Doubling property for the Laplacian and its applications (Course Chengdu 2007)

Similar documents
Measure Estimates of Nodal Sets of Polyharmonic Functions

15 Solving the Laplace equation by Fourier method

Math 124B February 02, 2012

SOME SOLVABILITY THEOREMS FOR NONLINEAR EQUATIONS

Journal of Inequalities in Pure and Applied Mathematics

MATH 417 Homework 3 Instructor: D. Cabrera Due June 30. sin θ v x = v r cos θ v θ r. (b) Then use the Cauchy-Riemann equations in polar coordinates

Numerical approximation to ζ(2n+1)

Asymptotically Lacunary Statistical Equivalent Sequence Spaces Defined by Ideal Convergence and an Orlicz Function

A NOTE ON VERY WEAK SOLUTIONS FOR A CLASS OF NONLINEAR ELLIPTIC EQUATIONS

Compactly Supported Radial Basis Functions

A STABILITY RESULT FOR p-harmonic SYSTEMS WITH DISCONTINUOUS COEFFICIENTS. Bianca Stroffolini. 0. Introduction

Brief summary of functional analysis APPM 5440 Fall 2014 Applied Analysis

Exceptional regular singular points of second-order ODEs. 1. Solving second-order ODEs

GROWTH ESTIMATES THROUGH SCALING FOR QUASILINEAR PARTIAL DIFFERENTIAL EQUATIONS

SOME GENERAL NUMERICAL RADIUS INEQUALITIES FOR THE OFF-DIAGONAL PARTS OF 2 2 OPERATOR MATRICES

PROBLEM SET #1 SOLUTIONS by Robert A. DiStasio Jr.

Green s Identities and Green s Functions

STUDY OF SOLUTIONS OF LOGARITHMIC ORDER TO HIGHER ORDER LINEAR DIFFERENTIAL-DIFFERENCE EQUATIONS WITH COEFFICIENTS HAVING THE SAME LOGARITHMIC ORDER

ESSENTIAL NORM OF AN INTEGRAL-TYPE OPERATOR ON THE UNIT BALL. Juntao Du and Xiangling Zhu

RADIAL POSITIVE SOLUTIONS FOR A NONPOSITONE PROBLEM IN AN ANNULUS

KOEBE DOMAINS FOR THE CLASSES OF FUNCTIONS WITH RANGES INCLUDED IN GIVEN SETS

-Δ u = λ u. u(x,y) = u 1. (x) u 2. (y) u(r,θ) = R(r) Θ(θ) Δu = 2 u + 2 u. r = x 2 + y 2. tan(θ) = y/x. r cos(θ) = cos(θ) r.

On absence of solutions of a semi-linear elliptic equation with biharmonic operator in the exterior of a ball

f(k) e p 2 (k) e iax 2 (k a) r 2 e a x a a 2 + k 2 e a2 x 1 2 H(x) ik p (k) 4 r 3 cos Y 2 = 4

On the integration of the equations of hydrodynamics

d 2 x 0a d d =0. Relative to an arbitrary (accelerating frame) specified by x a = x a (x 0b ), the latter becomes: d 2 x a d 2 + a dx b dx c

A THREE CRITICAL POINTS THEOREM AND ITS APPLICATIONS TO THE ORDINARY DIRICHLET PROBLEM

3.1 Random variables

A generalization of the Bernstein polynomials

< 1. max x B(0,1) f. ν ds(y) Use Poisson s formula for the ball to prove. (r x ) x y n ds(y) (x B0 (0, r)). 1 nα(n)r n 1

THE LAPLACE EQUATION. The Laplace (or potential) equation is the equation. u = 0. = 2 x 2. x y 2 in R 2

Vanishing lines in generalized Adams spectral sequences are generic

Appendix A. Appendices. A.1 ɛ ijk and cross products. Vector Operations: δ ij and ɛ ijk

MASSACHUSETTS INSTITUTE OF TECHNOLOGY Physics Department Physics 8.07: Electromagnetism II September 15, 2012 Prof. Alan Guth PROBLEM SET 2

ON INDEPENDENT SETS IN PURELY ATOMIC PROBABILITY SPACES WITH GEOMETRIC DISTRIBUTION. 1. Introduction. 1 r r. r k for every set E A, E \ {0},

Mean Curvature and Shape Operator of Slant Immersions in a Sasakian Space Form


arxiv: v1 [math.na] 8 Feb 2013

Numerical Integration

A CHARACTERIZATION ON BREAKDOWN OF SMOOTH SPHERICALLY SYMMETRIC SOLUTIONS OF THE ISENTROPIC SYSTEM OF COMPRESSIBLE NAVIER STOKES EQUATIONS

ON THE INVERSE SIGNED TOTAL DOMINATION NUMBER IN GRAPHS. D.A. Mojdeh and B. Samadi

=0, (x, y) Ω (10.1) Depending on the nature of these boundary conditions, forced, natural or mixed type, the elliptic problems are classified as

FRACTIONAL HERMITE-HADAMARD TYPE INEQUALITIES FOR FUNCTIONS WHOSE SECOND DERIVATIVE ARE

On a quantity that is analogous to potential and a theorem that relates to it

An Exact Solution of Navier Stokes Equation

FOURIER-BESSEL SERIES AND BOUNDARY VALUE PROBLEMS IN CYLINDRICAL COORDINATES

Goodness-of-fit for composite hypotheses.

Research Article On Alzer and Qiu s Conjecture for Complete Elliptic Integral and Inverse Hyperbolic Tangent Function

f h = u, h g = v, we have u + v = f g. So, we wish

arxiv: v1 [math.nt] 12 May 2017

PHYS 301 HOMEWORK #10 (Optional HW)

EM Boundary Value Problems

A Survey of Azimuthal Angle and Eigenvalues of the Laplace Equation

JANOWSKI STARLIKE LOG-HARMONIC UNIVALENT FUNCTIONS

Math 301: The Erdős-Stone-Simonovitz Theorem and Extremal Numbers for Bipartite Graphs

In the previous section we considered problems where the

A Lower Bound for the First Steklov Eigenvalue on a Domain

BOUNDARY REGULARITY FOR THE POISSON EQUATION IN REIFENBERG-FLAT DOMAINS. Antoine Lemenant. Yannick Sire

ENGI 4430 Non-Cartesian Coordinates Page xi Fy j Fzk from Cartesian coordinates z to another orthonormal coordinate system u, v, ˆ i ˆ ˆi

FREE TRANSVERSE VIBRATIONS OF NON-UNIFORM BEAMS

What Form of Gravitation Ensures Weakened Kepler s Third Law?

x x2 2 B A ) v(0, t) = 0 and v(l, t) = 0. L 2. This is a familiar heat equation initial/boundary-value problem and has solution

Boundedness for Marcinkiewicz integrals associated with Schrödinger operators

Question Bank. Section A. is skew-hermitian matrix. is diagonalizable. (, ) , Evaluate (, ) 12 about = 1 and = Find, if

BEST CONSTANTS FOR UNCENTERED MAXIMAL FUNCTIONS. Loukas Grafakos and Stephen Montgomery-Smith University of Missouri, Columbia

Problem Set #10 Math 471 Real Analysis Assignment: Chapter 8 #2, 3, 6, 8

ON LACUNARY INVARIANT SEQUENCE SPACES DEFINED BY A SEQUENCE OF MODULUS FUNCTIONS

MATH 220: SECOND ORDER CONSTANT COEFFICIENT PDE. We consider second order constant coefficient scalar linear PDEs on R n. These have the form

Functions Defined on Fuzzy Real Numbers According to Zadeh s Extension

On the global uniform asymptotic stability of time-varying dynamical systems

Method for Approximating Irrational Numbers

Regularity for Fully Nonlinear Elliptic Equations with Neumann Boundary Data

CENTRAL INDEX BASED SOME COMPARATIVE GROWTH ANALYSIS OF COMPOSITE ENTIRE FUNCTIONS FROM THE VIEW POINT OF L -ORDER. Tanmay Biswas

Surveillance Points in High Dimensional Spaces

(n 1)n(n + 1)(n + 2) + 1 = (n 1)(n + 2)n(n + 1) + 1 = ( (n 2 + n 1) 1 )( (n 2 + n 1) + 1 ) + 1 = (n 2 + n 1) 2.

8 Separation of Variables in Other Coordinate Systems

Quantitative, uniqueness, and vortex degree estimates for solutions of the Ginzburg-Landau equation

I. CONSTRUCTION OF THE GREEN S FUNCTION

A PROOF OF THE INF-SUP CONDITION FOR THE STOKES EQUATIONS ON LIPSCHITZ DOMAINS

Lacunary I-Convergent Sequences

( ) [ ] [ ] [ ] δf φ = F φ+δφ F. xdx.

As is natural, our Aerospace Structures will be described in a Euclidean three-dimensional space R 3.

Multiple Criteria Secretary Problem: A New Approach

In many engineering and other applications, the. variable) will often depend on several other quantities (independent variables).

Analytical solutions to the Navier Stokes equations

On the Poisson Approximation to the Negative Hypergeometric Distribution

is the instantaneous position vector of any grid point or fluid

A Multivariate Normal Law for Turing s Formulae

Do not turn over until you are told to do so by the Invigilator.

COORDINATE TRANSFORMATIONS - THE JACOBIAN DETERMINANT

Euclidean Figures and Solids without Incircles or Inspheres

HIGHER REGULARITY OF THE FREE BOUNDARY IN THE OBSTACLE PROBLEM FOR THE FRACTIONAL LAPLACIAN

Geometry of the homogeneous and isotropic spaces

Results on the Commutative Neutrix Convolution Product Involving the Logarithmic Integral li(

PES 3950/PHYS 6950: Homework Assignment 6

Math 2263 Solutions for Spring 2003 Final Exam

7 Wave Equation in Higher Dimensions

Physics 121 Hour Exam #5 Solution

Perturbation to Symmetries and Adiabatic Invariants of Nonholonomic Dynamical System of Relative Motion

Solution to HW 3, Ma 1a Fall 2016

Transcription:

Doubling popety fo the Laplacian and its applications Couse Chengdu 007) K.-D. PHUNG The oiginal appoach of N. Gaofalo and F.H. Lin Fo simplicity, we epoduce the poof of N. Gaofalo and F.H. Lin in the simplest case of the Laplacian. Let B yo,r = { y R N+, y y o < R }, with y o R N+ and R > 0. The unit sphee is defined by S N = B 0,. Theoem 4..- Let D R N+, N, be a connected bounded open set such that B 0, D. If v = v y) H D) is a solution of y v = 0 in D, then fo any 0 < M /, we have v y) dy N+ B exp ln 4) 0, v y) ) dy B 0,M B 0, v y) v y) dy. dσ B 0,M Poof.- We will use the following thee fomulas. Let R > 0, R f y) dy = f s) N ddσ s), 4.) B 0,R 0 S N d f y) dy = f Rs) R N dσ s) = f y) dσ y), 4.) dr B 0,R S N B 0,R B 0,R f y i y) dy = S N f Rs) s i R N dσ s). 4.3) The identity 4.) is the fomula of change of vaiable in spheical coodinate. 4.) comes fom 4.) when f is a continuous function. The identity 4.3), available when f is integable, taduces the Geen fomula. Indeed, B 0,R f y i y) dy = B 0,R f y) ν i y) dσ y) = S N f Rs) s i R N dσ s) because on B 0,R, ν x) = x R. Now, we will intoduce the following quantities. Let > 0, H ) = B 0, v y) dσ y), D ) = B 0, v y) dy, 4.4)

and N ) = D ) H ). 4.5) The goal is to show that N is a non-deceasing function fo 0 <. To this end, we will compute the deivatives of H and D. The computation of the deivative of H ) = S N v s) N dσ s) gives H ) = N N v s) dσ s) + v s) v) S N S N s s N dσ s) = N H ) + ) S v N s sn dσ s) = N H ) + div v ) N ddσ s) 0 S N = N H ) + B 0, v ) dy, but v ) = v v + v. Since y v = 0, one finds H ) H ) = N + B 0, v y) dy H ) Next, one emaks that indeed, B 0, yi v yi vdy Consequently, 4.6) becomes fom 4.7) H ) H ) = N +. 4.6) v y) y dy = v y) v y) dσ y), 4.7) B 0, B 0, y = B 0, yi [v yi v] dy B 0, v y i vdy = B 0, [v yi v] y i y dσ y) because yv = 0 and on B 0,R, ν y) = y y. B 0, v y) v y) H ) y y dσ y) On anothe hand, the deivative of D ) = v) ρs 0 S ρ N dρdσ s) is N. 4.8) D ) = v) s S N dσ s) N = v) s S s s N dσ s) N = ) div 0 S v) s s N ddσ s) N ) = B 0, div v y dy, ) but div v y = v divy+ v ) y, with divy = N+. It emains to compute B 0, v ) ydy, when v = v y) H D). One have yj B 0, yi v ) ) y i dy = B 0, yj v y iy j vy i dy = [ ] B 0, yj yj v yi vy i dy B 0, y j v yi vy i dy B 0, yj v yi v yj y i dy = ) ) B 0, yj v yj y yi v y i y dσ y) B 0, yi v dy because y v = 0 and on B 0,R, ν y) = y y = y R. Consequently, D ) = N D ) + v B 0, y) y y dσ y), that is v D ) B D ) = N ) + 0, y) y y dσ y). 4.9) D )

Finally, the computation of the deivative of N ) = D) H) gives [ ] N ) = N ) + D ) D ) H ) H ) and one conclude fom 4.4), 4.8), 4.9) and 4.0) that y v N ) = N ) B0, y dy) ) B v 0, dy D ) H ), 4.0) v v y B0, ) y dy. Thanks to the Cauchy-Schwaz inequality, one deduce that N ) 0 i.e., N is non-deceasing on ]0, ]. Theefoe, fo, N ) N ) that is D) H) N ). Thus, fom 4.6) and 4.5), one have Consequently,, H ) H ) N N ). )) d H ) ln d N N ) d d ln ) By integating 4.) between R > 0 and R, one finds ) H R) ln H R) N N ) ln ), that is 0 < R /, S N v Rs) R) N dσ s) N e N ) ln 4. 4.) S N v Rs) R N dσ s). One conclude that fo any M /, B 0,M v y) dy = M v s) N ddσ s) 0 S N = M v Rs) R) N drdσ s) 0 S N N+ e N ) ln 4 M v Rs) R N drdσ s) 0 S N N+ e N ) ln 4 B 0,M v y) dy. Comment.- The above computations can be genealized to an elliptic opeato of second ode see [ GaL], [ Ku]). The impoved appoach of I. Kukavica It seems moe natual to conside the monotonicity popeties of the fequency function defined by B 0, v y) y ) dy B 0, v y) dy instead of B 0, v y) dy B 0, v y) dσ y). Following the ideas of Kukavica [ Ku], [ KN], see also [ AE]), one obtains the following thee lemmas. 3

. Monotonicity fomula We pesent the following lemmas. Lemma A.- Let D R N+, N, be a connected bounded open set such that B yo,r o D with y o D and R o > 0. If v = v y) H D) is a solution of y v = 0 in D, then Φ ) = B yo, v y) y y o ) dy B y o, v y) dy is non-deceasing on 0 < < R o, and d d ln v y) dy = N + + Φ )). B y o, Lemma B.- Let D R N+, N, be a connected bounded open set such that B yo,r o D with y o D and R o > 0. Let,, 3 be thee eal numbes such that 0 < < < 3 < R o. If v = v y) H D) is a solution of y v = 0 in D, then whee α = ln B y o, v y) dy ln + ln 3 ) ]0, [. B y o, v y) dy ) α B y o, 3 v y) dy ) α, The above two esults ae still available when we ae closed to a pat Γ of the bounday unde the homogeneous Diichlet bounday condition on Γ. Lemma C.- Let D R N+, N, be a connected bounded open set with bounday D. Let Γ be a non-empty Lipschitz open subset of D. Let o,,, 3, R o be five eal numbes such that 0 < < o < < 3 < R o. Suppose that y o D satisfies the following thee conditions: i). B yo, D is sta-shaped with espect to y o ]0, R o [, ii). B yo, D ]0, o [, iii). B yo, D Γ [ o, R o [. If v = v y) H D) is a solution of y v = 0 in D and v = 0 on Γ, then whee α = ln B yo, D ln + ln 3 ) α α v y) dy v y) dy v y) dy), B yo, B yo,3 D ) ]0, [.. Poof of Lemma B Let By applying Lemma A, we know that H ) = v y) dy. B yo, d d lnh ) = N + + Φ )). 4

Next, fom the monotonicity popety of Φ, one deduces that the following two inequalities ) ln H) H ) = N++Φ) d Consequently, ln ln ) H3 ) H ) and theefoe the desied estimate holds whee α = ln ln + ln 3 N + + Φ )) ln, = 3 N++Φ) d N + + Φ )) ln 3. ) ) H) H ) ln H3) H ) N + ) + Φ ln ), ln 3 ). H ) H )) α H 3 )) α,.3 Poof of Lemma A We intoduce the following two functions H and D fo 0 < < R o : H ) = B y o, v y) dy, D ) = B y o, v y) y y o ) dy. Fist, the deivative of H ) = v ρs + y 0 S N o ) ρ N dρdσ s) is H ) = B v y) y dσ y). Next, o, ecall the Geen fomula v ν Gdσ y) ν v ) Gdσ y) = v Gdy v ) Gdy. B y o, B y o, B y o, B y o, We apply it with G y) = y y o whee G Byo, = 0, ν G Byo, =, and G = N + ). It gives H ) = B N + ) v y dy + o, B yo, v ) y y o ) dy = N+ H ) + B yo, div v v) y y o ) dy ) = N+ H ) + v + v v y y o ) dy. Consequently, when y v = 0, that is H ) H) = N+ + D) H) B yo, H ) = N + H ) + D ), the second equality in Lemma A. A.) Now, we compute the deivative of D ) = v) ρs+yo 0 S ρ ) ρ N dρdσ s): N D ) = d d ) v) ρs+yo 0 S ρ N dρdσ s) N = v) ρs+yo 0 S ρ N dρdσ s) N = B yo, v dy. S N v) s+yo N dσ s) A.) 5

Hee, we emak that B v y dy = N+ o, D ) + 4 B y y y o) v dy o, B yo, v y y o ) v y y o ) dy, indeed, A.3) N + ) B v y y y o ) dy o, = B yo, div v y y o ) ) y y o ) dy B yo, v y y o )) y y o ) dy = B y y i v y y o )) y i y oi ) dy because on B yo,, = y y o o, = B v y y i v y y o ) y i y oi ) dy o, B v y y i y oi )) y i y oi ) dy o, = B v y y i v y y o ) y i y oi ) dy + o, B v y y y o dy, o, Theefoe, and B yo, yj v y i y j v = B y o, y j y i y oi ) yj v yi v + B y o, y j y i y oi ) yj v yi v + B yo, y i y oi ) y j v yi v y y o ) y i y oi ) dy y y o )) dy y y o ) dy y y o ) dy + B yo, y i y oi ) yj v yi v yj y y o ) dy = 0 because on B yo,, = y y o + B yo, v y y o ) dy + B y y y o) v v y y o ) dy o, B y o, y y o) v dy. N + ) B v y y y o ) dy = o, B v y dy 4 o, B y y y o) v dy o, + B y y y o) v v y y o ) dy, o, and this is the desied estimate A.3). Consequently, fom A.) and A.3), we obtain, when y v = 0, the following fomula D ) = N + D ) + 4 y y o ) v dy. B yo, A.4) The computation of the deivative of Φ ) = D) H) gives Φ ) = H ) [D ) H ) D ) H )], which implies using A.) and A.4) that [ ] H ) Φ N+ ) = D ) + 4 B y y y o) v dy H ) [ N+ o, H ) + D )] D ) 4 ) B yo, y y o ) v dyh ) D ) 0, = indeed, thanks to an integation by pats and using Cauchy-Schwaz inequality, we have D ) = 4 B v v y y y o) dy o, ) 4 B y y yo, o) v dy) B v yo, dy ) 4 B y y yo, o) v dy H ). ) 6

Theefoe, we have poved the desied monotonicity fo Φ and this completes the poof of Lemma A..4 Poof of Lemma C Unde the assumption B yo, D Γ fo any [ o, R o ), we extend v by zeo in B yo,r o \D and denote by v its extension. Since v = 0 on Γ, we have v = v D in B yo,r o, v = 0 on B yo,r o D, v = v D in B yo,r o. Now, we denote = B yo, D, when 0 < < R o. In paticula, = B yo,, when 0 < < o. We intoduce the following thee functions: and H ) = v y) dy, D ) = v y) y y o ) dy, Φ ) = D ) H ) 0. Ou goal is to show that Φ is a non-deceasing function. Indeed, we will pove that the following equality holds d d lnh ) = N + ) d d ln + Φ ). C.) Theefoe, fom the monotonicity of Φ, we will deduce that in a simila way than in the poof of Lemma A) that ) ) ln H) H ) ln H3) H ) ln N + ) + Φ ) ln, 3 and this will imply the desied estimate whee α = ln v y) dy ln + ln 3 ). B y o, v y) dy ) α 3 v y) dy ) α, Fist, we compute the deivative of H ) = B yo, v y) dy = v ρs + y 0 S N o ) ρ N dρdσ s). H ) = v s + y S N o ) N dσ s) = v s + y S N o ) s s N dσ s) ) = B yo, div v y) y y o ) dy ) = B y N + ) v y) + v y) y y o ) dy o, = N+ H ) + v y) v y) y y o ) dy. C.) Next, when y v = 0 in D and v Γ = 0, we emak that D ) = v y) v y) y y o ) dy, C.3) 7

indeed, v y y o ) dy = div v v y y o )) dy vdiv v = v v y y o ) dy v v because on B yo,, = y y o and v Γ = 0 = v v y y o ) dy because y v = 0 in D. Consequently, fom C.) and C.3), we obtain y y o )) dy y y o ) dy H ) = N + H ) + D ), C.4) and this is C.). On anothe hand, the deivative of D ) = v) ρs+yo 0 S ρ ) ρ N dρdσ s) is N D ) = v) ρs+yo 0 S ρ N dρdσ s) N = v y) dy. C.5) Hee, when y v = 0 in D and v Γ = 0, we will emak that v y) dy = N+ D ) + 4 B y y y o) v y) dy o, + Γ B y νv y y o ) y y o ) νdσ y), o, C.6) indeed, N + ) v y y o ) dy = div v y y o ) ) y y o ) dy v y y o )) y y o ) dy = Γ B yo, v y y o ) y y o ) νdσ y) yi v y y o )) y i y oi ) dy = Γ B yo, v y y o ) y y o ) νdσ y) v yi v y y o ) y i y oi ) dy + v y y o dy, and yj v y i y j v Theefoe, when y v = 0 in D, we have = yj y i y oi ) yj v yi v + yj y i y oi ) yj v yi v + y i y oi ) y j v yi v y y o ) y i y oi ) dy y y o )) dy y y o ) dy y y o ) dy + y i y oi ) yj v yi v yj y y o ) dy = Γ B ν y j y i y oi ) yj v yi v y y o )) dσ y) o, + v y y o ) dy +0 because y v = 0 in D y y o ) v dy. N + ) v y y o ) dy = Γ B v y y y o ) y y o ) νdσ y) o, Γ B y y j vν j y i y oi ) yi v) y y o ) dσ y) o, + u dy 4 y y o ) v dy. 8

By using the fact that v Γ = 0, we get v = v ν) ν on Γ and we deduce that N + ) v y y o ) dy = Γ B yo, ν v y y o ) y y o ) νdσ y) and this is C.6). + v dy 4 y y o ) v dy, Consequently, fom C.5) and C.6), when y v = 0 in D and v Γ = 0, we have D ) = N + D )+ 4 y y o ) v y) dy+ ν v y y o ) y y o ) νdσ y). Γ B yo, C.7) The computation of the deivative of Φ ) = D) H) gives Φ ) = H ) [D ) H ) D ) H )], which implies fom C.4) and C.7), that H ) Φ ) = 4 ) y y o ) v y) dy H ) D ) + Γ B yo, ν v y y o ) y y o ) νdσ y) H ). Thanks to C.3) and Cauchy-Schwaz inequality, we obtain that 0 4 y y o ) v y) dy H ) D ). the inequality 0 y y o ) ν on Γ hols when B yo, D is sta-shaped with espect to y o fo any ]0, R o [. Theefoe, we get the desied monotonicity fo Φ which completes the poof of Lemma C. 3 Quantitative unique continuation popety fo the Laplacian Let D R N+, N, be a connected bounded open set with bounday D. Let Γ be a non-empty Lipschitz open pat of D. We conside the Laplacian in D, with a homogeneous Diichlet bounday condition on Γ : y v = 0 in D, v = 0 on Γ, 4.) v = v y) H D). The goal of this section is to descibe intepolation inequalities associated to solutions v of 4.). Theoem 4..- Let ω be a non-empty open subset of D. Then, fo any D D such that D D Γ and D \Γ D ) D, thee exist C > 0 and ]0, [ such that fo any v solution of 4.), we have ) v y) dy C v y) dy v y) dy). D ω D O in a equivalent way, 9

Theoem 4.3.- Let ω be a non-empty open subset of D. Then, fo any D D such that D D Γ and D \Γ D ) D, thee exist C > 0 and ]0, [ such that fo any v solution of 4.), we have D v y) dy C ) v y) dy + ε v y) dy ε > 0. ε ω D Poof of Theoem 4..- We divide the poof into two steps. Step.- We apply Lemma B, and use a standad agument voi e.g., [ Ro]) which consists to constuct a sequence of balls chained along a cuve. Moe pecisely, we claim that fo any non-empty compact sets in D, K and K, such that meask ) > 0, thee exists ]0, [ such that fo any v = v y) H D), solution of y v = 0 in D, we have ) v y) dy v y) dy v y) dy). 4.3) K K D Indeed, let δ > 0 and q j R N fo j = 0,,, m, one can constuct a sequence of balls { } B qj,δ j=0,..,m, such that the following inclusion hold K B q0,δ K B qm,δ o fo some δ o > 0 B qj+,δ B qj,δ j = 0,.., m B qj,3δ D j = 0,.., m. Then, thanks to Lemma B, thee exist α, α, ]0, [, such that K v y) dy B q v y) dy m,δo B q m,δ v y) dy) α B q m,3δ v y) dy B qm,δ v y) dy) α D v y) dy ) α ) α B qm,δ v y) dy) α D v y) dy) α ) α D v y) dy which implies the desied inequality 4.3). B q0,δ v y) dy) D v y) dy), Step.- We apply Lemma C, and choose y o in a neighbohood of the pat Γ such that the conditions i, ii, iii, hold. Next, by an adequate patition of D, we deduce fom 4.3) that fo any D D such that D D Γ and D \Γ D ) D, thee exist C > 0 and ]0, [ such that fo any v = v y) H D) such that y v = 0 on D and v = 0 on Γ, we have This completes the poof. ) v y) dy C v y) dy v y) dy). D ω D ) α Remak.- Fom standad minimization technique, the above inequality implies D v y) dy C ) v y) dy + ε v y) dy ε > 0. ε ω D 0

Indeed, we denote A = D v y) dy 0, B = ω v y) dy and E = D v y) dy. We know that thee exist C > 0 and ]0, [ such that A CB E. Theefoe, A C B E A ). Now, if E A ε, then A C B ) ε. And, if E A > ε, then A εe. Consequently, one obtain the desied intepolation inequality. Convesely, suppose that thee exist C > 0 and ]0, [ such that D v y) dy C A then, we choose ε = E in ode to get A CB E. ) v y) dy + ε v y) dy ε > 0, ε ω D Comment.- The above computations can be genealized to solutions of the following elliptic system y u = f in D, u = 0 on Γ, u = u y) H D), y ) f H H0 D), in ode to get the following estimate D u y) dy C y ) ) f + u y) dy u y) dy). L D) ω D 4 Quantitative unique continuation popety fo the elliptic opeato t + In this section, we pesent the following esult to be compaed to [ LeR]). Theoem 4.4.- Let be a Lipschitz connected bounded open set of R N, N. We choose T > 0 and δ ]0, T/[. We conside the elliptic opeato of second ode in ]0, T [ with a homogeneous Diichlet bounday condition on 0, T ), t u + u = 0 on ]0, T [, u = 0 in ]0, T [, 4.4) u = u x, t) H ]0, T [). Then, fo any ϕ C0 0, T )), ϕ 0, thee exist C > 0 and ]0, [ such that fo any u solution of 4.4), we have T δ δ T u x, t) dxdt C 0 ) T u x, t) dxdt 0 ϕu x, t) dxdt). Poof.- We apply Theoem 4. with D = ]0, T [, ]δ, T δ[ D, y = x, t), y = t +.

5 Quantitative unique continuation popety fo the sum of eigenfunctions The goal of this section is to obtain the following esults to be compaed to [ LZ] o [ JL]). Theoem 4.5.- Let be a bounded open set in R N, N, eithe convex o C and connected. Let ω be a non-empty open subset in. Then, thee exists C > 0 such that fo any sequence {a j } j of eal numbes and any intege M >, we have a j Ce C λ M ω a j e j x) dx, whee {λ j } j and {e j } j ae the eigenvalues and eigenfunctions of in H 0 ), constituting an othonomal basis in L ). O in an equivalent way, Theoem 4.6.- Let be a bounded open set in R N, N, eithe convex o C and connected. Let ω be a non-empty open subset in. Then, thee exists C > 0 such that fo any sequence {a j } j of eal numbes and any R > λ, we have {j;λ j R} a j Ce C R ω {j;λ j R} a j e j x) dx, whee {λ j } j and {e j } j ae the eigenvalues and eigenfunctions of in H 0 ), constituting an othonomal basis in L ). Poof of Theoem 4.5.- We divide the poof into thee steps. Step.- Fo any a j R, we intoduce the solution w x, t) = λj ) a j e j x) ch t χ x) a j e j x), whee χ C 0 ω), χ = in ω ω. Recall that cht = e t + e t ) /. Theefoe, w solves t w + w = f in ]0, T [, w = 0 on ]0, T [, w = t w = 0 on ω {0}, w = w x, t) H ]0, T [), fo any T > 0, whee f = χ a j e j H 0 ). We denote by w the extension of w by zeo in ω ] T, 0[. Theefoe, w solves t w + w = f 0,T ) in ]0, T [ ω ] T, 0[, w = 0 in ω ] T, 0[, w = 0 on ]0, T [. At pesent, we define D, a connected open set in R N+, satisfying the following six conditions: i). ]δ, T δ[ D fo some δ ]0, T/[ ; ii). ]δ, T δ[ D ; iii). D ]0, T [ ω ] T, 0[ ;

iv). thee exists a non-empty open subset ω o D ω ] T o, 0[ fo some T o ]0, T [ ; v). D C if is C and connected ; vi). D is convex with an adequate choice of δ, T o ) if is convex. In paticula, w H D). Step.- We claim that thee exists g H ] T, T [) H0 ] T, T [) H D) such that { t g + g = f 0,T ) in ] T, T [, g = 0 on ] T, T [), and g L D) f L ]0,T [). 4.5) Indeed, we will poceed with six substeps when is C and connected the case whee is convex is well-known since then ] T, T [ is convex). We denote h = f 0,T ) L ] T, T [). Substep : one ecall that h L ] T, T [) implies the existence of g H 0 ] T, T [). Substep : thanks to the inteio egulaity fo elliptic systems, fo any D 0 ] T, T [, g H D 0 ). Substep 3: thank to the bounday egulaity fo elliptic systems, but not closed to the bounday { T, T }, g is also locally in H because is C. Substep 4: we extend the solution at t = T as follows. Let h x, t) = h x, t) fo x, t) ] T, T [ and h x, t) = h x, T t) fo x, t) ]T, 3T [. Thus h L ] T, 3T [). Let g x, t) = g x, t) fo x, t) [ T, T [ and g x, t) = g x, T t) fo x, t) [T, 3T ]. Thus, g solves { t g + g = h in ] T, 3T [, g = 0 on ] T, 3T [). By applying the bounday egulaity as in substep 3, one obtain that g H ]0, T [). In paticula, g H ]0, T [). Substep 5: we extend in a simila way at t = T in ode to conclude that g H ] T, 0[). Substep 6: finally, we multiply t g + g = h by ) g and integate by pats ove ] T, T [, in ode to obtain T T tg H ) dt + g L ] T,T [) which gives the desied inequality 4.5). = T 0 f x) ) g x, t) dxdt = T 0 f x) ) ) g x, t) dxdt because f H0 ) f L ]0,T [) g L ] T,T [) fom Cauchy-Schwaz. Step 3.- Finally, we apply Theoem 4.3 with y = t +, v = w g in D with Γ = ]0, T [ and ]δ, T δ[ D D such that D D Γ and D \Γ D ) D in ode that which implies that D w g dy C D w dy C ) w g dy + ε w g dy ε > 0, ε ωo D ) g dy + ε w dy ε ]0, [, ε D D whee we have used that w = 0 in ω o. Fom 4.5), we conclude that thee exist C > 0 and ]0, [ such that T δ δ w x, t) dxdt C ) T T f x) dxdt + ε w x, t) dxdt ε > 0. ε 0 0 3

On anothe hand, we have the following inequalities T 0 f x) dxdt = T 0 χ x) a j e j x) dxdt T ω χ x) a j e j x) dx, T δ δ T 0 w x, t) dxdt = T 0 a j e j x) ch λ j t ) χ x) a j e j x) dxdt T e λ M T a j + T ω χ x) a j e j x) dx, t) λj a j e j x) ch dxdt T δ δ w x, t) dxdt + T Consequently, fom the last fou inequalities, we deduce that fo any ε > 0, T δ) a j T δ δ a j e j x) ch λ j t ) dxdt C ) ε T ω χ x) a j e j x) dx +4ε T e λ M T a j + T ω χ x) a j e j x) dx +T ω χ x) a j e j x) dx. Choosing ε = 8 T δ) T e λ M T, we obtain the existence of C > 0 such that a j Ce C λ M ω a j e j x) dx. ω χ x) a j e j x) dx. 6 Application to the wave equation Fom the idea of L. Robbiano which consists to use an intepolation inequality of Hölde type fo the elliptic opeato t + and the FBI tansfom intoduced by G. Lebeau et L. Robbiano, we obtain the following estimate of logaithmic type. Theoem 4.9.- Let be a bounded open set in R N, N, eithe convex o C and connected. Let ω be a non-empty open subset in. Then, fo any β ]0, [, thee exist C > 0 and T > 0 such that fo any solution u of t u u = 0 in ]0, T [, u = 0 on ]0, T [, u, t u), 0) = u 0, u ), 4

with non-identically zeo initial data u 0, u ) H 0 ) L ), we have u 0, u ) H 0 ) L ) e C u 0,u! ) /β H 0 ) L ) u 0,u ) L ) H ) u L ω ]0,T [). 7 Application to the heat equation In this section, we popose the following esult. Theoem 4.7.- Let be a bounded open set in R N, N, eithe convex o C and connected. Let ω be a non-empty open subset in. Then, fo any T > 0, thee exists C > 0 such that fo any u solution of t u u = 0 in ]0, T [, u = 0 on ]0, T [,.3) u, 0) = u o, with non-identically zeo initial data u o H0 ), and fo any t o ]0, T [, we have 0 uo C@ u o L ) Ce to +t H 0 ) o A uo L ) u, t o ) L ω). Comment.- We also have that u o H ) Ce C uo to +t L ) o uo H )! u, t o ) L ω). Remak.- The quantitative unique continuation popety fom ω {t o } fo paabolic opeato with space-time coefficients was established by L. Escauiaza, F.J. Fenandez and S. Vessella [ EFV]). Poof of Theoem 4.7.- We decompose the poof into two steps. Fist, in step, we will pove that the solution u of.3) satisfies the following estimate ) u o L ) exp t o u o H 0 ) u o L ) u, t o ) L ). Next, in step, we will pove that the solution u of.3) satisfies the following estimate ) u x, t o ) dx C e C to u x, 0) dx u x, t o ) dx). ω Finally, the above inequalities imply the existence of C > 0 such that u o L ) C C/ u o ) H e 0 to exp t ) o u o u x, t o ) dx. L ) ω Poof of step.- Let us intoduce fo almost t [0, T ] such that the solution of.3) satisfies u, t) 0, the following quantity. Φ t) = u x, t) H 0 ) u x, t) L ). 5

We begin to check that Φ is a non-inceasing function on [0, T ]. This monotonicity popety holds because fo any initial data in a dense set of H0 ), we have that d dtφ t) 0. Indeed, fom the following two equalities { d d dt u x, t) L ) + u x, t) H o ) = 0, dt u x, t) H 0 ) + u x, t) L ) = 0, we can deduce that d dt Φ t) = u x, t) 4 L ) [ u x, t) L ) u x, t) L ) + u x, t) 4H 0 ) ]. Theefoe, we get by classical density agument and Cauchy-Schwaz inequality that fo any solution u of.3), u, t) 0 a.e., and any t [0, T ], Φ t) Φ 0). On anothe hand, we also have that d dt u x, t) L ) + Φ t) u x, t) L ) = 0, which implies that 0 d dt u x, t) L ) + Φ 0) u x, t) L ). Theefoe, by Gonwall Lemma, we get the desied estimate Poof of step.- Let λ,λ, and e,e, be the eigenvalues and eigenfunctions of in H0 ), constituting an othonomal basis in L ). Fo any u o = u, 0) = α j e j in L ) whee j α j = u oe j dx, the solution u of.3), can be witten u x, t) = α j e j x) e λjt. Let t o ]0, T [. We intoduce see [ Lin] o [ CRV]) the solution w x, t) = λj ) α j e j x) e λjto ch t χ x) α j e j x) e λjto, j j j whee χ C 0 ω), χ = in ω ω. Theefoe, w solves t w + w = f in ]0, T [, w = 0 on ]0, T [, w = t w = 0 on ω {0}, w = w x, t) H ]0, T [), whee f = χu, t o ) H 0 ). Consequently, in a simila way than in the poof of Theoem 4.5, fo any δ ]0, T/[, thee exist C > 0 and ]0, [ such that we have T δ δ w x, t) dxdt C On anothe hand, the following inequalities hold. T f x) dxdt T T 0 0 w x, t) dxdt T j ) T T f x) dxdt + ε w x, t) dxdt ε > 0 ε 0 0 ω χ x) u x, t o ) dx, α je λ jt o λ j T) + T ω χ x) u x, t o ) dx, αj e λ jt o λ j T) = α j {j ; λ j T } j e λjto λj T) + α {j ; λ to j> T } j e λjto λj T) to e T to αj, j 6

T δ δ t) λj α j e j x) e λ jt o ch dxdt j T δ δ w x, t) dxdt+t Consequently, fom the fast five inequalities, we deduce that fo any ε > 0, T δ) αj e λjto T δ) αj e λ jt o λ j δ) j j T δ δ α j e j x) e λjto ch λ j t ) dxdt T C ε +4T ε j ) Finally, thee exists C > 0 such that fo any t o > 0, u x, t o) dx = αj e λjto j C ε which implies the desied estimate, u x, t o ) dx C e C ) to ω χ x) u x, t o) dx αj + ) ω χ x) u x, t o) dx e T to j +T ω χ x) u x, t o) dx. ω χ x) u x, t o ) dx, ) ω u x, t o) dx + εe C to u x, 0) dx ε ]0, [, ) u x, 0) dx u x, t o ) dxdt). ω 8 Notes on the papes in efeence... to be completed. some efeence to papes of Alessandini have to be add too...) Refeences [ A] S. Angenent, The zeo set of a paabolic equation, J. eine angew. Math. 390 988), 79 96. [ AE] V. Adolfsson and L. Escauiaza, C,α domains and unique continuation at the bounday, Comm. Pue Appl. Math., 50, 3 997) 935-969. [ CRV] B. Canuto, E. Rosset and S. Vessella, Quantitative estimates of unique continuation fo paabolic equations and invese initial-bounday value poblems with unknows boundaies, Tans. Ame. Math. Soc. 354 00), 49 535. [ E] L. Escauiaza, Communication to E. Zuazua. [ EFV] L. Escauiaza, F.J. Fenandez and S. Vessella, Doubling popeties of caloic functions, Appl. Anal. 85, -3 006) 05-3. [ GaL] N. Gaofalo and F H. Lin, Monotonicity popeties of vaiational integals, A p -weights and unique continuation, Indiana Univ. Math. J. 35, 986) 45-68. 7

[ JL] D. Jeison and G. Lebeau, Nodal sets of sum of eigenfunctions, Hamonic analysis and patial diffeential equations Chicago, IL, 996), Univ. Chicago Pess, Illinois, 999) 3-39. [ Ku] I. Kukavica, Level sets fo the stationay Ginzbug-Landau equation, Calc. Va. 5 997), 5-5. [ Ku] I. Kukavica, Quantitative uniqueness fo second ode elliptic opeatos, Duke Math. J. 9 998) 5-40. [ KN] I. Kukavica and K. Nystöm, unique continuation on the bounday fo Dini domains, Poc. of Ame. Math. Soc. 6, 998) 44-446. [ Lin] F. H. Lin, A uniqueness theoem fo the paabolic equations, Comm. Pue Appl. Math. 43 990), 7-36. [ LeR] G. Lebeau and L. Robbiano, Contôle exacte de l équation de la chaleu, Comm. Pat. Diff. Eq. 0 995) 335-356. [ LeR] G. Lebeau and L. Robbiano, Stabilisation de l équation des ondes pa le bod, Duke Math. J. 86, 3 997) 465-49. [ LZ] G. Lebeau and E. Zuazua, Null-contollability of a system of linea themoelasticity, Ach. Mech. Anal. 4, 4 998) 97-39. [ M] L. Mille, Unique continuation estimates fo the laplacian and the heat equation on non-compact manifolds, Math. Res. Lettes, 005) 37-47. [ Ro] L. Robbiano, Théoème d unicité adapté au contôle des solutions des poblèmes hypeboliques, Comm. Pat. Diff. Eq. 6 99) 789-800. [ Ro] L. Robbiano, Fonction de coût et contôle des solutions des équations hypeboliques, Asymptotic Analysis, 0 995) 95-5. 8