A RIGIDITY THEOREM ON THE SECOND FUNDAMENTAL FORM FOR SELF-SHRINKERS QI DING 7 1 0 2 Abstract. In Theorem 3.1 of [12], we proved a rigidity result for self-shrinkers under the integral condition on the norm of the second fundamental form. In this paper, we n a relax thesuch boundto any finite constant (see Theorem 4.4 for details). J 7 2 ] G 1. Introduction D Self-similar solutions for mean curvature flow play a key role in the understanding . h the possible singularities that the flow goes through. Self-shrinkers are type I singularity t a modelsoftheflow. Huiskenmadeapioneerworkonself-shrinkingsolutionsoftheflow[22, m 23]. ColdingandMinicozzi [8] gave acomprehensive studyforself-shrinkinghypersurfaces [ and solve a long-standing conjecture raised by Huisken. 1 v Colding-Ilmanen-Minicozzi [9] showed that cylindrical self-shrinkers are rigid in a very 5 strong sense. Namely, any other shrinker that is sufficiently close to one of them on 4 a large, but compact set must itself be a round cylinder. See [25] by Guang-Zhu for 9 further results. Lu Wang in [37,38] proved strong uniqueness theorems for self-shrinkers 7 0 asymptotic to regular cones or generalized cylinders of infinite order. . 1 For Bernstein type theorems, Ecker-Huisken [17] and Wang [36] showed the nonexis- 0 7 tence of nontrivial graphic self-shrinkinghypersurfaces in Euclidean space. For 2 n 6, 1 Guang-Zhu showed that any smooth complete self-shrinker in Rn+1 which is grap≤hica≤l in- : v sidealarge, butcompact, setmustbeahyperplane. Ding-Xin-Yang [14] studiedthesharp Xi rigidity theoremswiththeconditiononGaussmapofself-shrinkers. Inhighcodimensions, see [2,3,10,13,26] for more Bernstein type theorems. r a Le-Sesum [30] showed that any complete embedded self-shrinking hypersurface with polynomial volume growth mustbeahyperplaneprovided thesquared normof thesecond fundamental form B 2 < 1. Cao-Li [1] showed that any complete self-shrinker (with high | | 2 codimension) with polynomial volume growth must be a generalized cylinder provided B 2 1. Later, Cheng-Peng [5] removed the condition of polynomial volume growth in | | ≤ 2 the case of B 2 < 1 (See [4,6,12,42] for more results on the gap theorems of the norm of | | 2 the second fundamental form). In [12], Ding-Xin proved a rigidity result for self-shrinkers if the integration of B n is small. In this paper, we improve the small constant to any | | finite constant. For a complete properly immersed self-shrinker Σn Rn+1, Ilmanen showed that there existsacone Rn+1 withthecrosssection beingaco⊂mpactsetinSn suchthatλΣn C ⊂ → C as λ 0 locally in the Hausdorff metric on closed sets (see [28] Lecture 2, B, remark + → The author would like to thank Yuanlong Xin for his interest in this work. He is supported partially by NSFC. 1 2 QIDING on p.8). In [35], Song gave a simple proof by ”maximum principle for self-shrinkers”. For high codimensions, with backward heat kernel (see [8]) we show the uniqueness of tangent cones at infinity for self-shrinkers with Euclidean volume growth in the current sense with the condition on mean curvature(see Theorem 3.3). ǫ-regularity theorems for the mean curvature flow have been studied by Ecker [15,16], Han-Sun [19], Ilmanen [27], Le-Sesum [29]. Now we use the one showed by Ecker [16] starting from self-similar solutions, and obtain the curvature estimates for self-shrinkers, see Theorem 4.2. Combining Theorem 3.3, Theorem 4.2 and backward uniqueness for parabolic operators [21], we can show that self-shrinkers with finite integration on B n | | must be planes, which improves a previous rigidity theorem in [12]. A litter more, we obtain the following Theorem. Theorem 1.1. Let M be an n-dimensional properly non-compact self-shrinker with com- pact boundary in Rn+m, B denote the second fundamental form of M. If (1.1) lim B ndµ = 0, r | | →∞ZM B2r Br ∩ \ then M must be an n-plane through the origin. 2. Preliminary Let M be an n-dimensional C2-submanifold in Rn+m with the induced metric. Let and betheLevi-Civita connections on M and Rn+m, respectively. We definethesecon∇d ∇ fundamental form B of M by B(V,W)= ( W)N = W W V V V ∇ ∇ −∇ for any V,W Γ(TM), where the mean curvature vector H of M is given by H = trace(B) = n∈ B(e ,e ), where e is a local orthonormal frame field of M. i=1 i i { i} In this pPaper, Mn is said to be a self-shrinker in Rn+m if its mean curvature vector satisfies XN (2.1) H = , − 2 where X = (x , ,x ) Rn+m is the position vector of M in Rn+m, and ( )N 1 n+m ··· ∈ ··· stands for the orthogonal projection into the normal bundle NM. Let ( )T denote the ··· orthogonal projection into the tangent bundle TM. We define a second order differential operator as in [8] by L X2 X2 1 f = e| 4| div e−| 4| f = ∆f X, f L ∇ − 2h ∇ i (cid:18) (cid:19) for any f C2(M). Let ∆ be the Laplacian of M, then for self-shrinkers, ∈ (2.2) ∆ X 2 = 2 X,∆X +2 X 2 = 2 X,H +2n = XN 2+2n. | | h i |∇ | h i −| | In [8], Colding and Minicozzi defined a function F for self-shrinking hypersurfaces X0,t0 inEuclideanspace. Obviously, hypersurfacescan begeneralized tosubmanifoldsnaturally in this definition. Set Φ C (Rn+m) for any t > 0 by t ∞ ∈ 1 X2 Φt(X) = e−|4t| . (4πt)n/2 A RIGIDITY THEOREM ON THE SECOND FUNDAMENTAL FORM FOR SELF-SHRINKERS 3 For an n-complete submanifold M in Rn+m, we define a functional F on M by t 1 X2 Ft(M) = Φtdµ = e−|4t| dµ for t > 0, (4πt)n/2 ZM ZM where dµ is the volume element of M. Sometimes, we write F for simplicity if no ambigu- t ous in the text. If a self-shrinker is proper, then it is equivalent to that it has Euclidean volume growth at most by [7] and [11]. We shall only consider proper self-shrinkers in the following text. Now we use the backward heat kernel to give a monotonicity formula for self-shrinkers with arbitrary codimensions, which is essentially same as self-shrinking hypersurfaces es- tablished by Colding-Minicozzi in [8]. Lemma 2.1. For any 0 < t t , each complete immersed self-shrinker Mn with 1 2 boundary ∂M (may be empty)≤in Rn≤+m∞satisfies t2 Φ (X) F (M) F (M) = XT,ν s ds t2 − t1 − h ∂Mi 2s (2.3) Zt1 (cid:18)Z∂M (cid:19) t2 1 1 + 1 XN 2Φ (X)dµ ds. s 4s − s | | Zt1 (cid:18) (cid:19)(cid:18)ZM (cid:19) Proof. We differential F (M) with respect to t, t (2.4) Ft′ = (4π)−n2t−(n2+1) −n2 + |X4t|2 e−|X4t|2dµ. ZM (cid:18) (cid:19) A straightforward calculation shows (see also [11]) e|X4t|2div e−|X4t|2 X 2 = ∆ X 2+ 1 X 2 X 2 − ∇| | − | | 4t∇| | ·∇| | (cid:18) (cid:19) 1 = 2 H,X 2n+ XT 2 − h i− t| | (2.5) XT 2 = XN 2+ | | 2n | | t − 1 X 2 = 1 XN 2+ | | 2n, − t | | t − (cid:18) (cid:19) where the third equality above uses the self-shrinkers’ equation (2.1). Then Ft′ =(4π)−n2t−(n2+1) − 41div e−|X4t|2∇|X|2 − 14 1− 1t |XN|2e−|X4t|2 dµ ZM (cid:18) (cid:18) (cid:19) (cid:18) (cid:19) (cid:19) (2.6) =1(4π)−n2t−(n2+1) 2 XT,ν∂M e−|X4t|2 1 1 XN 2e−|X4t|2dµ 4 − h i − − t | | (cid:18) Z∂M (cid:18) (cid:19)ZM (cid:19) 1 1 1 = XT,ν Φ (X) 1 XN 2Φ (X)dµ, ∂M t t − 2t h i − 4t − t | | Z∂M (cid:18) (cid:19)ZM where ν is the normal vector of ∂M in Γ(TM). Then we complete the proof by inte- ∂M gration from t to t . (cid:3) 1 2 Denote 1 1 1 (2.7) G (M) , F (M)+ XT,ν Φ (X) = 1 XN 2Φ (X)dµ. t t′ 2t h ∂Mi t −4t − t | | t Z∂M (cid:18) (cid:19)ZM 4 QIDING The above Lemma implies G (M) 0 for each self-shrinker and t 1. If ∂M is bounded t ≤ ≥ and has finite (n 1)-dimensional Hausdorff measure, then the limit − t lim G (M)ds s t →∞(cid:18)Z1 (cid:19) always exists, and is a finite negative number. Hence, it’s clear that lim F (M) exists. t t →∞ 3. Uniqueness of tangent cones at infinity for self-shrinkers For any n-rectifiable varifold V Rn+m with multiplicity one, we define a functional ⊂ Ξ by t 1 X2 Ξt(V,f) = (4πt)n/2 fe−|4t| dµV ZsptV for any t > 0, where µ is a measure on Rn+m associated with the Radon measure of V V in Rn+m G(n,n+m). × We suppose that M is a self-shrinker in Rn+m B with boundary ∂M ∂B for R R some R 1 and n 1(∂M) < . Let φ C1(Rn+m\ 0 ) be a homogeneous⊂function of − degree z≥ero. NamHely, for any 0∞= X Rn∈+m, \{ } 6 ∈ φ(X) = φ(X ξ) = φ(ξ) | | with ξ = X . Then X | | δ x x ij i j (3.1) ∂xiφ = X − X 3 ∂ξjφ, j (cid:18)| | | | (cid:19) X and δ x x (3.2) |∇φ|2 = Xjk2 − Xj 4k ∂ξjφ∂ξkφ≤ |X|−2 ∂ξjφ 2 , |X|−2|φ|21. j,k (cid:18)| | | | (cid:19) j X X(cid:0) (cid:1) Taking the derivative of Ξ (M,φ) on t gets t ∂tΞt(M,φ) = (4π)−n2t−(n2+1) n + |X|2 φe−|X4t|2dµ −2 4t (3.3) ZM (cid:18) (cid:19) =(4π)−n2t−(n2+1) φdiv e−|X4t|2 X 2 φ 1 1 XN 2e−|X4t|2 dµ. −4 ∇| | − 4 − t | | ZM (cid:18) (cid:18) (cid:19) (cid:18) (cid:19) (cid:19) Combining X φ =0, we have ·∇ φdiv e−|X4t|2 X 2 dµ −4 ∇| | ZM (cid:18) (cid:19) = 1div φe−|X4t|2 X 2 dµ+ 1 φ X 2e−|X4t|2dµ −4 ∇| | 4∇ ·∇| | (3.4) ZM (cid:18) (cid:19) ZM = 1 φ XT,ν∂M e−|X4t|2 + 1X φe−|X4t|2dµ − 2 h i 2 ·∇ Z∂M ZM = 1 φ XT,ν∂M e−R4t2 1 XN φe−|X4t|2dµ. − 2 h i − 2 ·∇ Z∂M ZM A RIGIDITY THEOREM ON THE SECOND FUNDAMENTAL FORM FOR SELF-SHRINKERS 5 Set cR = 2−1(4π)−n2R n−1(∂M). Substituting (3.2) and (3.4) into (3.3) gets ·H ∂tΞt(M,φ) 2−1(4π)−n2t−(n2+1) XN φe−|X4t|2dµ | | ≤ | |·|∇ | (cid:18)ZM + φ0Re−R4t2 n−1(∂M) + φ0 Gt(M) | | H | | | | (cid:19) 2−1(4π)−n2t−(n2+1) |XN| φ1e−|X4t|2dµ+ φ0 Gt(M) +cRt−(n2+1) ≤ X | | | | | | ZM | | (cid:16) (cid:17) (3.5) φ0 Gt(M) +cRt−(n2+1) ≤| | | | (cid:16) (cid:17) 1/2 1/2 +2−1(4π)−n2t−(n2+1) φ1 XN 2e−|X4t|2dµ X −2e−|X4t|2dµ | | | | | | (cid:18)ZM (cid:19) (cid:18)ZM (cid:19) φ0 Gt(M) +cRt−(n2+1) ≤| | | | + φ(cid:16)1 Gt(M) 1/2 t (cid:17)(4π)−n2t−(n2+2) X −2e−|X4t|2dµ 1/2. | | | | t 1 | | r − (cid:18) ZM (cid:19) Put D = M B for every r > 0. There is a constant c > 0 depending only on M such r r 0 ∩ that for all r > 0 1dµ < c rn. 0 ZDr Note M Rn+m B . Then for n 2, t R2, one has R ⊂ \ ≥ ≥ n t X2 n ∞ t X2 t−2 e−|4t| dµ t−2 e−|4t| dµ X 2 ≤ X 2 ZM | | k= 1 X[log(tR−2)]ZD2k+1√t\D2k√t | | − − 2log2 t−n2 ∞ 1 e−4k−1 1dµ ≤ 4k k= 1 X[log(tR−2)] ZD2k+1√t\D2k√t − − 2log2 1 (3.6) c ∞ 4 ke 4k2(k+1)n +c − 4 k2(k+1)n 0 − − 0 − ≤ Xk=0 k= 1 X[log(tR−2)] − − 2log2 1+[log(tR−2)] 2log2 c0 ∞ 2k(n−2)+ne−4k−1 +c0 2−k(n−2)+n ≤ k=0 k=1 X X n (4π)2c1(1+logt 2logR), ≤ − where c is a constant depending only on n,c . Therefore 1 0 √1+logt t 1/2 ∂tΞt(M,φ) √c1 φ1 Gt(M) + φ0 Gt(M) +cRt−(n2+1) | | ≤ t | | t 1 | | | | (3.7) 1+logt (cid:12)(cid:12) − (cid:12)(cid:12) (cid:16) (cid:17) c1 φ1 +cR(cid:12)(cid:12) t−(n2+1) φ0(cid:12)(cid:12)+(φ0 + φ1) Gt(M). ≤ 4t(t 1)| | | | | | | | | | − Theorem 3.1. Let M be an n-dimensional self-shrinker in Rn+m with Euclidean volume growth and boundary ∂M ∂B . If R ⊂ (3.8) limsup r1 n H < , − | | ∞ r→∞ (cid:18) ZM∩Br (cid:19) 6 QIDING then there is a sequence t such that i → ∞ M , t 1M = X Rn+m t X M ti −i { ∈ | i ∈ } converges to a cone C in Rn+m. Proof. By co-area formula, we can choose R > 0 so that n 1(∂M) < with ∂M ′ − H ∞ ⊂ ∂B . Denote R by R for convenience. Let M , t 1M = X Rn+m tX M for R ′ t − ′ { ∈ | ∈ } any t > 0. Since M has Euclidean volume growth and (3.8) holds, then by compactness of varifolds, there exists an n-rectifiable varifold T in Rn+m with integer multiplicity and a sequence of t such that M = t 1M ⇀ T in the sense of Radon measure (See 42.7 i ti −i Theorem of [34] for example). Denote φ and Ξ (M,φ) as above. Set µ be the volume element of M . Since t t t 1 X2 1 X2 (3.9) Ξt2(M,φ) = (4πt2)n/2 φe−|4t|2 dµ = (4π)n/2 φe−| 4| dµt = Ξ1(Mt,φ), ZM ZMt then for all R > 0 1 X2 (3.10) il→im∞Ξ1(MtiR,φ) = il→im∞ΞR2(Mti,φ) = (4πR2)n/2 ZT φ e−|4R|2 dµT = ΞR2(T,φ). Note that G (M) does not change sign for t > 1. Fixing 0 < r < R < , from (3.7) we t ∞ have t2R2 i Ξ (M,φ) Ξ (M,φ) ∂ Ξ (M,φ)ds (cid:12) t2ir2 − t2iR2 (cid:12) ≤ Zt2ir2 | s s | (3.11) (cid:12)(cid:12) t2iR2 c1 1+logs φ1+c(cid:12)(cid:12)R φ0s−(n2+1)+( φ0+ φ1) Gs(M) ds ≤ 4s(s 1)| | | | | | | | | | Zt2ir2 (cid:18) − (cid:19) c1 φ1 ∞ 1+logsds+ 2 (tir)−n−2cR φ0+(φ0 + φ1) t2iR2Gs(M)ds ≤ 4 | | Zt2ir2 s(s−1) n | | | | | | (cid:12)(cid:12)Zt2ir2 (cid:12)(cid:12) (cid:12) (cid:12) for all ti with rti 2. Since (cid:12) (cid:12) ≥ (cid:12) (cid:12) t2iR2 t2iR2 t2iR2 1 G (M)ds F (M)ds+ XT,ν Φ (X) ds (cid:12)(cid:12)Zt2ir2 s (cid:12)(cid:12) ≤ (cid:12)(cid:12)Zt2ir2 t′ Zt2ir2 (cid:18)2s Z∂Mh ∂Mi s (cid:19) (cid:12)(cid:12) (3.12) (cid:12)(cid:12) (cid:12)(cid:12) (cid:12)(cid:12) t2iR2 R (cid:12)(cid:12) ≤(cid:12)(cid:12)Ft2ir2(M)−Ft2iR(cid:12)2(M(cid:12))(cid:12)+Zt2ir2 (cid:18)2sHn−1(∂M)(4πs)−n/2(cid:19)ds (cid:12) (cid:12) (cid:12) R =(cid:12)F (M) F (M)(cid:12)+ (4π) n/2 n 1(∂M)(t r) n t2ir2 − t2iR2 n − H − i − (cid:12) (cid:12) and limt (cid:12) Ft exists, we obtain(cid:12) →∞(cid:12) (cid:12) (3.13) ilim Ξ1(Mtir,φ) = ilim Ξ1(MtiR,φ) = ΞR2(T,φ). →∞ →∞ Hence 1 X2 (3.14) Ξt(T,φ) = φe−|4t| dµT (4πt)n/2 ZT is independent of t (0, ). ∈ ∞ Clearly, 0< n(T B ) c rn r 2 H ∩ ≤ A RIGIDITY THEOREM ON THE SECOND FUNDAMENTAL FORM FOR SELF-SHRINKERS 7 for some constant c > 0 and all r > 0. By the following lemma for V(r) = φ dµ , 2 T Br T we conclude that ∩ R (3.15) r−n φ dµT ZT∩Br is a constant independent of r. An analog argument as the proof of 19.3 in [34] implies that T is a cone. (cid:3) Lemma 3.2. Let V(r) be a monotone nondecreasing continuous function on [0, ) with ∞ V(0) = 0 and V(r) c rn for some constant c > 0. If the quantity 3 3 ≤ 1 ∞ r2 (3.16) e−4tdV(r) (4πt)n/2 Z0 is a constant for any t > 0, then r nV(r) is a constant. − Proof. There are constants κ ,κ > 0 such that for all t > 0 0 1 (3.17) ∞e−rt2dV(r) = κ0tn/2 = κ1 ∞e−rt2drn, Z0 Z0 namely, (3.18) ∞e−rt2d(V(r) κ1rn)= 0. − Z0 Integrating by parts implies (3.19) ∞(V(r) κ1rn)re−rt2dr = 0. − Z0 Suppose that there is a constant r > 0 such that V(r ) κ rn > 0 (Or else we complete the proof by (3.19)). Then there is0a 0 < δ < r0 and ǫ0>−0 s1uc0h that V(r) κ rn ǫ for 2 − 1 ≥ all r (r δ,r +δ). Set t = 2r2, then in (0, ) the function ∈ 0− 0 p p 0 ∞ r2 rpe−tp attains its maximal value at r = r . 0 Now we claim (3.20) plim pr1p2+e1p2 r0+δrpe−rtp2dr = ∞ e−t2dt = √π. →∞ 0 Zr0−δ Z−∞ In fact, I(p) ,pr0p12+e1p2 Zr0r−0+δδrpe−rtp2dr = p21ep2 Z−rrδ0δ0(1+s)pe−p2(1+s)2ds (3.21) = rδ0√p 1+ t pe−p2(cid:16)√2tp+tp2(cid:17)dt Z−rδ0√p(cid:18) √p(cid:19) = rδ0√p eplog(cid:16)1+√tp(cid:17)e−√pt−t22dt. Z−rδ0√p When 1 s < , a simple calculation implies −2 ≤ ∞ 8 s2 s3 min 0, s3 log(1+s) s+ . 3 ≤ − 2 ≤ 3 (cid:26) (cid:27) 8 QIDING Combining the above inequality, we get limsupI(p) limsup rδ0√p e−t2+3t√3pdt (3.22) p→∞ ≤ p→∞ Z−rδ0√p δ √p = lim r0 e−t2(1−3√tp)dt = ∞ e−t2dt, p→∞Z−rδ0√p Z−∞ and liminfI(p) lim rδ0√pe−t2dt+liminf 0 e−t2+38√t3pdt (3.23) p→∞ ≥p→∞Z0 p→∞ Z−rδ0√p 0 = ∞e−t2dt+ lim e−t2(cid:16)1−38√tp(cid:17)dt = ∞ e−t2dt. Z0 p→∞Z−rδ0√p Z−∞ Hence we have shown (3.20). For p > 1, (3.24) pr021p+e1p2 Zr0∞+δrn+pe−rtp2dr = r0nZrδ0∞√pe(n+p)log(cid:16)1+√tp(cid:17)e−√pt−t22dt ≤r0nZrδ0∞√pe(n+p)√tpe−√pt−t22dt ≤ r0nZrδ0∞√pe√npt−t22dt. Then (3.25) lipminf pr12p+e1p2 ∞(V(r)−κ1rn)rpe−rtp2dr →∞ 0 Z0 ≥lipminf pr12p+e1p2 ǫ r0+δrpe−rtp2dr−κ1 r0−δrn+pe−rtp2dr−κ1 ∞ rn+pe−rtp2dr →∞ 0 (cid:18) Zr0 δ Z0 Zr0+δ (cid:19) − ≥ǫ√π−κ1r0nlimp→s∞up pr012p+e1p2 Z0r0−δrpe−p2rr022dr+Zrδ0∞√pe√npt−t22dr! =ǫ√π−κ1r0nlimp→s∞up Z−−√rpδ0√peplog(cid:16)1+√tp(cid:17)e−√pt−t22dt+Zrδ0∞√pe−t2(cid:16)12−√npt(cid:17)dr! ≥ǫ√π−κ1r0nlimsup −rδ0√pe√pte−√pt−t22dt = ǫ√π. p→∞ Z−√p ! Taking the derivative of t in (3.19) yields (3.26) ∞(V(r) κ1rn)r2k+1e−rt2dr = 0 − Z0 for any t > 0 and k = 0,1,2 . If we choose p = 2k+1, r2 > e, t = 2r2 in (3.25), then ··· 0 p p 0 we get the contradiction provided k is sufficiently large. Hence V(r) κ rn 0. (cid:3) 1 − ≡ Theorem 3.3. Let M be an n dimensional smooth self-shrinker with Euclidean volume growth and boundary ∂M ∂B in Rn+m. If (3.8) holds, then the limit lim r 1M R r − ⊂ →∞ exists and is cone, namely, the tangent cone at infinity of M is a unique cone. A RIGIDITY THEOREM ON THE SECOND FUNDAMENTAL FORM FOR SELF-SHRINKERS 9 Proof. We claim (3.27) lim r n φdµ − r →∞(cid:18) ZM∩Br (cid:19) exists for every homogeneous function φ C1(Rn+m 0 ) with degree zero. Suppose ∈ \{ } (3.28) limsupr−n φdµ > liminfr−n φdµ r→∞ ZM∩Br r→∞ ZM∩Br for some homogeneous function φ C1(Rn+m 0 ) with degree zero. Then there exist ∈ \{ } two sequences p and q such that i i → ∞ → ∞ (3.29) lim p−i n φdµ > lim qi−n φdµ. i→∞ ZM∩Bpi i→∞ ZM∩Bqi By compactness of varifolds and Theorem 3.1, there exist two cones C ,C in Rn+m with 1 2 integer multiplicities and subsequences p of p and q of q such that M ⇀ C and ki i ki i pki 1 M ⇀ C in the sense of Radon measure. So we have qki 2 φdµ = lim φdµ = lim p n φdµ ZC1∩B1 C1 i→∞ZMpki∩B1 pki i→∞ −ki ZM∩Bpki (3.30) >il→im∞qk−inZM∩Bqki φdµ = il→im∞ZMqki∩B1φdµqki = φdµ , C2 ZC2∩B1 which implies X2 X2 (3.31) φe−| 4| dµC1 > φe−| 4| dµC2 ZC1 ZC2 by co-area formula. From the previous argument, the limit 1 X2 (3.32) lim Ξt(M,φ) = lim φe−|4t| dµ t t (4πt)n/2 →∞ →∞ ZM exists. It infers that X2 X2 1 X2 (3.33) ZC1φe−| 4| dµC1 =il→im∞ZMpki φe−| 4| = tl→im∞ tn/2 ZM φe−|4t| dµ X2 X2 =il→im∞ZMqki φe−| 4| = ZC2φe−| 4| dµC2. However, (3.33) contradicts (3.31). Hence, the claim (3.27) holds. Iflim r 1M ⇀ C+, lim s 1M ⇀ C andC+ = C arecones, thenfrom(3.33) i→∞ i− i→∞ −i − 6 − one has X2 X2 (3.34) φe−| 4| dµC+ = φe−| 4| dµC − ZC+ ZC− for every homogeneous function φ C1(Rn+m 0 ) with degree zero. It’s clear that ∈ \{ } (3.35) φ = φ. ZC+∩∂B1 ZC−∩∂B1 10 QIDING Arbitrariness of φ implies C+ = C . Therefore, the tangent cone at infinity of M is a − unique cone. (cid:3) 4. A rigidity theorem for self-shrinkers Let us recall an ǫ-regularity theorem for mean curvature flow showed by Ecker (A litter different from Theorem 1.8 in [16]). Theorem 4.1. For p [n,n+2], there exists a constant ǫ > 0 such that for any smooth 0 properly immersed solu∈tion = ( ) of mean curvature flow in Rn+m, every X t t ( 4,0) 0 which the solution reaches aMt timeMt ∈[ −1,0), the assumption 0 ∈ − 1 ρ2 (4.1) I , sup − B p ǫ X0,t0 √−t0≤ρ<ρ′≤2(ρ′2−ρ2)n+22−p Z−ρ′2 ZMt∩B2(X0)| | ≤ 0 implies 2 (4.2) sup σ2 sup sup |B|2 ≤ ǫ−01IX0,t0 p . σ∈[0,1] t∈(t0−(1−σ)2,t0) Mt∩B1−σ(X0) ! (cid:0) (cid:1) For completeness, we give the proof in appendix which is based on Ecker’s proof. Let us consider the mean curvature flow in Theorem 4.1 which starts from a self-shrinker. Let M be a self shrinker, then the one-parameter family = √ tM is a mean curvature t M − flow for 4 t < 0. In this case, − ≤ IX0,t0 = sup ρ′2−ρ2 −n+22−p −ρ2 |B|p dt √−t0≤ρ<ρ′≤2(cid:0) (cid:1) Z−ρ′2 Z√−tM∩B2(X0) ! 1 (4.3) = sup ρ′2 ρ2 −n+22−p ρ B p 2 dr √−t0≤ρ<ρ′≤2(cid:0) − (cid:1) Zρ1′ Zr1M∩B2(X0)| | ! r3 1 = sup 2 ρ′2 ρ2 −n+22−p ρ rp−n−3 B pdµ dr. √−t0≤ρ<ρ′≤2 (cid:0) − (cid:1) Zρ1′ ZM∩B2r(rX0)| | ! For any 1 < t < 0 and X √ t M, I ǫ implies −4 0 0 ∈ − 0 X0,t0 ≤ 0 1 2 sup sup B 2 ǫ 1I p . (4.4) 4 | | ≤ −0 X0,t0 t∈(t0−14,t0) √−tM∩B12(X0) (cid:0) (cid:1) Hence 2 (4.5) sup sup |B|2 ≤ 4 ǫ−01IX0,t0 p . t∈(2,(−t0)−1/2) 1tM∩B12(X0) (cid:0) (cid:1) Now we have the following curvature estimates for self-shrinkers. Theorem 4.2. Let M be an n dimensional proper self-shrinker in Rn+m. If for some p [n,n+2) there is ∈ (4.6) lim B pdµ = 0, R→∞ZM∩B2R\BR| |