christoffel–Minkowski problem (1.2).

The scalar function η(t) in (1.1) is usually used to keep Mt normalized in a certain sense; see for example [4, 21, 33]. In this paper, 𝜂 is given by

η(t)=Sn-1hf(x)φ(h)dxSn-1hσkdx,

where h(,t) is the support function of the convex hypersurface Mt. It will be proved in Section 2 that Sn-1hσkdx is non-decreasing along the flow under this choice of 𝜂.

To obtain the long-time existence of flow (1.1), we need some constraints on 𝜑.

  1. φ(s) is a positive and continuous function defined in (0,+) such that φ>αs-k-ε for some positive constants 𝜀 and 𝛼 for 𝑠 near 0 and ϕ(s)=0s1φ(τ)dτ is unbounded as s+. Here 𝑘 is the order of σk.

The main results of this paper are stated as follows.

Theorem 1

Assume M0 is a smooth, closed and strictly convex hypersurface in Rn. Suppose 𝑘 is an integer with 1k<n-1 and φC(0,+) satisfying (A). Moreover, for any s>0,

s(ss(logφ(s)))0and-ass(logφ(s))-1,

where 𝑎 is a positive constant. Suppose 𝑓 is a smooth function on Sn-1 such that

(k+1)f-1k+aeij+(k+a)ij(f-1k+a)

is positive definite. Then flow (1.1) has a unique smooth solution Mt for all times t>0. Moreover, when t, a subsequence of Mt converges in C to a smooth, closed, strictly convex hypersurface, whose support function is a smooth solution to equation (1.2) for some positive constant 𝑐.

When f1, we have the following result.

Theorem 2

Assume M0 is a smooth, closed and strictly convex hypersurface in Rn. If f1, φC(0,+) satisfying (A), and 𝑘 is an integer with 1k<n-1. Moreover, for any s>0,

s(ss(logφ(s)))0andss(logφ(s))-1.

Then flow (1.1) has a unique smooth solution Mt for all times t>0. Moreover, when t, a subsequence of Mt converges in C to a smooth, closed, strictly convex hypersurface, whose support function is a smooth solution to equation (1.2) for some positive constant 𝑐.

As an application, we have the following corollary.

Corollary 1

Under the assumptions of Theorem 1 or Theorem 2, there exists a smooth solution to equation (1.2) for some positive constant 𝑐.

From the proof of Lemma 7 in Section 3, we will see if φ(s)sφ(s)=a0 for some negative constant a0, then the convexity condition on 𝑓 reduces to f-1k-a0eij+jj(f-1k-a0) being positive definite. Hence, when φ(s)=s1-p for pk+1 with the above condition on 𝑓, our conclusion recovers the existence results to the Lp-Christoffel–Minkowski problem which have been obtained in [17, 21, 33].

This paper is organized as follows. In Section 2, we give some basic knowledge about flow (1.1) and evolution equations of some geometric quantities. In Section 3, the long-time existence of flow (1.1) will be obtained. First, under assumption (A), uniform positive upper and lower bounds for support functions of {Mt} are derived. Based on the bounds of support functions, we obtain the uniform bounds of principal curvatures by constructing proper auxiliary functions. The long-time existence of flow (1.1) then follows by standard arguments. In Section 4, by considering a related geometric functional, we prove that a subsequence of {Mt} converges to a smooth solution to equation (1.2), completing the proofs of Theorem 1 and Theorem 2.

2 Preliminaries

Let Rn be the 𝑛-dimensional Euclidean space, and let Sn-1 be the unit sphere in Rn. Assume 𝑀 is a smooth closed strictly convex hypersurface in Rn. Without loss of generality, we may assume that 𝑀 encloses the origin. The support function ℎ of 𝑀 is defined as h(x):=maxyMy,x for all xSn-1, where , is the standard inner product in Rn.

Denote the Gauss map of 𝑀 by νM. Then 𝑀 can be parametrized by the inverse Gauss map X:Sn-1M with X(x)=νM-1(x). The support function ℎ of 𝑀 can be computed by

(2.1)h(x)=x,X(x),xSn-1.

Note that 𝑥 is just the unit outer normal of 𝑀 at X(x). Differentiating (2.1), we have

ih=ix,X(x)+x,iX(x).

Since iX(x) is tangent to 𝑀 at X(x), we have ih=ix,X(x). It follows that

(2.2)X(x)=h+hx.

By differentiating (2.1) twice, the second fundamental form Aij of 𝑀 can be computed in terms of the support function (see for example [34]),

(2.3)Aij=ijh+heij,

where ij=ij denotes the second-order covariant derivative with respect to eij. The induced metric matrix gij of 𝑀 can be derived by Weingarten’s formula,

(2.4)eij=ix,jx=AimAljgml.

The principal radii of curvature are the eigenvalues of the matrix bij=Aikgjk. When considering a smooth local orthonormal frame on Sn-1, by virtue of (2.3) and (2.4), we have

(2.5)bij=Aij=ijh+hδij.

We will use bij to denote the inverse matrix of bij.

From the evolution equation of X(x,t) in flow (1.1), we derive the evolution equation of the corresponding support function h(x,t),

(2.6)h(x,t)t=1f(x)σk(x,t)φ(h)h(x,t)η(t)-h(x,t).

The radial function 𝜌 of 𝑀 is given by ρ(u):=max{λ>0:λuM} for all uSn-1. Note that ρ(u)uM. From (2.2), 𝑢 and 𝑥 are related by ρ(u)u=h(x)+h(x)x and ρ2=|h|2+h2.

In the rest of the paper, we take a local orthonormal frame {e1,,en-1} on Sn-1 such that the standard metric on Sn-1 is {δij}. Double indices always mean to sum from 1 to n-1. We denote partial derivatives σkbij and 2σkbpqbmn by σkij and σkpq,mn respectively. For convenience, we also write N=1f(x)φ(h)h, F=Nσkη(t).

Now, we can prove that the mixed volume Sn-1h(x,t)σk(x,t)dx is non-decreasing along flow (1.1).

Lemma 1

S n - 1 h ( x , t ) σ k ( x , t ) d x is non-decreasing along flow (1.1).

Proof

According to the evolution equation of ℎ in (2.6), we get

tσk=σkijt(ijh+δijh)=σkijij(th)+σkijδijth=σkijijF-σkijijh+σkijδijF-σkijδijh=σkijijF+σkijδijF-kσk;

the last equality holds because σk is homogeneous of degree 𝑘 and σkijbij=kσk. Hence

tSn-1hσkdx=Sn-1(tσk)hdx+Sn-1σkthdx=Sn-1(hσkijijF+hσkijδijF-khσk)dx+Sn-1Fσkdx-Sn-1hσkdx=(k+1)Sn-1Fσkdx-(k+1)Sn-1hσkdx+Sn-1(hσkijijF-Fσkijijh)dx=(k+1)Sn-1Fσkdx-(k+1)Sn-1hσkdx,

where, in the last equality, we use the fact ii(σkij)=0.

By Hölder’s inequality, we have

1k+1tSn-1hσkdx=Sn-11f(x)σk2φ(h)hηdx-Sn-1hσkdx=1Sn-1hσkdx[Sn-11f(x)σk2φ(h)hdxSn-1hφ(h)f(x)dx-(Sn-1hσkdx)2]0,

and the equality holds if and only if cφ(h)σk(x)=f(x) for some positive constant 𝑐. ∎

By flow equation (1.1), we can derive evolution equations of some geometric quantities.

Lemma 2

The following evolution equations hold along flow (1.1):

tbij-Nη(t)σkpqpqbij=(k+1)Nη(t)σkδij-Nη(t)σkpqδpqbij+Nη(t)(σkipbjp-σkjpbip)
+Nη(t)σkpq,mnjbpqibmn+η(t)(σkijN+jσkiN+iσkjN)-bij,
tbij-Nη(t)σkpqpqbij=-(k+1)Nη(t)σkbipbjp+Nη(t)σkpqδpqbij-Nη(t)bipbjq(σkrpbrq-σkrqbrp)
-Nη(t)bilbjs(σkpq,mn+2σkpmbnq)lbpqsbmn
-η(t)bipbjq(σkijN+jσkiN+iσkjN)+bij,
t(ρ22)-Nη(t)σkijij(ρ22)=(k+1)hNη(t)σk-ρ2+η(t)σkihiN-Nη(t)σkijbmibmj.

Proof

From (1.1),

tijh=ij(th)=η(t)(σkijN+jσkiN+iσkjN)+Nη(t)ijσk-hij,

where ijσk=σkpq,mnjbpqibmn+σkpqijbpq. By the Gauss equation,

ijbpq=pqbij+δijpqh-δpqijh+δiqpjh-δpjiqh.

Hence

thij=Nη(t)σkpqpqbij+kNη(t)σkδij-Nη(t)σkpqδpqbij+Nη(t)(σkipbjp-σkjpbip)+Nη(t)σkpq,mnjbpqibmn+η(t)(σkijN+jσkiN+iσkjN)-hij.

This together with (2.5) gives the evolution equation of bij. The evolution equation of bij then follows from tbij=-bimbljtbml. For more details of computations about the evolution equations of bij and bij, one can refer to [8, 34].

Recalling that ρ2=h2+|h|2, we have

t(ρ22)-Nη(t)σkijij(ρ22)=hth+ihith-Nη(t)σkij(hijh+ihjh+mhjmih+mihmjh)=hth+ihi(Nη(t)σk-h)-Nη(t)σkij[ihjh+mhj(bmi-hδmi)]-Nη(t)σkijh(bij-hδij)-Nη(t)σkij(bmi-hδmi)(bmj-hδmj)=(k+1)hNη(t)σk-ρ2+η(t)σkihiN-Nη(t)σkijbmibmj.

3 The Long-Time Existence of the Flow

In this section, we will give a priori estimates about support functions and curvatures to obtain the long-time existence of flow (1.1) under assumptions of Theorem 1 and Theorem 2.

In the rest of this paper, we assume that M0 is a smooth, closed, strictly convex hypersurface in Rn and h:Sn-1×[0,T)R is a smooth solution to the evolution equation (2.6) with the initial h(,0) the support function of M0. Here 𝑇 is the maximal time for which the solution exists. Let Mt be the convex hypersurface determined by h(,t), and let ρ(,t) be the corresponding radial function.

We first give the uniform positive upper and lower bounds of h(,t) and ρ(u,t) for t[0,T).

Lemma 3

Let ℎ be a smooth solution of (2.6) on Sn-1×[0,T), 𝑓 a positive, smooth function on Sn-1 and φC(0,+) a decreasing function satisfying (A). Then

1Ch(x,t)C,1Cρ(u,t)C,

where 𝐶 is a positive constant independent of 𝑡.

Proof

Let J(t)=Sn-1ϕ(h(x,t))f(x)dx. We claim that J(t) is unchanged along flow (1.1). It is because

J(t)=Sn-1ϕ(h)thf(x)dx=Sn-1f(x)φ(h)thdx=Sn-1f(x)φ(h)(1f(x)σk(x)φ(h)hη(t)-h)dx=Sn-1σk(x)hη(t)dx-Sn-1hφ(h)f(x)dx=0.

For each t[0,T), suppose that the maximum of radial function ρ(,t) is attained at some utSn-1. Let

Rt=maxuSn-1ρ(u,t)=ρ(ut,t)

for some utSn-1. By the definition of support function, we have h(x,t)Rtx,ut for all xSn-1. Denote the hemisphere containing ut by Sut+={xSn-1:x,ut>0}. Since ϕ(h)=1φ(h)>0 implies that ϕ(h) is strictly increasing about ℎ, we have

J(0)=J(t)Sut+ϕ(h(x,t))f(x)dxSut+ϕ(Rtx,ut)f(x)dxfminSut+ϕ(Rtx,ut)dx=fminS+ϕ(Rtx1)dx,

where S+={xSn-1:x1>0}.

Denote S1={xSn-1:x112}. Then

J(0)fminS1ϕ(Rt2)dx=fminϕ(Rt2)|S1|,

which implies that ϕ(Rt2) is uniformly bounded from above. By assumption (A), ϕ(s) is strictly increasing and tends to + as s+. Thus Rt is uniformly bounded from above.

Now we can prove that η(t) has a positive lower bound. Since mixed volumes are monotonic increasing (see [32, page 282]), we have, for each t[0,T),

hmink+1(t)Sn-1hσkdxωn-1hmaxk+1(t),

here hmin(t)=minxSn-1h(x,t), and hmax(t)=maxxSn-1h(x,t).

This together with Lemma 1 and the upper bound of ℎ implies that there exist positive constants c1 and c2 such that

Sn-1hσkdxc1andhmax(t)c2.

Recalling the definition of η(t) and noticing that 1φ(s) is an increasing function, we have

η(t)=Sn-1hf(x)φ(h)dxSn-1hσkdx1c1Sut+hf(x)φ(h)dx1c1Sut+Rtx,utfmin1φ(Rtx,ut)dx=1c1fminS+Rtx11φ(Rtx1)dx1c1fmin|S1|12Rt1φ(12Rt)c3,

where c3 is a positive constant independent of 𝑡.

Suppose the minimum of h(x,t) is attained at a point (xt,t). At (xt,t), ijh is non-negative. It follows that σk(xt,t)hmink(t). Then, in the sense of the lim inf of a difference quotient (see [15]), we have

hmin(t)t1fmaxhmin(t)[η(t)hmink(t)φ(hmin(t))-f(x)]1fmaxhmin(t)[c3hmink(t)φ(hmin(t))-fmax].

If φ(s)>αs-k-ε for some positive constant 𝜀 for 𝑠 near 0, then

hmin(t)t1fmaxhmin(t)(hmin-ε(t)αc3-fmax).

The right hand of the above inequality is positive for hmin(t) small enough, and the lower bound of hmin(t) follows from the maximum principle in [15]. ∎

By the equality ρ2=h2+|h|2, we can obtain the gradient estimate of the support function from Lemma 3.

Corollary 2

Under the assumptions of Lemma 3, we have

|h(x,t)|Cfor all(x,t)Sn-1×[0,T),

where 𝐶 is a positive constant depending only on constants in Lemma 3.

The uniform bounds of η(t) can be derived from Lemmas 1 and 3.

Lemma 4

Under the assumptions of Lemma 3, η(t) is uniformly bounded above and below from zero.

Proof

In term of the proof of Lemma 3, η(t) has uniform positive lower bound. From Lemma 1, we know that Sn-1hσkdx is monotonic decreasing about 𝑡, which gives a positive lower bound on Sn-1hσkdx. This together with the uniform bounds of h(x,t) implies that η(t) is bounded from above. ∎

To obtain the long-time existence of flow (1.1), we need to establish the uniform bounds on principal curvatures. By Lemma 3, for any t[0,T), h(,t) always ranges within a bounded interval I=[1C,C], where 𝐶 is the constant in Lemma 3. First, we give the estimates of σk.

Lemma 5

Under the assumptions of Lemma 3, σk(x,t)C for all (x,t)Sn-1×[0,T), where 𝐶 is a positive constant independent of 𝑡.

Proof

Consider the auxiliary function Q=logM-Aρ22, where M=Nσk=1f(x)φ(h)hσk and 𝐴 is a positive constant to be determined later. The evolution equation of 𝑀 is given by

tM=Ntσk+σktN=N(σkijijF+σkijδijF-kσk)+Mh(1+φhφ)(F-h)=NσkijijF+NσkijδijF-kM+M2hη(t)(1+φhφ)-M(1+φhφ)=Nη(t)σkijijM+MNη(t)σkijδij-M(k+1+φhφ)+M2hη(t)(1+φhφ).

It is easy to compute

iQ=iMM-Ai(ρ22),ijQ=ijMM-1M2iMjM-Aij(ρ22).

Due to the evolution equation of ρ22 in Lemma 2, the evolution equation of 𝑄 is

tQ-Nη(t)σkijijQ=1M2Nη(t)σkijiMjM+Nη(t)σkijδij-(k+1+φhφ)+Mhη(t)(1+φhφ)-(k+1)AhNη(t)σk+Aρ2-Aη(t)σkihiN+ANη(t)σkijbmibmj.

For fixed 𝑡, at a point where 𝑄 attains its spatial minimum, we have

tQAρ2-(k+1+φhφ)+Mhη(t)(1+φhφ)-(k+1)AhNη(t)σk-Aη(t)σkihiN=12Aρ2-(k+1+φhφ)+1heQ+Aρ22η(t)(1+φhφ)+ANη(t)σk(ρ22eQ+Aρ22η(t)-h(k+1)-1NihiN).

Now we choose A>2ρ2(k+1). Notice that 𝜑 is a monotonic decreasing, positive function and we have obtained uniform bounds of ℎ, 𝜌, |h| and η(t). If 𝑄 is negatively large enough, the right-hand side is positive, and the lower bound of 𝑄 follows. ∎

Lemma 6

Under the assumptions of Lemma 3, σkC for all (x,t)Sn-1×[0,T), where 𝐶 is a positive constant independent of 𝑡.

Proof

By Lemma 3, there exists a positive constant 𝐵 such that B<ρ2<1B for all t>0. Define

P(x,t)=φσkf(1-Bρ22)=Mh11-Bρ22.

By the evolution equation of 𝑀 in Lemma 5, we have

tMh-Nη(t)σkijijMh=-Mh(k+φhφ)+M2h2η(t)(k+φhφ)+2Nhη(t)σkijihjMh.

Hence

tP-Nη(t)σkijijP=11-Bρ22[-Mh(k+φhφ)+M2h2η(t)(k+φhφ)+2Nhη(t)σkijihjMh]+MBh(1-Bρ22)2[(k+1)Nhη(t)σk-ρ2+η(t)σkihiN-Nη(t)σkijbmibmj]-2B1-Bρ22Nη(t)σkijiρ22jP.

At a point where P(,t) attains its maximum, we have

jMh=-MhBjρ221-Bρ22=-MhBbjmhm1-Bρ22.

Due to the inverse concavity of (σk)1k, we have from [1, Lemma 5], ((σk)1k)ijbimbjm(σk)2k, which means σkijbimbjmk(σk)1+1k. Then, at the point where P(,t) attains its maximum, we have

tP11-Bρ22[-Mh(k+φhφ)+M2h2η(t)(k+φhφ)]+MhB(1-Bρ22)2[(k+1)Nhη(t)σk-ρ2+η(t)σkihiN-kNη(t)(σk)1+1k].

From Lemmas 3 and 4, we have tPc1P+c2P2-c3P2+1k. By the maximum principle, we see that P(x,t) is uniformly bounded from above. The upper bound of σk follows from the uniform bounds on ℎ and 𝜌. ∎

Now we can derive the upper bounds of principal curvatures κi(x,t) of Mt for i=1,,n-1.

Lemma 7

Under the assumptions of Theorem 1, we have κiC for all (x,t)Sn-1×[0,T), where 𝐶 is a positive constant independent of 𝑡.

Proof

By rotation, we assume that the maximal eigenvalue of bij at 𝑡 is attained at point xt in the direction of the unit vector e1TxtSn-1. We also choose the orthonormal vector field such that bij is diagonal. By the evolution equation of bij in Lemma 2, we get

tb11h-Nη(t)σkpqpqb11h=2hNη(t)σkpqpb11hqh+Nh2η(t)b11σkpqpqh-(k+1)Nhη(t)σk(b11)2+Nhη(t)σkpqδpqb11-Nhη(t)(b11)2(σkpq,mn+2σkpmbnq)1bpq1bmn-η(t)h(b11)2(11Nσk+21σk1N)-b11h2Nη(t)σk+2b11h=2hNη(t)σkpqpb11hqh-(k+1)Nhη(t)σk(b11)2-Nhη(t)(b11)2(σkpq,mn+2σkpmbnq)1bpq1bmn-η(t)h(b11)2(11Nσk+21σk1N)+(k-1)b11h2Nη(t)σk+2b11h.

According to inverse concavity of (σk)1k, we obtain, by [34] or [1],

(σkpq,mn+2σkpmbnq)1bpq1bmnk+1k(1σk)2σk.

On the other hand, by the Schwartz inequality, the following inequality holds:

2|1σk1N|k+1kN(1σk)2σk+kk+1σk(1N)2N.

Hence we have, at (xt,t),

tb11h-(b11)2hσkη(t)[11N-kk+1(1N)2N+(k+1)N+(1-k)Nb11h]+2b11h.

Let 𝜏 be the arc-length of the great circle passing through xt with the unit tangent vector e1. Notice that

11N-kk+1(1N)2N+(k+1)N=(k+1)Nkk+1(N1k+1+(N1k+1)ττ).

Since

Nτ=(f-1)τφh+f-1φhτ(1+φhφ),
Nττ=(f-1)ττφh+2(f-1)τφhτ(1+φhφ)+f-1φhτ2(1+φhφ)+f-1φhττ(1+φhφ)+f-1φhτ2(1+φhφ);
here f-1 is 1f.

We have by direct computations

1+N-1k+1(N1k+1)ττ=1+1k+1N-1Nττ-k(k+1)2N-2Nτ2=1+1k+1f(f-1)ττ+2f(k+1)h(f-1)τhτ(1+φhφ)+φ(k+1)φhhτ2(1+φhφ)+hττ(k+1)h(1+φhφ)+hτ2(k+1)h(1+φhφ)-k(k+1)2f2(f-1)τ2-2kf(k+1)2h(1+φhφ)(f-1)τhτ-khτ2(k+1)2h2(1+φhφ)2=1+1k+1f(f-1)ττ+2f(k+1)2h(f-1)τhτ(1+φhφ)+hττ(k+1)h(1+φhφ)+hτ2(k+1)h(1+φhφ)-k(k+1)2f2(f-1)τ2+hτ2(k+1)2h2(1+φhφ)(φhφ-k)=1+φhφk+1hττ+hh+hτ2(k+1)h(1+φhφ)-1+φhφh(k+1)2f[hτ(k-φhφfh)12-(f-1)τ(hfk-φhφ)12]2+1k+1[(k-φhφ)-(f-1)τ2f2(kk+1+1k+11+φhφφhφ-k)+(f-1)ττf]1+φhφk+1hττ+hh+1k+1[(k-φhφ)-(f-1)τ2f2(kk+1+1k+11+φhφφhφ-k)+(f-1)ττf],

where, in the last inequality, we use the conditions φhφ-1 and (φhφ)0. Since

(k+1)f-1k+aeij+(k+a)(f-1k+a)ij

is positive definite and -aφhφ-1, thus we can estimate

(k-φhφ)-(f-1)τ2f2k-φhφ-1k-φhφ+(f-1)ττfk+1-(f-1)τ2f2k+a-1k+a+(f-1)ττf=k+1+(k+a)f1k+a(f-1k+a)ττ=f1k+a[(k+1)f-1k+a+(k+a)(f-1k+a)ττ]cf,

where cf is a positive constant depending on 𝑓 and the minimal eigenvalue of (k+1)f-1k+aeij+(k+a)(f-1k+a)ij.

Now we can derive

tb11h-(b11h)2Nσkη(t)(cfh+(2-a-k)b11)+2b11h.

By the uniform bounds on ℎ, 𝑓, 𝜂 and σk, we conclude

tb11h-c1(b11h)2+c2b11h.

Here c1 and c2 are positive constants independent of 𝑡. The maximum principle then gives the upper bound of b11, and the result follows. ∎

When f1, it can be seen from the proof of Lemma 7 that the conditions on 𝑓 and the lower bound of φhφ can be removed.

Corollary 3

Under the assumptions of Theorem 2, we have κiC for all (x,t)Sn-1×[0,T), where 𝐶 is a positive constant independent of 𝑡.

Combining Lemma 5, Lemma 6 and Lemma 7, we see that the principal curvatures of Mt have uniform positive upper and lower bounds. This together with Lemma 3 and Corollary 2 implies that the evolution equation (2.6) is uniformly parabolic on any finite time interval. Thus the result of [25] and the standard parabolic theory show that the smooth solution of (2.6) exists for all time. And by these estimates again, a subsequence of Mt converges in C to a positive, smooth, strictly convex hypersurface M in Rn. To complete the proofs of Theorem 1 and Theorem 2, it only needs to be checked that the support function of M satisfies equation (1.2).

4 Convergence of the Flow

By Lemma 1, Lemma 3 and Lemma 6, the functional

V(t)=Sn-1h(x,t)σk(x,t)dx

is non-decreasing along the flow, and V(t)C for all t0. This tells that

0tV(t)dt=V(t)-V(0)V(t)C,

which leads to 0V(t)dtC. This implies that there exists a subsequence of times tj such that V(tj)0 as tj.

By Lemma 1, we have

(k+1)V(tj)V(tj)=Sn-11f(x)σk2(x)hφ(h)dxSn-1hφ(h)f(x)dx-(Sn-1hσkdx)2.

Since ℎ and σk have uniform positive upper and lower bounds, by passing to the limit, we obtain

Sn-11f(x)σk~2(x)φ(h~)h~dxSn-1h~φ(h~)f(x)dx=(Sn-1h~σk~dx)2,

where σk~ and h~ are the 𝑘-th elementary symmetric function for principal curvature radii and the support function of M. According to the equality condition for the Hölder inequality, there exists a constant c0 such that cφ(h~)σk~(x)=f on Sn-1. Noticing that h~ and σk~ have positive upper and lower bounds, 𝑐 should be positive. The proofs of Theorems 1 and 2 are finished.

Award Identifier / Grant number: 11871432

Award Identifier / Grant number: 11871102

Award Identifier / Grant number: 12071017

Award Identifier / Grant number: 1172005

Funding statement: This work was supported by Natural Science Foundation of China (11871432, 11871102 and 12071017) and Beijing Natural Science Foundation (1172005).

Acknowledgements

The authors would like to thank the anonymous referee for helpful suggestions.

  1. Communicated by: Guozhen Lu

References

[1] B. Andrews, J. McCoy and Y. Zheng, Contracting convex hypersurfaces by curvature, Calc. Var. Partial Differential Equations 47 (2013), no. 3–4, 611–665. 10.1007/s00526-012-0530-3Search in Google Scholar

[2] K. J. Böröczky, E. Lutwak, D. Yang and G. Zhang, The logarithmic Minkowski problem, J. Amer. Math. Soc. 26 (2013), no. 3, 831–852. 10.1090/S0894-0347-2012-00741-3Search in Google Scholar

[3] P. Bryan, M. N. Ivaki and J. Scheuer, A unified flow approach to smooth, even Lp-Minkowski problems, Anal. PDE 12 (2019), no. 2, 259–280. 10.2140/apde.2019.12.259Search in Google Scholar

[4] C.-Q. Chen, Y. Huang and Y.-M. Zhao, Smooth solutions to the Lp dual Minkowski problem, Math. Ann. 373 (2019), no. 3–4, 953–976. 10.1007/s00208-018-1727-3Search in Google Scholar

[5] H. Chen and Q.-R. Li, The Lp dual Minkowski problem and related parabolic flows, preprint. Search in Google Scholar

[6] K.-S. Chou and X.-J. Wang, A logarithmic Gauss curvature flow and the Minkowski problem, Ann. Inst. H. Poincaré Anal. Non Linéaire 17 (2000), no. 6, 733–751. 10.1016/s0294-1449(00)00053-6Search in Google Scholar

[7] K.-S. Chou and X.-J. Wang, The Lp-Minkowski problem and the Minkowski problem in centroaffine geometry, Adv. Math. 205 (2006), no. 1, 33–83. 10.1016/j.aim.2005.07.004Search in Google Scholar

[8] B. Chow and D.-H. Tsai, Expansion of convex hypersurfaces by nonhomogeneous functions of curvature, Asian J. Math. 1 (1997), no. 4, 769–784. 10.4310/AJM.1997.v1.n4.a7Search in Google Scholar

[9] R. J. Gardner, D. Hug, W. Weil, S. Xing and D. Ye, General volumes in the Orlicz–Brunn–Minkowski theory and a related Minkowski problem I, Calc. Var. Partial Differential Equations 58 (2019), no. 1, Paper No. 12. Search in Google Scholar

[10] R. J. Gardner, D. Hug, S. Xing and D. Ye, General volumes in the Orlicz–Brunn–Minkowski theory and a related Minkowski problem II, Calc. Var. Partial Differential Equations 59 (2020), no. 1, Paper No. 15. 10.1007/s00526-019-1657-2Search in Google Scholar

[11] C. Gerhardt, Non-scale-invariant inverse curvature flows in Euclidean space, Calc. Var. Partial Differential Equations 49 (2014), no. 1–2, 471–489. 10.1007/s00526-012-0589-xSearch in Google Scholar

[12] P.-F. Guan and X.-N. Ma, The Christoffel–Minkowski problem. I. Convexity of solutions of a Hessian equation, Invent. Math. 151 (2003), no. 3, 553–577. 10.1007/s00222-002-0259-2Search in Google Scholar

[13] P.-F. Guan and C. Xia, Lp Christoffel–Minkowski problem: The case 1<p<k+1, Calc. Var. Partial Differential Equations 57 (2018), no. 2, Paper No. 69. 10.1007/s00526-018-1341-ySearch in Google Scholar

[14] C. Haberl, E. Lutwak, D. Yang and G. Zhang, The even Orlicz Minkowski problem, Adv. Math. 224 (2010), no. 6, 2485–2510. 10.1016/j.aim.2010.02.006Search in Google Scholar

[15] R. S. Hamilton, Four-manifolds with positive curvature operator, J. Differential Geom. 24 (1986), no. 2, 153–179. 10.4310/jdg/1214440433Search in Google Scholar

[16] Y. He, Q.-R. Li and X.-J. Wang, Multiple solutions of the Lp-Minkowski problem, Calc. Var. Partial Differential Equations 55 (2016), no. 5, Article ID 117. Search in Google Scholar

[17] C. Hu, X.-N. Ma and C. Shen, On the Christoffel–Minkowski problem of Firey’s 𝑝-sum, Calc. Var. Partial Differential Equations 21 (2004), no. 2, 137–155. Search in Google Scholar

[18] Y. Huang, J. Liu and L. Xu, On the uniqueness of Lp-Minkowski problems: The constant 𝑝-curvature case in R3, Adv. Math. 281 (2015), 906–927. 10.1016/j.aim.2015.02.021Search in Google Scholar

[19] Y. Huang, E. Lutwak, D. Yang and G. Zhang, Geometric measures in the dual Brunn–Minkowski theory and their associated Minkowski problems, Acta Math. 216 (2016), no. 2, 325–388. 10.1007/s11511-016-0140-6Search in Google Scholar

[20] D. Hug, E. Lutwak, D. Yang and G. Zhang, On the Lp Minkowski problem for polytopes, Discrete Comput. Geom. 33 (2005), no. 4, 699–715. 10.1007/s00454-004-1149-8Search in Google Scholar

[21] M. N. Ivaki, Deforming a hypersurface by principal radii of curvature and support function, Calc. Var. Partial Differential Equations 58 (2019), no. 1, 2133–2165. 10.1007/s00526-018-1462-3Search in Google Scholar

[22] H.-Y. Jian and J. Lu, Existence of solutions to the Orlicz–Minkowski problem, Adv. Math. 344 (2019), 262–288. 10.1016/j.aim.2019.01.004Search in Google Scholar

[23] H.-Y. Jian, J. Lu and X.-J. Wang, Nonuniqueness of solutions to the Lp-Minkowski problem, Adv. Math. 281 (2015), 845–856. 10.1016/j.aim.2015.05.010Search in Google Scholar

[24] H.-Y. Jian, J. Lu and G. Zhu, Mirror symmetric solutions to the centro-affine Minkowski problem, Calc. Var. Partial Differential Equations 55 (2016), no. 2, Article ID 41. 10.1007/s00526-016-0976-9Search in Google Scholar

[25] N. V. Krylov and M. V. Safonov, A property of the solutions of parabolic equations with measurable coefficients, Izv. Akad. Nauk SSSR Ser. Mat. 44 (1980), no. 1, 161–175, 239. 10.1070/IM1981v016n01ABEH001283Search in Google Scholar

[26] Q.-R. Li, W. Sheng and X.-J. Wang, Flow by Gauss curvature to the Aleksandrov and dual Minkowski problems, J. Eur. Math. Soc. (JEMS) 22 (2020), no. 3, 893–923. 10.4171/JEMS/936Search in Google Scholar

[27] Y.-N. Liu and J. Lu, A flow method for the dual Orlicz–Minkowski problem, Trans. Amer. Math. Soc. 373 (2020), no. 8, 5833–5853. 10.1090/tran/8130Search in Google Scholar

[28] Y.-N. Liu and J. Lu, A generalized Gauss curvature flow related to the Orlicz–Minkowski problem, preprint (2020), https://arxiv.org/abs/2005.02376. 10.29007/qwpkSearch in Google Scholar

[29] J. Lu, Nonexistence of maximizers for the functional of the centroaffine Minkowski problem, Sci. China Math. 61 (2018), no. 3, 511–516. 10.1007/s11425-016-0539-xSearch in Google Scholar

[30] J. Lu and X.-J. Wang, Rotationally symmetric solutions to the Lp-Minkowski problem, J. Differential Equations 254 (2013), no. 3, 983–1005. 10.1016/j.jde.2012.10.008Search in Google Scholar

[31] E. Lutwak, The Brunn–Minkowski–Firey theory. I. Mixed volumes and the Minkowski problem, J. Differential Geom. 38 (1993), no. 1, 131–150. 10.4310/jdg/1214454097Search in Google Scholar

[32] R. Schneider, Convex Bodies: the Brunn–Minkowski Theory, Encyclopedia of Mathematics and its Applications 151, Cambridge University, Cambridge, 2014. Search in Google Scholar

[33] W.-M. Sheng and C.-H. Yi, A class of anisotropic expanding curvature flow, Discrete Contin. Dyn. Syst. Ser. A 40 (2020), 2017–2035. 10.3934/dcds.2020104Search in Google Scholar

[34] J. I. E. Urbas, An expansion of convex hypersurfaces, J. Differential Geom. 33 (1991), no. 1, 91–125. 10.4310/jdg/1214446031Search in Google Scholar

[35] D. Xi, H. Jin and G. Leng, The Orlicz Brunn–Minkowski inequality, Adv. Math. 260 (2014), 350–374. 10.1016/j.aim.2014.02.036Search in Google Scholar

[36] C. Xia, Inverse anisotropic curvature flow from convex hypersurfaces, J. Geom. Anal. 27 (2017), no. 3, 2131–2154. 10.1007/s12220-016-9755-2Search in Google Scholar

[37] Y.-J. Sun and Y.-M. Long, The planar Orlicz Minkowski problem in the L1-sense, Adv. Math. 281 (2015), 1364–1383. 10.1016/j.aim.2015.03.032Search in Google Scholar

[38] G. Zhu, The logarithmic Minkowski problem for polytopes, Adv. Math. 262 (2014), 909–931. 10.1016/j.aim.2014.06.004Search in Google Scholar

[39] D. Zou and G. Xiong, Orlicz–John ellipsoids, Adv. Math. 265 (2014), 132–168. 10.1016/j.aim.2014.07.034Search in Google Scholar

Received: 2020-08-07
Revised: 2020-09-13
Accepted: 2020-09-14
Published Online: 2020-10-02
Published in Print: 2021-02-01

© 2020 Walter de Gruyter GmbH, Berlin/Boston

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

Downloaded on 26.4.2024 from https://www.degruyter.com/document/doi/10.1515/ans-2020-2108/html
Scroll to top button