Université de Cergy-Pontoise. Insitut Universitaire de France. joint work with Frank Merle. Hatem Zaag. wave equation

Similar documents
DETERMINATION OF THE BLOW-UP RATE FOR THE SEMILINEAR WAVE EQUATION

arxiv: v1 [math.ap] 9 Jun 2016

A Liouville theorem for vector valued semilinear heat equations with no gradient structure and applications to blow-up

THE L 2 -HODGE THEORY AND REPRESENTATION ON R n

arxiv: v3 [math.ap] 20 Jun 2017

Math The Laplacian. 1 Green s Identities, Fundamental Solution

The Schrödinger equation with spatial white noise potential

Anisotropic partial regularity criteria for the Navier-Stokes equations

On the uniqueness of heat flow of harmonic maps and hydrodynamic flow of nematic liquid crystals

On Smoothness of Suitable Weak Solutions to the Navier-Stokes Equations

Regularity of Weak Solution to Parabolic Fractional p-laplacian

arxiv:math/ v1 [math.ap] 28 Oct 2005

BLOWUP THEORY FOR THE CRITICAL NONLINEAR SCHRÖDINGER EQUATIONS REVISITED

CRITERIA FOR THE 3D NAVIER-STOKES SYSTEM

BIHARMONIC WAVE MAPS INTO SPHERES

A note on W 1,p estimates for quasilinear parabolic equations

MATH 425, HOMEWORK 3 SOLUTIONS

THE WAVE EQUATION. d = 1: D Alembert s formula We begin with the initial value problem in 1 space dimension { u = utt u xx = 0, in (0, ) R, (2)

Regularity of the p-poisson equation in the plane

Stationary Kirchhoff equations with powers by Emmanuel Hebey (Université de Cergy-Pontoise)

Some lecture notes for Math 6050E: PDEs, Fall 2016

LORENTZ ESTIMATES FOR WEAK SOLUTIONS OF QUASI-LINEAR PARABOLIC EQUATIONS WITH SINGULAR DIVERGENCE-FREE DRIFTS TUOC PHAN

Liquid crystal flows in two dimensions

Laplace s Equation. Chapter Mean Value Formulas

A Liouville theorem for a heat equation and applications for quenching

A new regularity criterion for weak solutions to the Navier-Stokes equations

Integro-differential equations: Regularity theory and Pohozaev identities

Régularité des équations de Hamilton-Jacobi du premier ordre et applications aux jeux à champ moyen

Ancient solutions to Geometric Flows Lecture No 2

Free energy estimates for the two-dimensional Keller-Segel model

Sobolev Spaces. Chapter 10

The Concentration-compactness/ Rigidity Method for Critical Dispersive and Wave Equations

Elliptic PDEs of 2nd Order, Gilbarg and Trudinger

The Sine-Gordon regime of the Landau-Lifshitz equation with a strong easy-plane anisotropy

The continuity method

Math 54 - HW Solutions 5

On the Brezis and Mironescu conjecture concerning a Gagliardo-Nirenberg inequality for fractional Sobolev norms

Lecture No 1 Introduction to Diffusion equations The heat equat

Solutions to the Nonlinear Schrödinger Equation in Hyperbolic Space

Strong uniqueness for stochastic evolution equations with possibly unbounded measurable drift term

A REMARK ON AN EQUATION OF WAVE MAPS TYPE WITH VARIABLE COEFFICIENTS

ESTIMATES FOR ELLIPTIC HOMOGENIZATION PROBLEMS IN NONSMOOTH DOMAINS. Zhongwei Shen

A class of domains with fractal boundaries: Functions spaces and numerical methods

Sharp energy estimates and 1D symmetry for nonlinear equations involving fractional Laplacians

Homogenization and error estimates of free boundary velocities in periodic media

L 3, -Solutions to the Navier-Stokes Equations and Backward Uniqueness

Some definitions. Math 1080: Numerical Linear Algebra Chapter 5, Solving Ax = b by Optimization. A-inner product. Important facts

S chauder Theory. x 2. = log( x 1 + x 2 ) + 1 ( x 1 + x 2 ) 2. ( 5) x 1 + x 2 x 1 + x 2. 2 = 2 x 1. x 1 x 2. 1 x 1.

Existence and Continuation for Euler Equations

Partial Differential Equations, 2nd Edition, L.C.Evans Chapter 5 Sobolev Spaces

Sobolev regularity for the Monge-Ampère equation, with application to the semigeostrophic equations

THE INVERSE FUNCTION THEOREM

Spectral Gap and Concentration for Some Spherically Symmetric Probability Measures

ON WEAKLY NONLINEAR BACKWARD PARABOLIC PROBLEM

The Gaussian free field, Gibbs measures and NLS on planar domains

Singular Integrals. 1 Calderon-Zygmund decomposition

Relation between Distributional and Leray-Hopf Solutions to the Navier-Stokes Equations

NONLINEAR PROPAGATION OF WAVE PACKETS. Ritsumeikan University, and 22

Lecture No 2 Degenerate Diffusion Free boundary problems

HOMEOMORPHISMS OF BOUNDED VARIATION

arxiv: v1 [math.ap] 25 Jul 2012

NONLOCAL DIFFUSION EQUATIONS

Growth Theorems and Harnack Inequality for Second Order Parabolic Equations

Improved estimates for the Ginzburg-Landau equation: the elliptic case

On non negative solutions of some quasilinear elliptic inequalities

Applied Math Qualifying Exam 11 October Instructions: Work 2 out of 3 problems in each of the 3 parts for a total of 6 problems.

REVIEW OF ESSENTIAL MATH 346 TOPICS

Observability and measurable sets

SYMMETRY OF POSITIVE SOLUTIONS OF SOME NONLINEAR EQUATIONS. M. Grossi S. Kesavan F. Pacella M. Ramaswamy. 1. Introduction

Wave Equations: Explicit Formulas In this lecture we derive the representation formulas for the wave equation in the whole space:

Final Exam. Monday March 19, 3:30-5:30pm MAT 21D, Temple, Winter 2018

Math 497 R1 Winter 2018 Navier-Stokes Regularity

SHARP BOUNDARY TRACE INEQUALITIES. 1. Introduction

The Kato square root problem on vector bundles with generalised bounded geometry

The Kato square root problem on vector bundles with generalised bounded geometry

GENERATORS WITH INTERIOR DEGENERACY ON SPACES OF L 2 TYPE

u xx + u yy = 0. (5.1)

A Priori Bounds, Nodal Equilibria and Connecting Orbits in Indefinite Superlinear Parabolic Problems

A Product Property of Sobolev Spaces with Application to Elliptic Estimates

Regularity and Decay Estimates of the Navier-Stokes Equations

Partial regularity for suitable weak solutions to Navier-Stokes equations

Two dimensional exterior mixed problem for semilinear damped wave equations

An introduction to Birkhoff normal form

The Navier Stokes Equations for Incompressible Flows: Solution Properties at Potential Blow Up Times

Uniqueness of ground state solutions of non-local equations in R N

Universität des Saarlandes. Fachrichtung 6.1 Mathematik

The Dirichlet problem for non-divergence parabolic equations with discontinuous in time coefficients in a wedge

arxiv: v1 [math.ap] 27 Feb 2011

Ancient solutions to geometric flows

Dynamics of energy-critical wave equation

ISABELLE GALLAGHER AND MARIUS PAICU

Recent developments on the global behavior to critical nonlinear dispersive equations. Carlos E. Kenig

Non-radial solutions to a bi-harmonic equation with negative exponent

A Dirichlet problem in the strip

Blow up and regularity for fractal Burgers equation

Nonlinear aspects of Calderón-Zygmund theory

TD M1 EDP 2018 no 2 Elliptic equations: regularity, maximum principle

A GENERAL CLASS OF FREE BOUNDARY PROBLEMS FOR FULLY NONLINEAR PARABOLIC EQUATIONS

Sharp Upper Bound on the Blow up Rate for critical nonlinear Schrödinger Equation

HARDY INEQUALITIES WITH BOUNDARY TERMS. x 2 dx u 2 dx. (1.2) u 2 = u 2 dx.

Transcription:

The blow-up rate for the critical semilinear wave equation Hatem Zaag CNRS École Normale Supérieure joint work with Frank Merle Insitut Universitaire de France Université de Cergy-Pontoise

utt = u + u p 1 u, u(0) = u0 et ut(0) = u1, where u(t) : x IR N u(x, t) IR, u0 H 1 loc,u (IRN ) and u1 L 2 loc,u (IRN ). v L 2 loc,u (IR N ) = sup a IR N ( x a <1 v(x) 2 dx ) 1/2. N 2 and p = pc 1 + 4 N 1. Earlier work: 1 < p < pc (Amer. J. Math.). 1

Critical why? 1- When p = pc, there is a conformal invariance in the equation: if U(ξ, τ) is defined by U(ξ, τ) = ( x 2 t 2 ) N 1 2 u(x, t), ξ = x x 2 t 2, τ = t x 2 t 2, then U satisfies the same equation as u. 2- The subritical case 1 < p < pc has been solved in an earlier work (Amer. J. Math.), where major difficulties to adapt to the ciritical case appeared. The presentation is done for 1 < p pc. 2

CAUCHY PROBLEM IN H 1 loc,u (IRN ) L 2 loc,u (IRN ) Since p < N+2 N 2, it follows from : - the solution of the Cauchy problem in H 1 L 2 (IR N ) (Lindblad and Sogge, Shatah and Struwe) - the finite speed of propagation. 3

FINITE TIME BLOW-UP SOLUTIONS Existence: John, Caffarelli and Friedman, Alinhac, Kichenassamy and Litman QUESTION (was open before this work) Evaluate the norm of u, u and ut in L 2 loc,u (IRN ) near the blow-up time T. 4

A HINT : THE ASSOCIATED ODE utt = u p, u(t ) = + gives u(t) κ(t t) 2 p 1 where κ = κ(p) is explicitly given. 5

SELF-SIMILAR VARIABLES wa(y, s) = (T t) 2 p 1u(x, t), y = x a T t, s = log(t t). Equivalent problem: For all y IR N and s log T : 2 s w + p + 3 p 1 sw + 2y. sw + i,j (yiyj δi,j) 2 yiyj w + 2(p + 1) p 1 y. w = w p 1 w 2(p + 1) (p 1) 2 w, 6

Equivalent problem in divergence form: For all y IR N and s log T : 2 s w 1 ρ div [ρ w ρ(y. w)y] + 2(p+1) (p 1) 2 w w p 1 w = p+3 p 1 sw 2y. sw where ρ(y) = (1 y 2 ) α If p < pc 1 + 4 N 1, then α 2 p 1 N 1 2 > 0. If p = pc, then α = 0 and ρ 1. 7

Th. (p pc) For any sol. u blowing up at time T, for all s log T + 1 and a IR N, wa(s) H 1 (B) + swa(s) L 2 (B) K where B = B(0, 1), K = K(N, p, u0, T ). Rk. From scaling arguments and the solution of the Cauchy Problem in H 1 L 2 (IR N ), we get for all s log T + 1, sup wa(s) H a IR N 1 (B) + swa(s) L 2 (B) ɛ0(n, p) > 0. = We are at the good scale 8

IN THE ORIGINAL VARIABLES Th. (p pc) For any sol. u blowing-up at time T, for any t [T (1 e 1 ), T ), u L 2 loc,u (IR N ) K(T t) 2 p 1 ut L 2 loc,u (IR N ) + u L 2 K(T t) p 1 2 1 loc,u (IRN ) where K = K(N, p, u0, T ). 9

THE ARGUMENTS OF THE PROOF - Existence of a Lyapunov functional for the equation on w and energy-type estimates. - Interpolation to gain more regularity. - Gagliardo-Nirenberg type estimates. 10

Equivalent problem in divergence form: For all y IR N and s log T : 2 s w 1 ρ div [ρ w ρ(y. w)y] + 2(p+1) (p 1) 2 w w p 1 w = p+3 p 1 sw 2y. sw where ρ(y) = (1 y 2 ) α If p < pc 1 + 4 N 1, then α 2 p 1 N 1 2 > 0. If p = pc, then α = 0 and ρ 1. 11

A LYAPUNOV FUNCTIONAL Antonini-Merle E(w) = B 1 2 ( sw) 2 (p + 1) + (p 1) 2w2 1 p + 1 w p+1 ρdy + 1 2 B ( w 2 (y. w) 2) ρdy ρ(y) = (1 y 2 ) α with α = 2 If p < pc, then α > 0. If p = pc, then α = 0 and ρ 1. If p > pc, then E is not even defined. Hence, pc is critical. p 1 N 1 2 0. 12

Lemma 1 (Monotonicity) For all s1 and s2: (p < pc, Antonini-Merle), E(w(s2)) E(w(s1)) = 2α s2 s1 B ( sw) 2 (1 y 2 ) α 1 dyds. (p = pc: degeneracy), E(w(s2)) E(w(s1)) = s2 s1 B ( sw(σ, s)) 2 dσds. Rk. 2α(1 y 2 ) α 1 δ B as p pc. 13

Lemma 2 (Blow-up criterion (Antonini-Merle)) If a solution W satisfies E(W (s0)) < 0 for some s0 IR, then W blows up in finite time. BOUNDS ON E For all s log T, s2 s1 log T 0 E(w(s)) E(w( log T )) C0 where C0 = C0( u0, T ). 14

BOUNDS ON THE DISSIPATION OF E For all s log T, s2 s1 log T (p < pc, supported in a cylinder), s 2 s1 B ( sw) 2 (y, s)(1 y 2 ) α 1 dyds C 0 2α, (p = pc, supported in the boundary of the cylinder), s 2 s1 B ( sw(σ, s)) 2 dσds C0. This degeneracy is a major difficulty in adapting the subcritical case to the critical. 15

(p < pc) SUPPORT OF THE DISSIPATION IN THE (x, t) VAR. (remember w = wa): In the interior of the light cone with vertex (a, T ). t T (a,t) t_2 t_1 x a 16

(p = pc) SUPPORT OF THE DISSIPATION IN THE (x, t) VAR. (remember w = wa): On the EDGE of the light cone with vertex (a, T ). t T (a,t) t_2 t_1 x a 17

Since all is uniform in a, we move the a, and then integrate in a... t T (a,t) t_2 t_1 x a We recover an estimate in the interior of the light cone, between t1 and t2 (like the subcritical case). 18

More precisely (p = pc), Proposition 1 For all a IR N and s2 s1 log T, s 2 s1 B swa(y, s) 2 dyds C0. 19

Even better: NON CONCENTRATION OF ( sw) 2 Proposition 2 For any ball B(b, r0) B(0, 3) with r0 < 1, s 2 s1 B(b,r0) swa(y, s) 2 dyds C0r0. 20

With this adaptation, we concentrate (from now on) on the subcritical case. Recall Bounds on E and its dissipation: For all s log T, s2 s1 log T 0 E(w(s)) E(w( log T )) C0, s 2 s1 B ( sw) 2 (y, s)(1 y 2 ) α 1 dyds C 0 2α, where C0 = C0( u0, T ). 21

GOAL Prove that w, w and sw are bounded in L 2 (B(0, 1)). Since they appear in E (with a weight): E(w) = B 1 2 ( sw) 2 (p + 1) + (p 1) 2w2 1 p + 1 w p+1 ρdy + 1 2 B ( w 2 (y. w) 2) ρdy, it is enough to bound w p+1 ρ. Rk. We get rid of the weights through a covering argument. 22

Rk. Since we have an average (in time) estimate on sw, s 2 s1 B ( sw) 2 (y, s)(1 y 2 ) α 1 dyds C0, we will look for average (in time) estimates of the terms in the functional E. 23

CONTROL OF SPACE-TIME INTEGRALS. First, we integrate E(w) between s1 and s2 : s 2 s1 E(w(s))ds = s2 s1 B 1 2 ( sw) 2 (p + 1) + (p 1) 2w2 1 p + 1 w p+1 ρdy + 1 2 s 2 s1 B ( w 2 (y. w) 2) ρdy Remark that w p+1 controls all the other terms in E. 24

2nd IDENTITY We multiply the w-equation by wρ, integrate on B (s1, s2), IBP and use the definition of E to write : (p 1) 2(p + 1) s 2 s1 B w p+1 ρdy = s2 s1 E(w(s))ds + s2 s1 B ( ( sw) 2 ρ swy. wρ swwy. ρ ) dyds + 1 2 B w sw + p + 3 2(p 1) N w 2 ρdy s2 s1. 25

Proposition 3 (p pc) For all a IR N and s log T + 1, s+1 s B w p+1 ρdyds C(C0, N, p, T ). Rk. There is a weight... PROOF (p < pc) : We will control all terms on the RHS of the previous identity by C ɛ + Cɛ s2 s1 B w p+1 ρdyds and then take ɛ small. 26

Back to p pc Corollary 1 For all a IR N and s log T + 1, s+1 s B 1/2 ( ( swa) 2 + wa 2 + wa p+1 + wa 2) dyds C where B 1/2 B(0, 1/2), C = C(N, p, u0, T ). Rk. If p < pc, we first get estimates on B B(0, 1) with the weight ρ. If p = pc, we directly get estimates on B. 27

We are ready for the proof of the Thoerem that I recall here : Th. (p pc) For any sol. u blowing up at time T, for all s log T + 1 and a IR N, B ( ( swa) 2 + wa 2 + wa 2) dy K where B = B(0, 1), K = K(N, p, u0, T ). 28

Step 1 : (p pc) Control of B w a(y, s) 2 dy We start from s+1 s B 1/2 ( ( swa) 2 + wa 2) dyds C. We first get rid of ds, and then extend the integration in space to B. Let g(s) = wa(y, s) 2 dy 12. B We write 1/2 29

g L 2 (s,s+1) = ( s+1 sg 2 L 2 (s,s+1) = s+1 s s ds B w2 dyds )1 2 C B 1/2 w swdy 4 B 1/2 w 2 dy 2 1 4 s 2 s1 ds B 1/2 ( sw) 2 dy C0. Hence, g H 1 (s, s + 1). From the Sobolev injection in one dimension, g L (s, s + 1), i.e. B 1/2 wa 2 dy C. 30

Now, we extend the integration to B. We take a = 0. Since b IR N, y < 1 2 w b (y, s) 2 dy C w b (y, s) = w0(y + be s, s), then (z = y + be s ) b IR N, z be s < 1 2 w0(z, s) 2 dz C + covering, this yields z <1 w 0(z, s) 2 dz C. 31

Step 2 : (p pc) Control of wa(s) in L r loc Proposition 4 For all s log T + 1 and a IR N, B w a(y, s) p+3 2 dy C where B = B(0, 1). Proof : Follows from w 2 and w p+1 by interpolation (H 1 L in one dimension). 32

Step 3 : (p pc) Control of the gradient in L 2 loc,u Lemma 3 For all s log T + 1 et a IR N, B w a p+1 C B w a p+3 γ ( 2 dy B w a 2 dy ) β, where γ(p, N) > 0 and if p < pc, β = β(p, N) [0, 1) if p = pc, β = 1. Proof : Gagliardo-Nirenberg. 33

Proposition 5 For all s log T + 1 and a IR N, B w a(y, s) 2 dy C. Formal proof : If all the weights were equal to 1, we would have from the functional E : B w a 2 dy C + B w a p+1 dy + Gagliardo Nirenberg B w a p+1 C B w a p+3 γ ( 2 dy B w a 2 dy ) β. If p < pc, then β < 1 and we get the conclusion. If p = pc, then β = 1. we can conclude only if B w a p+3 2 dy is small. 34

This is possible if we replace B by B(b, r0) for small r0. Indeed, remember the NON CONCENTRATION result: Proposition 6 For any ball B(b, r0) B(0, 3) with r0 < 1, s 2 s1 B(b,r0) swa(y, s) 2 dyds C0r0. Corollary 2 B(b,r0) w a(y, s) p+3 2 dy C0 r 0. 35

We then cover the unit ball B by balls of radius r0 with overlapping less than some C(N). 36