INCREASING STABILITY IN THE INVERSE PROBLEM FOR THE SCHRÖDINGER EQUATION. A Dissertation by. Li Liang

Size: px
Start display at page:

Download "INCREASING STABILITY IN THE INVERSE PROBLEM FOR THE SCHRÖDINGER EQUATION. A Dissertation by. Li Liang"

Transcription

1 INCREASING STABILITY IN THE INVERSE PROBLEM FOR THE SCHRÖDINGER EQUATION A Dissertation by Li Liang Bachelor of Science, Wichita State University, 20 Master of Science, Wichita State University, 203 Submitted to the Department of Mathematics, Statistics, and Physics and the faculty of the Graduate School of Wichita State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy December 205

2 c Copyright 205 by Li Liang All Rights Reserved

3 INCREASING STABILITY IN THE INVERSE PROBLEM FOR THE SCHRÖDINGER EQUATION The following faculty members have examined the final copy of this dissertation for form and content, and recommend that it be accepted in partial fulfillment of the requirement for the degree of Doctor of Philosophy with a major in Mathematics. Victor Isakov, Committee Chair Thomas DeLillo, Committee Member Alexander Bukhgeym, Committee Member Tianshi Lu, Committee Member Hyuck M. Kwon, Committee Member Accepted for the College of Liberal Arts and Sciences Ron Matson, Dean Accepted for the Graduate School Kerry Wilks, Interim Dean iii

4 ABSTRACT The Schrödinger equation is a partial differential equation that describes how the quantum state of a physical system changes with time. It was formulated in late 925 by the Austrian physicist Erwin Schrödinger. The study of the inverse problem for the Schrödinger equation focuses on finding the potential c from the prescribed boundary condition, which is generally given as Cauchy data containing both solution on the boundary and its normal derivative, or the Dirichlet-to-Neumann operator which maps the solution on the boundary to its normal derivative. The result of research has direct application to optical tomography, which is an inverse problem of reconstructing medical images through transmission of light. More precisely, one can detect cancer by recovering the absorption and scattering coefficients in the transport equation. The paper [30] discussed the simplification of the transport equation into the Schödinger equation. Optical tomography with partial data is considered extremely valuable, since we do not have access to the full boundary in real application. The research in the dissertation assumes partial data, which can be applied to breast cancer detection. The main result of this dissertation demonstrates the increasing stability phenomenon in the inverse problem for the Schrödinger equation with partial data. We establish the theorem which contains the stability estimate bound for c. The bound decays as the energy k grows in a certain interval, and hence shows a better stability of recovering c there. In addition, we found a numerical algorithm for the linearized (simplified) inverse problem by using the Neumann-to-Dirichlet boundary map. The algorithm gives numerical evidence of increasing stability, which confirmed the theoretical prediction. The proof of uniqueness for this inverse problem was established before. The proof used almost exponential solution for the Schrödinger equation and the Fourier transform of c. A similar technique will be used in this dissertation to obtain the stability bound, but our choice of ζ in the almost exponential solutions is new. iv

5 TABLE OF CONTENTS Chapter Page INTRODUCTION NOTATION PRELIMINARY UNIQUENESS OF THE INVERSE PROBLEM THE SCHRÖDINGER EQUATION THE PROOF OF UNIQUENESS THE INCREASING STABILITY PHENOMENON THE STABILITY ESTIMATE BOUND THE ALMOST EXPONENTIAL SOLUTIONS THE PROOF OF INCREASING STABILITY THE CAUCHY BOUNDARY DATA THE LINEARIZED INVERSE PROBLEM FOR THE SCHRÖDINGER EQUA- TION NUMERICAL SIMULATION STABILITY ESTIMATE FOR PARTIAL DATA NUMERICAL SIMULATION FOR PARTIAL DATA THE CONCLUSION AND FUTURE CHALLENGES REFERENCES v

6 LIST OF FIGURES Figure Page vi

7 CHAPTER INTRODUCTION The study of inverse problems allows us to calculate parameters which can t be directly observed. In particular, inverse problems of partial differential equations are to recover coefficients from boundary measurements. The coefficients often represent important physical objects contained in a domain which we can t get in. Uniqueness and stability play important roles in these problems, since the first one implies that we have enough data to determine the coefficients, and the later one describes how errors on the measurements impact the accuracy of the recovered solution. In this Dissertation, we will show the increasing stability of recovering the coefficient c C () in the Schrödinger equation when the Dirichlet-to-Neumann map is given on part of the boundary. The proof uses almost exponential solutions of the Schrödinger equation as the primary technique for achieving the result. The idea of these solutions was first introduced by Faddeev [4], then rediscovered by Sylvester and Uhlmann [] in the proof of the global uniqueness of c in the three dimensional case. Alessandrini in the paper [5] has obtained a logarithmic stability estimate for c from the Dirichlet-to-Neumann map. Mandache [6] demonstrated the optimality of log-type stability. However, the logarithmic stability was disappointing in applications, since small errors in the data of the inverse problem result in large errors in numerical reconstruction of physical properties of the medium. Consequently, it restricts resolution in the electrical impedance tomography. Isakov proved the uniqueness of c in the Schrödinger equation with partial data in the paper [7], where he used Riemann-Lebesgue Lemma and the methods of reflection for almost exponential solutions. He also derived the increasing stability bound for c in different ranges of frequency k for full boundary data in [8], where both complexand real-valued geometrical optics solutions were used in the proof. The proof was simplified in [9] where only complex-valued geometrical optics solution were used. Similar results

8 were obtained by Isaev and Novikov [0] by less explicit and more complicated methods of scattering theory. Since the access to full boundary data can be very expensive in real life, inverse problems with local data generally have valuable applications. This thesis will show increasing stability of c in the Schrödinger equation when the Dirichlet-to-Neumann map is given on an arbitrary part of the boundary, assuming that the remaining part is contained in a plane in 3 dimension. The result has applications to situations where the part of the boundary of the domain is inaccessible. For instance, when detecting breast cancer, one can assume the domain is a half unit ball. The measurements can be made on the top of the half ball by putting sensors along the top surface. However the bottom surface is inside human body so it is impossible to take measurements from there. The result of this paper allows us to better recover potential c inside the half ball from prescribed partial data for large frequencies k. We use methods of even reflection to build almost complex exponential solutions which have vanishing Neumann data on the partial boundary contained in a plane, then give bounds on these solutions by using sharp bounds on regular fundamental solutions of some linear partial differential operators. In the proof, we encountered some new difficulties compared to [7], [8]. More precisely, this paper uses the type of almost exponential solutions introduced in [7]. The advantage is that, in contrast to [8], they will work in both cases of high and low frequencies. However the product of these solutions will produce an extra term which decays in an uncontrollable way, so in addition to [8], we use integration by parts for this extra term, then use the stability of analytic continuation to obtain the bound for c. Moreover, the dissertation contains both theoretical and numerical evidence of increasing stability phenomenon in the linearized inverse problem for the Schrödinger equation with full/partial Neumann-to-Dirichlet map. The phenomenon shows a better stability in recovering potential c as the energy frequency k grows. We will first justify the linearization by assuming that c is a small perturbation, then obtain an increasing stability bound for c with explicit constants as the theoretical evidence. In the case of partial data, the methods 2

9 of reflection and analytic continuation are also needed to derive such bound. For numerical experiments, we take measurements of Dirichlet data along the boundary surface, then compute c for different choices of frequency k to confirm the prediction of the theory. Currently, there are many researches regarding the linearized inverse conductivity problem with zero energy frequency. These works can be traced back to the 980s when Calderón [7] proposed the idea of determining electrical conductivity of a medium by making voltage and current measurements along its boundary. Inspired by his idea, Isaacson [8]-[20] made many contributions to the research in Electrical Impedance Tomography, which has great potential for the medical application. He [2] also introduced the concepts of distinguish-ability as a means of characterizing unrecoverable information in the presence of measurement errors. In the paper [22], Isaacson and Cheney have made analysis on the effects of measurements precision on the linearized inverse problem for the conductivity equation. Dobson and Santosa [23] computed conductivities by prescribing a set of dipole current patterns along the circular boundary and examined the resolution limit. Their analysis shows severe limitations on the image of the recovered solutions when k = 0. This paper will focus on the linearized Schrödinger equation, which is an alternative form of the conductivity equation, and will show that the stability can be improved if one increases the energy k. The thesis is organized as follows. In next chapter, we present readers with the inverse problem for the Schrödinger equation and give proofs for uniqueness and stability. In chapter 3, we justify the linearization for the Schödinger equation, then derive the increasing stability bound for c from the prescribed Neumann-to-Dirichlet map on the full boundary. chapter 4 contains the numerical result in the full data case, which is found by forming a system of equations, and solve the system for different values of k. Then, we obtain similar results when the partial data is given. The methods of reflection and analytic continuation are used here. Finally, in the last section, we outline future developments and challenges. 3

10 . NOTATION The notation introduced here will be used throughout the entire thesis. ˆf(ξ) = R 3 f(x)e iξ x denotes the Fourier transform of f. H s () = {f L 2 (), ( + ξ 2 ) s ˆf 2 (ξ) < + } denotes Sobolev space with order R 3 s R. In the case of s Z +, the definition is equivalent to H s () = { α s Dα f 2 < + }. H s () denotes the space of functions f H s () and supp f. f (s) () = [ R 3 ( + ξ 2 ) s ˆf 2 (ξ) ] 2 denotes the norm for Sobolev space. f p () = ( f p ) p denotes the norm for Lp space for p <. C, C(M, ), C 0, C denote the generic constants whose values change from line to line, and only depend on the indicated arguments. 4

11 .2 PRELIMINARY Theorem.2. (Green s theorem) Let v, u H 2 (), where is a Lipschitz domain, we have v u = u v + ( ν u)v, v u = v u + v ν u u ν v. Theorem.2.2 (Trace s theorem) Let be a bounded Lipschitz domain. We define a linear operator T : C ( ) L 2 ( ) as T f = f, then there exist a unique extension of T such that T : H s () H s 2 ( ), for 2 < s, and T is a bounded linear operator. Theorem.2.3 (Sobolev embedding theorem) Let H s,p () denote the Sobolev space consisting of all functions on whose first k weak derivatives are functions in L p () for p <. If k > l and p < q < are real numbers such that (k l)p < n and then q = p k l n, H s,p () H l,q () Definition An open set R n is called Lipschitz domain if for every point p, there exits a open ball B(p, r) centered at p with radius r > 0, and a map V p : B(p, r) B, where B denotes the unit ball, such that. V p is bijection, 5

12 2. V p and Vp are both Lipschitz continuous functions, 3. V p ( B(p, r)) = B 0, 4. V p ( B(p, r)) = B +, where B 0 = {x B x n = 0}, B + = {x B x n > 0}. Definition The domain R n is said to satisfy cone condition if each point x is vertex of a finite, right-spherical cone C x contained. (C x is the union of all points on line segments from x to points of a ball not containing x.) Theorem.2.4 (the Riemann-Lebesgue Lemma) If f is L integrable on R n, then lim f(x)e iξ x = 0 ξ 0 R n 6

13 CHAPTER 2 UNIQUENESS OF THE INVERSE PROBLEM In this chapter, we will introduce the inverse problem for the Schrödinger equation with partial data, then prove uniqueness by the method of reflection and Fourier transform. 2. THE SCHRÖDINGER EQUATION Let be a domain in R 3 with Lipschitz boundary. We assume { x i <, i =, 2, 3, x 3 > 0} and a part Γ of lies inside the plane {x R 3 x 3 = 0}. We consider the Schrödinger equation u k 2 u + cu = 0 in (2..) with the boundary data u = g 0 H 2 ( + ), ν u= 0 on Γ, (2..2) where + = {x 3 > 0} and the potential c C (). Suppose that k 2 σ( + c), where σ( + c) is the set of eigenvalues of Dirichlet problem (2..)(2..2) for the operator +c. By [29], σ( +c) forms a discrete set. Thus we can define the Dirichlet-to-Neumann operator which maps g 0 H 2 ( + ) into ν u H 2 ( + ) as Λ c g 0 = ν u on +. (2..3) The trace v u is understood as follows. One can take a sequence u m convergent to u in the space H (). From trace theorem, it follows that v u m ( 2 )( ) is bounded by u m () (). Since u m converges in H (), their boundary trace ν u m must converge in H 2() and the limit of ν u m is the desired trace of u. In the practical situation, one takes measurements on ν u for each prescribed function g 0. Thus, in order for Λ maps g 0 into ν u, there must exist a solution to the boundary value problem (2..)(2..2). 7

14 The inverse problem for the Schrödinger equation is to find the potential c in (2..) from the boundary conditions (2..2) and (2..3). It is known that these conditions can determine the coefficient c uniquely. The proof for this claim will be given in the next section. 2.2 THE PROOF OF UNIQUENESS The following theory establishes the uniqueness of recovering c from the Dirichlet-to- Neumann map given as (2..3). Theorem 2.2. If Λ c = Λ c2, then c = c 2. Proof: Let u, u 2 solve the boundary value problems (2..) (2..2) corresponding to c and c 2 respectively. By the application of the Green s theorem and Λ c = Λ c2, we have the orthogonality relation (c 2 c )u u 2 = 0. (2.2.) Now, suppose that ξ R 3 and ξ = (ξ, ξ 2, ξ 3 ). We introduce vectors e() = (ξ 2 + ξ 2 2) 2 (ξ, ξ 2, 0), e(3) = (0, 0, ) and e(2) to form orthonormal basis e(), e(2), e(3) in R 3. Denote the coordinate system in this basis by (x e, x 2e, x 3e ) e. Observe it preserves inner product in the sense that xy = x y + x 2 y 2 + x 3 y 3 = x e y e + x 2e y 2e + x 3e y 3e Let ζ() = ( ξ e 2 τξ 3, i( ξ 2 ( 4 + τ 2 ) k 2 ) 2, ξ τξ e) e, ζ () = ( ξ e 2 τξ 3, i( ξ 2 ( 4 + τ 2 ) k 2 ) 2, ξ 3 2 τξ e) e, ζ(2) = ( ξ e 2 + τξ 3, i( ξ 2 ( 4 + τ 2 ) k 2 ) 2, ξ 3 2 τξ e) e, ζ (2) = ( ξ e 2 + τξ 3, i( ξ 2 ( 4 + τ 2 ) k 2 ) 2, ξ τξ e) e, (2.2.2) 8

15 where ξ e = ξ 2 + ξ2 2 and τ is a positive real number. One can check that ζ() ζ() = ζ () ζ () = ζ(2) ζ(2) = ζ (2) ζ (2) = k 2. Assume c = 0 on {x 3 > 0} \, then extend c onto R 3 as even function of x 3. Denote V (x, x 2, x 3 ) = V (x, x 2, x 3 ) for a given function V. By [6], section 5.3, there are almost exponential solutions u(x; j) = e iζ(j) x ( + V (x; j)) + e iζ (j) x ( + V (x; j)), j =, 2, (2.2.3) to (2..) with c = c, c 2 respectively, and V (; j) (0) () 0, V (; j) (0) () 0 as τ. (2.2.4) Then it is obvious that ν u(x; ) = ν u(x; 2) = 0 on Γ 0 (2.2.5) Now, let c(x) = c (x) c 2 (x), by (2.2.), (2.2.3), (2.2.5), we have that 0 = cu(x; )u(x; 2) = c(x)(e i(ζ()+ζ(2))x ( + V (; ))( + V (; 2) +e i(ζ ()+ζ(2))x ( + V (; ))( + V (; 2)) +e i(ζ()+ζ (2))x ( + V (; ))( + V (; 2)) (2.2.6) +e i(ζ ()+ζ (2))x ( + V (; ))( + V (; 2)))dx. Due to (2.2.2), the above equation implies that c(x)(e iξx ( + V (x; ))( + V (x; 2)) + e iξ x ( + V (x; ))( + V (x; 2)) + e iξ e(x e 2τx 3 ) ( + V (x; ))( + V (x; 2)) (2.2.7) + e iξ e(x e +2τx 3 ) ( + V (x; ))( + V (; 2))) = 0. If we send τ, since all moduli of exponents are bounded by, then the limits of all terms containing V and V are 0 due to (2.2.4). By the Riemann-Lebesgue Lemma, limits of e iξ e(x e 2τx 3 ), e iξ e(x e +2τx 3 ) (2.2.8) 9

16 are 0 as τ provided ξ e 0. Thus from (2.2.7) we have that c(x)(e iξx + e iξ x ) = 0 (2.2.9) for ξ e 0. Since c is a compactly supported function, the left side of (2.2.9) is analytic with respect to ξ e. Thus (2.2.9) holds for all ξ R 3. Because c is an even function of x 3, c(x)(e iξx + e iξ x ) = c(x)e iξx. R 3 (2.2.0) By (2.2.7) and the definition of Fourier transform, ĉ(ξ) = c(x)e iξx = 0, R 3 (2.2.) which implies c = 0 or c = c 2 by the uniqueness of inverse Fourier transform. 0

17 CHAPTER 3 THE INCREASING STABILITY PHENOMENON The increasing stability phenomenon shows that the error for recovered solution c decreases when k grows in a certain interval. In this chapter, we will obtain some bound for c which can be viewed as an evidence of the phenomenon. The proof uses almost exponential solutions and methods of reflection. The following lemma is needed for establishing the main result. Lemma The Dirichlet-to-Neumann map defined in (2..3) is a linear bounded operator from L 2 ( + ) into L 2 ( + ). Proof: To prove Λ c Λ c2 is a linear bounded operator from L 2 ( + ) to L 2 ( + ), we will firstly show the same result for Λ c Λ 0. Λ 0 denotes the Dirichlet-to-Neumann map for Laplace equation. Throughout the proof, we will use the generic constant C which depends on. Let u 0 and u c be the solutions for the following equations. u 0 = 0 in, (3.0.) u 0 = g 0 on +, (3.0.2) and u c + c u c = 0 in, (3.0.3) u c = g 0 on +, (3.0.4) We define the operator (Λ c Λ 0 )g 0 = ν u c ν u 0 on +. Denote Γ = {x 3 = 0}. Now let us reflect functions u 0, u c, and c across Γ as even functions of x 3. Denote the domain after reflection. Assume is the support of c contained in, so. Then by the theorem , p.56 [5], u 0 (0) ( ) C g 0 (0) ( ). (3.0.5)

18 Now let u = u c u 0, then subtracting equation (3.0.3) from (3.0.) results in u + c u = c u 0 in, (3.0.6) u = 0 on. (3.0.7) By the theorem 4., p.90 [6] for elliptic equations, it follows that u () ( ) C c u 0 (0) ( ) C u 0 (0) ( ). (3.0.8) Introduce a larger domain B with the smooth boundary and a function U satisfying U = c u c u 0 in B, (3.0.9) U = 0 on B, (3.0.0) where right side of (3.0.9) equals 0 in B \. Since B is a smooth domain, U (2) (B) C( u (0) ( ) + u 0 (0) ( )) C g 0 (0) ( ), (3.0.) where the second inequality is due to (3.0.8) and (3.0.5). Let U = U u, then subtracting equation (3.0.9) from (3.0.6) gives us U = 0 in, (3.0.2) U = U on. (3.0.3) It follows that ν U (0) ( ) C U () ( ) C( U (0) ( ) + U (0) ( )), (3.0.4) where the first inequality is due to the theorem (a), p.54 [5] together with the theorem , p.56, and the second one is by the definition of sobolev norm. Now, by applying the trace theorem for the partial derivative j U, j =, 2, 3, we have j U (0) ( ) j U () ( ) U (2) ( ). By using the partition of unity, we can assume = {x R 3 x 3 = ϕ(x, x 2 )}, where ϕ is Lipschitz. Let x = (x, x 2 ) Dϕ, then 2

19 for j =, 2, j U(x, ϕ(x )) (0) (Dϕ) ( j U)(x, ϕ(x )) (0) (Dϕ)+ ( 3 U)(x, ϕ(x )) j ϕ(x ) (0) (Dϕ) j U (0) ( ) + C 3 U (0) ( ) C U (2) ( ), (3.0.5) due to the chain rule, bounds of j U (0) by the trace theorem, and the Lipschitz conditions of ϕ. The inequality (3.0.5) gives us a bound for U (0) ( ), which is the first term of right side of (3.0.4). The second term U (0) ( ) can be bounded by trace theorem, U (0) ( ) U ( 2 )( ) U (2) ( ). Combining these results with (3.0.)(3.0.4), we conclude that ν U (0) ( ) C U (2) ( ) C g 0 (0) ( ). Then since u = U U, it follows that ν u (0) ( ) ν U (0) ( ) + ν U (0) ( ) C( U (0) ( ) + U (0) ( )) g 0 ( ) (3.0.6) The above inequality shows Λ c Λ 0 is a bounded operator from L 2 ( + ) to L 2 ( + ). Hence the same result holds for Λ c Λ c2. 3. THE STABILITY ESTIMATE BOUND In this section, we will give the main theorem which contains the stability estimate bound for c. Consider the boundary value problem given in (2..)(2..2)(2..3). Theorem 3.. Let c j () M 0, c j () M, j =, 2, c = 0 near +, (3..) and ε = Λ c2 Λ c, E = logε, then there are constants C(, M), 0 < λ < and 0 < λ < such that c c 2 2 (0) () C(, M)ελ (E + k) λ C(, M) +, (3..2) (E + k) 2 3 λ where M = M0 2 + M 2. 3

20 For a given ε, one can minimize the bound of (3..2) with respect to k. We found that, at the minimum point k = ( 3 2 ε λ ) 3 5λ E, the right hand side C(, M)[ε λ (E + k) λ + ] = C(, M)ε 2 (E+k) 2 3 λ 5 λ, which is Hölder continuous in ε. Thus, as k grows in the zone k < ( 3 2 ε λ ) 3 5λ E, the bound decreases and become more like Hölder type when k approaches to the minimum point. This fact shows increasing stability phenomenon in the above zone. We need some lemmas to prove the theorem. The following inequality follows from the application of the Green s theorem. Lemma 3..2 For all solutions u j H () satisfying (2..))(2..2)(2..3) with potential c j, j=,2, we have (c 2 c )u u 2 = ((Λ c2 Λ c )u )u 2. + (3..3) Proof: Subtract equations u u 2 k 2 u u 2 +c u u 2 = 0 and u 2 u k 2 u 2 u +c 2 u 2 u = 0. Then integrate by parts to obtain (c 2 c )u u 2 = u u 2 + u 2 u = ν u u 2 + ν u 2 u + = ν u u 2 + ν u 2 u 2 ν u 2 u 2 + ν u 2 u + = ( ν u 2 ν u )u 2 + ν u 2 (u 2 u ) = ((Λ c2 Λ c )u )u 2, + + by the definition of Λ cj, (2..2), and u 2 = u = g 0 on THE ALMOST EXPONENTIAL SOLUTIONS Now, assume c(x) = 0 on {x R 3 x 3 > 0} \. We extend c onto the whole space R 3 as a even function of x 3. Denote V (x, x 2, x 3 ) = V (x, x 2, x 3 ) and ξ = (ξ, ξ 2, ξ 3 ). Under these abbreviations, we introduce the almost exponential solutions for the Schrödinger equation. Lemma 3.2. Let ξ R 3, ( + 4τ 2 ) 4k2, and ( + 4τ 2 ) 4C2 ξ 2 0 M 2 +2k 2 4 ξ 2 C 0 > 0, then there are almost exponential solutions for some constant u(x; j) = e iζ(j) x ( + V (x; j)) + e iζ (j) x ( + V (x; j)), j =, 2, (3.2.) 4

21 to equation u(x; j) k 2 u(x; j) + c j u(x; j) = 0 in, (3.2.2) with ζ() + ζ(2) = ξ, Imζ(j) = Imζ(j) = ξ 2 ( 4 + τ 2 ) k 2, (3.2.3) V (x; j) () () V (x; j) () () 2C 0 M ξ 2 ( + 4τ 2 ) 2k 2 + 4, (3.2.4) 2C 0 M ξ 2 ( + 4τ 2 ) 2k 2 + 4, (3.2.5) V (x; j) (2) () C 0 M( + 2), V (x; j) (2) () C 0 M( + 2) (3.2.6) and ν u(x; j) = 0 on {x 3 = 0}. (3.2.7) Proof: Suppose ξ R 3, ξ 0. We define a coordinate system (x e, x 2e, x 3e ) e in the following way. e() = (ξ 2 +ξ 2 2) 2 (ξ, ξ 2, 0), e(3) = (0, 0, ), e(2) is chosen to form an orthonormal basis e(), e(2), e(3) in R 3. Denote ξ e = ξ 2 + ξ 2 2. Now, let ζ() = ( ξ e 2 τξ 3, i( ξ 2 ( 4 + τ 2 ) k 2 ) 2, ξ τξ e) e, (3.2.8) ζ(2) = ( ξ e 2 + τξ 3, i( ξ 2 ( 4 + τ 2 ) k 2 ) 2, ξ 3 2 τξ e) e, (3.2.9) where τ is a positive real number. Then (3.2.3) is satisfied for the above ζ(), ζ(2). We will show the first term e iζ(j) x ( + V (x; j)) on right side of (3.2.) is a solution for (3.2.2) in set A = { x i <, i =, 2, 3}, then the second term is obtained by doing even reflection to e iζ(j) x ( + V (x; j)) across {x 3 = 0}, so it must be also a solution since k 2 + c is even with respect to x 3. Indeed, e iζ(j) x ( + V (x; j)) solves (3.2.2) in A if and only if V (; j) 2iζ(j) V (; j) = c j ( + V (; j)). (3.2.0) 5

22 Let P (ζ; j) = ζ ζ + 2ζ(j) ζ. By known results [2],[3], there is a regular fundamental solution E(j) of P (; j) such that for any linear partial differential operator Q with constant coefficients for any f L 2 (A), where QE(j)f (0) (A) C 0 sup ˆQ(ξ ) ˆP (ξ ) f (0) (A) (sup over ξ R 3 ) (3.2.) ˆP (ξ) = ( ξ α P (ξ) 2 ) 2. α 2 In our particular case, by letting ζ = ξ(j) + iη(j), ξ(j), η(j) R 3, for any ξ R 3, we have ˆP 2 (ξ ; j) = ( ξ 2 + 2ξ(j) ξ ) 2 + 4(η(j) ξ ) 2 + 4( ξ + ξ(j) 2 + η(j) 2 ) + 2 = ( ξ 2 + 2ξ ξ(j) + 2) 2 + 4(η(j) ξ ) 2 + 4( ξ(j) 2 + η(j) 2 ) + 8 (3.2.2) 4( ξ(j) 2 + η(j) 2 ) + 8 = 2( ξ 2 ( + 4τ 2 ) 2k 2 ) + 8, due to the choice of ζ(j) in (3.2.8) and (3.2.9). Similarly, ˆP 2 (ξ ; j) ( ξ + ξ(j) 2 ξ(j) 2 ) ξ + ξ(j) ( ξ + ξ(j) 2 + ξ(j) 2 ) + ξ 2 +, (3.2.3) due to the elementary inequalities (a b) 2 +2(a b)+ 0, or (a b) 2 +4a+2 2(a+b)+, with a = ξ + ξ(j) 2, b = ξ(j) 2. The regular fundamental solution in [2] is a convolution operator, so it commutes with differentiations, and hence from (3.2.) it follows that QE(j)f () (A) C 0 sup ˆQ(ξ ) ˆP (ξ ) f () (A) (sup over ξ R 3 ). (3.2.4) Since E(j) is a fundamental solution, any solution V (; j) to the equation V (; j) = E(j)(c j ( + V (; j))) on A (3.2.5) solves (3.2.0). From (3.2.4) with Q = and (3.2.2), it follows that E(j)f () (A) C 0 ( ξ 2 ( + 4τ 2 ) 2k 2 + 4) 2 f () (A), (3.2.6) 6

23 or E(j)f () (A) θ f () (A), θ = C 0 ξ 2 ( + 4τ 2 ) 2k (3.2.7) Observe that c j ( + V ) () (A) c j () (A) + c j V () (A) MV ol 2 (A) + 2M V () (A), (3.2.8) where we used the bound (3..) and M = M M 2. So the operator F (V (; j)) in the right side of (3.2.5) maps the ball B(ρ) = {V : V () (A) ρ} into the ball B(θMV ol 2 (A)+ θ 2Mρ), and hence into B(ρ) when θmv ol 2 (A) ( θ 2M)ρ. (3.2.9) The condition ( + 4τ 2 ) 4C2 0 M 2 +2k 2 4 ξ 2 and (3.2.7) imply that 2θM 2, and hence (3.2.9) holds with ρ = 2C 0 M ξ 2 ( + 4τ 2 ) 2k 2 + 4, (3.2.20) which implies (3.2.4), since A. Likewise, using (3.2.3) and (3.2.) with Q(ξ ) = ξ k, k =, 2, 3, from (3.2.5), we have that for V (; j) B(ρ) V (; j) (2) (A) C 0 c j ( + V (; j)) () (A) C 0 M( + 2 V (; j) () (A)) C 0 M( + 2 2C 0 M ξ 2 ( + 4τ 2 ) 2k ) C 0M( + 2), (3.2.2) due to the bound (3.2.4) given by (3.2.20) and the inequality ( + 4τ 2 ) 4C2 0 M 2 +2k 2 4 ξ 2. Therefore the operator F is continuous from H (A) into H 2 (A) and therefore compact from H (A) into itself. Now, this operator maps convex closed set B(ρ) H (A) into itself and is compact, hence by Schauder-Tikhonov Theorem, it has a fixed point V (; j) B(ρ). In view of (3.2.20) we have the bound (3.2.4) and due to (3.2.2) we have the first bound in (3.2.6). Similarly, one can apply the same argument for reflected solution e iζ (j) x ( + V (x; j)) to achieve the bound (3.2.5) and the second bound in (3.2.6). The proof is completed. 7

24 Lemma Let u(x; j) be the solutions (3.2.) to the Schrödinger equation u j k 2 u j + c j u j = 0 in from Lemma 2, then we have u(x; j) (0) ( + ) 2C()e ξ 2 ( 4 +τ 2 ) k 2. (3.2.22) Proof: Let e(x; j) = e iζ(j) x, and e (x; j) = e iζ (j) x. Clearly, we have e(x; j) (0) ( + ) + 2 e Imζ(j), e (x; j) (0) ( + ) + 2 e Imζ (j). Moreover, from trace theorems for Sobolev spaces V (; j) (0) ( + ) C() V (; j) () () C(), by (3.2.4) and the condition ( + 4τ 2 ) 4C2 0 M 2 +2k 2 4. Similarly, for the same reason ξ 2 V (; j) (0) ( + ) C() V (; j) () () C(). Therefore we have that u(; j) (0) ( + ) e(; j) (0) ( + ) + e Imζ(j) V (; j) (0) ( + ) + e (; j) (0) ( + ) + e Imζ (j) V (; j) (0) ( + ) + 2 e Imζ(j) + C()e Imζ(j) e Imζ (j) + C()e Imζ (j) = 2( C())e ξ 2 ( 4 +τ 2 ) k 2 = 2C()e ξ 2 ( 4 +τ 2 ) k 2 (3.2.23) due to (3.2.3). 3.3 THE PROOF OF INCREASING STABILITY In this section, we prove the Theorem (3..) which yields stability estimates for the potential c. 8

25 Proof of Theorem 3... Let c(x) = c 2 (x) c (x), then we find u(x; )u(x; 2) = e i(ζ()+ζ(2))x ( + V (; ))( + V (; 2) +e i(ζ ()+ζ(2))x ( + V (; ))( + V (; 2)) +e i(ζ()+ζ (2))x ( + V (; ))( + V (; 2)) (3.3.) +e i(ζ ()+ζ (2))x ( + V (; ))( + V (; 2)), Substituting u(x; )u(x; 2) into identity (3..3) and using triangle inequality give us c(x)(e iξx + e iξ x ) c(x)(v (; ) + V (; 2) + V (; )V (; 2)e iξx ) + c(x)(v (; ) + V (; 2) + V (; )V (; 2)e iξx ) + c(x)e iξ e(x e 2τx 3 ) ( + V (; ) + V (; 2) + V (; )V (; 2)) + c(x)e iξ e(x e +2τx 3 ) ( + V (; ) + V (; 2) + V (; )V (; 2)) + ε u(; ) (0) ( + ) u(; 2) (0) ( + ), (3.3.2) where ε = Λ c2 Λ c. Sobolev Embedding Theorems and (3.2.6) imply that V (; j) () C e V (; j) (2) () C e C 0 ( + 2)M. (3.3.3) Therefore, by Lemma 3.2., (3.3.3) and Hölder s inequality c(x)(v (; ) + V (; 2) + V (; )V (; 2)e iξx ) c(x) V (; ) + c(x) V (; 2) + c(x) V (; ) V (; 2) c (0) ()( V (; ) (0) + V (; 2) (0) ) + C e C 0 ( + 2)M c V (; ) c (0) ()(2 + C e C 0 ( + 2C 0 M 2)M) ξ 2 ( + 4τ 2 ) 2k (3.3.4) For the same reason c(x)(v (; ) + V (; 2) + V (; )V (; 2)e iξ x ) c (0) ()(2 + C e C 0 ( + 2C 0 M (3.3.5) 2)M) ξ 2 ( + 4τ 2 ) 2k

26 For the third term on right side of (3.3.2), let ξ e = ξ 2 + ξ 2 2, since c(x) = 0 near the boundary, we use integration by parts with respect to x 3, c(x)e iξ e(x e 2τx 3 ) ( + V (; ) + V (; 2) + V (; )V (; 2)) c(x) ( eiξe(xe 2τx3) ) + c(x)(v (; ) + V (; 2) + V (; )V (; 2)) x 3 2τξ e (c(x)) + c(x)v (; ) + c(x)v (; 2) + c(x)v (; )V (; 2) 2τ x 3 M V ol() + c 2τ (0) (2 + C e C 0 ( + 2C 0 M 2)M) ξ 2 ( + 4τ 2 ) 2k (3.3.6) For the same reason, the forth term is bounded by c(x)e iξ e(x e 2τx 3 ) ( + V (; ) + V (; 2) + V (; )V (; 2)) M V ol() 2τ + c (0) (2 + C e C 0 ( + (3.3.7) 2C 0 M 2)M) ξ 2 ( + 4τ 2 ) 2k Because c(x) is an even function of x 3, by Fourier transform, the left side of (3.3.2) gives us c(x)(e iξx + e iξ x ) = c(x)e iξx = ĉ (ξ). R 3 (3.3.8) Combining the above results (3.3.3)-(3.3.8), with help of (3.2.22), we obtain the bound for Fourier transform ĉ in ξ e >, ĉ (ξ) c (0) ()(2 + C e C 0 ( + 8C 0 M 2)M) ξ 2 ( + 4τ 2 ) 2k M V ol() + εc 2 ξ τ e 2 (+4τ 2 ) 4k 2 C(, M)( ξ 2 ( + 4τ 2 ) 2k ξ τ + εe 2 (+4τ 2 ) 4k 2 ) (3.3.9) for some constant C(, M). Taking square on both sides of (3.3.9), and using elementary inequality (a + b + c) 2 4a 2 + 4b 2 + 2c 2, then choose + 4τ 2 = 4k2 +( E 2 )2, we obtain ξ 2 ĉ 2 (ξ) C(, M)( 2k 2 + ( E 2 ) ξ 2 4k 2 + ( E 2 )2 ξ + ε) 2 C(, M)( 2k 2 + ( E 2 ) ρ 2 4k 2 + ( E 2 )2 2ρ + ε) 2 (3.3.0) for < ξ e < ρ and ξ 3 < ρ. 20

27 Now, writing f(ξ e ) = ĉ 2 (ξ e, ξ 3 ) = ( R 3 c(x)e i(x eξ e +x 3 ξ 3 ) ) 2, one can check that f(ξ e ) is an entire function if we extend ξ e = r + ir 2 onto C. Assume Γ = {0 < r <, r 2 = 0}, = { r i < 2, i =, 2}, and Γ 2 = { < r < 2, r 2 = 0} in C. Applying the corollary.2.2 [4] for analytic function f(ξ e ), ξ e C, we obtain the inequality sup Γ f(ξ e ) C(sup f(ξ e ) ) λ (sup f(ξ e ) ) λ (3.3.) Γ 2 for some 0 < λ <, and constant C. The bound of sup Γ2 f(ξ e ) is given by (3.3.0). Also by the definition of f, we have sup f(ξ e ) Me 2. Therefore, (3.3.) implies, for ξ e < and ξ 3 < ρ ĉ 2 (ξ) sup Γ f(ξ e ) CMe 2 (sup f(ξ e ) ) λ Γ 2 C(, M)[ (2k 2 + ( E 2 )2 + 4) λ + ρ 2λ (4k 2 + ( E 2 )2 2ρ 2 ) λ + ελ ]. (3.3.2) Adding the bounds of (3.3.0) and (3.3.2) gives us the bound for ĉ 2 in the domain ξ < ρ ĉ 2 (ξ) C(, M)[ (2k 2 + ( E 2 )2 + 4) + ρ 2λ λ (4k 2 + ( E 2 )2 2ρ 2 ) + λ ελ ], (3.3.3) where we have dropped all the terms in (3.3.0) due to their lower orders. Now we take integral on both sides of (3.3.3) over ξ < ρ. Let ξ = r and choose ρ = (E + k) λ 2λ+3. It follows that ξ <ρ ρ ĉ 2 (ξ) C(, M)[ C(, M)[ 0 r 2 dr (2k 2 + ( E 2 )2 + 4) λ + ρ 3 ρ 0 ρ 2λ r 2 dr (4k 2 + ( E 2 )2 2ρ 2 ) λ + (2k 2 + ( E 2 )2 + 4) λ + ρ 2λ+3 (4k 2 + ( E 2 )2 2ρ 2 ) λ + ελ ρ 3 ] (E + k) λ C(, M)[ (4k 2 + ( E 2 )2 ) + λ ελ (E + k) 3λ 2λ+3 ], ρ 0 ε λ r 2 dr] (3.3.4) where the first term on right side gets absorbed by the second one due to its lower order. Finally, in the domain ξ > ρ, we use properties of Fourier transform, ĉ 2 (ξ) ( + ξ 2 ) ĉ 2 (ξ)dξ = ρ< ξ + ρ 2 R + ρ 3 2 c 2 () () 4 + ρ M 2 4M 2 =. 2 + (E + k) 2λ 2λ+3 (3.3.5) 2

28 Adding inequalities (3.3.4) and (3.3.5), then using parseval s identity, we obtain the bound for c (0) (). Note that the first term in the bound (3.3.4) is lower order comparing to the bound (3.3.5) due to 0 < λ <, hence it gets absorbed. Therefore c c 2 2 (0) () C(, M)ελ (E + k) 3λ 2λ+3 + C(, M) (E + k) 2λ 2λ+3 (3.3.6) for some constant C(, M). Let λ = theorem THE CAUCHY BOUNDARY DATA 3λ, then (3.3.6) implies (3..2). This proves the 2λ+3 Some people argue that, since the eigenvalues for the problems (2..)(2..3) are unknown, as k, it hits infinitely many of them and we do not know when this happens. Thus the assumption that k 2 is not an eigenvalue is quite inconvenient. To avoid the spectral issue, one can replace the boundary condition (2..2)(2..3) by the Cauchy data, S c = {(u, ν u) H 2 ( + ) H 2 ( + )}, ν u= 0 on Γ, (3.4.) where + = {x 3 > 0} and the potential c C (). Denote. (s) () the norm for the Sobolev space H s (). Let S c and S c2 be two Cauchy data corresponding to c and c 2 respectively. To measure the distance between two Cauchy data, we define (f, g) ( f, g) ( 2 dist(s c, S c2 ) = max{ max min, 2 ), (f,g) S c ( f, g) S c2 (f, g) ( 2, 2 ) (f, g) ( f, g) ( 2 max min, 2 ) } (f,g) S c2 ( f, g) S c (f, g) ( 2, 2 ) where (f, g) ( 2, 2 ) = ( f 2 ( 2 )( +) + g 2 ( 2 )( +)) 2. (3.4.2) The main result (3..2) still holds under above assumptions with different choices of λ and λ 2. The following theorem gives details. 22

29 Theorem 3.4. Let c j () M 0, c j () M, j =, 2, c = 0 near +, (3.4.3) and dist(s c, S c2 ) = ε, E = logε, E > 0, then there are constants C(, M), 0 < λ < and 0 < λ i < 5, i =, 2 such that where M = M M 2. c c 2 2 (0) () C(, M)ελ (E + k) λ + C(, M) (E + k) λ 2, (3.4.4) For a given ε, one can minimize the bound of (3.4.4) with respect to k. We found that, at the minimum point k = ( λ 2 λ ε λ λ ) +λ 2 E, the right hand side C(, M)[ε λ (E + k) λ + λλ2 ] = C(, M)ε (E+k) λ λ +λ 2, which is Hölder continuous in ε. Thus, as k grows in the zone 2 k < ( λ 2 λ ε λ ) λ +λ 2 E, the bound decreases and become more like Hölder type when k approaches to the minimum point. This fact shows increasing stability phenomenon in the above zone. The following lemmas will be used in the proof of above theorem. Lemma For all solutions u j H () satisfying u j k 2 u j + c j u j = 0 in, j =, 2 with the the Cauchy boundary condition (3.4.), we have (c 2 c )u u 2 (u, ν u ) ( 2, 2 ) (u 2, ν u 2 ) ( 2, 2 ) dist(s c, S c2 ). (3.4.5) Proof: Define ν(u j ) on + in the weak form as < ν(u j ) u k >= u j u k + c j u j u k, for j, k {, 2}, where < > denotes the duality between H ( + ) and H ( + ). It is 2 2 clear that < ν(u ) u 2 > < ν(u 2 ) u >= 23 (c 2 c )u 2 u.

30 Now, since < ν u 2 u 2 > < ν(u 2 ) u 2 >= 0 for any (u 2, ν u 2 ) S c2, we have that < ν(u ) ν u 2 u 2 > < ν(u 2 ) u u 2 >= (c 2 c )u 2 u. (3.4.6) By the definition of dist(s c, S c2 ), this gives the desired result. Lemma Let u(x; j) be the solutions (3.2.) to the Schrödinger equation u j k 2 u j + c j u j = 0 in, then we have (u(; j), ν u(; j) ( 2, 2 ) C()( 5 + ξ 2 ( + 4τ 2 ) 2k 2 e ξ 2 ( 4 +τ 2 ) k 2 ). (3.4.7) Proof: Let e(x; j) = e iζ(j) x, and e (x; j) = e iζ (j) x. Due to (??) and the condition ( + 4τ 2 ) 4C2 0 M 2 +2k 2 4, we have V (; j) ξ 2 () (). Hence by the definition of Sobolev norm, e(; j)( + V (; j)) 2 () () =( e(; j) 2 + V (; j) i= Similarly, for the same reason 3 i= i e(; j) 2 + V (; j) 2 e(; j) 2 i V (; j) 2 ) C()(e 2 Imζ(j) ( + ζ(j) 2 )). e (; j)( + V (; j)) 2 () () C()(e2 Imζ(j) ( + ζ(j) 2 )). Therefore, by (3.4.2) and the trace theorem The proof is completed. (u(; j), ν u(; j) ( 2, 2 ) C() u(; j) () () C()( e(; j)( + V (; j)) () () + e (; j)( + V (; j)) () ()) C()( 5 + ξ 2 ( + 4τ 2 ) 2k 2 e ξ 2 ( 4 +τ 2 ) k 2 ). (3.4.8) Proof (Theorem 3.4.): Let c(x) = c 2 (x) c (x), then we find u(x; )u(x; 2) =e i(ζ()+ζ(2))x ( + V (; ))( + V (; 2) + e i(ζ ()+ζ(2))x ( + V (; ))( + V (; 2)) + e i(ζ()+ζ (2))x ( + V (; ))( + V (; 2)) + e i(ζ ()+ζ (2))x ( + V (; ))( + V (; 2)), 24

31 Substituting u(x; )u(x; 2) into identity (3.4.7) and using triangle inequality give us c(x)(e iξx + e iξ x ) c(x)(v (; ) + V (; 2) + V (; )V (; 2)e iξx ) + c(x)(v (; ) + V (; 2) + V (; )V (; 2)e iξx ) + c(x)e iξ e(x e 2τx 3 ) ( + V (; ) + V (; 2) + V (; )V (; 2)) + c(x)e iξ e(x e +2τx 3 ) ( + V (; ) + V (; 2) + V (; )V (; 2)) + ε (u, ν u ) ( 2, 2 ) (u 2, ν u 2 ) ( 2, 2 ). (3.4.9) Sobolev Embedding Theorems and (3.2.6) imply that V (; j) () C e V (; j) (2) () C e C 0 ( + 2)M. (3.4.0) Therefore, by the bounds of V (; j) (0) given in Lemma 3.2. c(x)(v (; ) + V (; 2) + V (; )V (; 2)e iξx ) c(x) V (; ) + c(x) V (; 2) + c(x) V (; ) V (; 2) c (0) ()( V (; ) (0) + V (; 2) (0) ) + C e C 0 ( + 2)M c V (; ) c (0) ()(2 + C e C 0 ( + 2C 0 M 2)M) ξ 2 ( + 4τ 2 ) 2k (3.4.) For the same reason c(x)(v (; ) + V (; 2) + V (; )V (; 2)e iξ x ) c (0) ()(2 + C e C 0 ( + 2C 0 M (3.4.2) 2)M) ξ 2 ( + 4τ 2 ) 2k For the third term on right side of (3.4.9), let ξ e = ξ 2 + ξ 2 2, since c(x) = 0 near the 25

32 boundary, we use integration by parts with respect to x 3, c(x)e iξ e(x e 2τx 3 ) ( + V (; ) + V (; 2) + V (; )V (; 2)) c(x) ( eiξe(xe 2τx3) ) + c(x)(v (; ) + V (; 2) + V (; )V (; 2)) x 3 2τξ e (c(x)) + c(x)v (; ) + c(x)v (; 2) + c(x)v (; )V (; 2) 2τ x 3 M V ol() + c 2τ (0) (2 + C e C 0 ( + 2C 0 M 2)M) ξ 2 ( + 4τ 2 ) 2k (3.4.3) For the same reason, the forth term is bounded by c(x)e iξ e(x e 2τx 3 ) ( + V (; ) + V (; 2) + V (; )V (; 2)) M V ol() 2τ + c (0) (2 + C e C 0 ( + (3.4.4) 2C 0 M 2)M) ξ 2 ( + 4τ 2 ) 2k Because c(x) is an even function of x 3, by Fourier transform, the left side of (3.4.9) gives us c(x)(e iξx + e iξ x ) = c(x)e iξx = ĉ (ξ). R 3 (3.4.5) Combining the above results (3.4.)-(3.4.5), with help of (3.4.7), we obtain the bound for Fourier transform ĉ in ξ e >, ĉ (ξ) c (0) ()(2 + C e C 0 ( + 8C 0 M 2)M) ξ 2 ( + 4τ 2 ) 2k M V ol() + εc()(5 + ξ 2 ( + 4τ 2 ) 2k 2 ξ )e 2 (+4τ 2 ) 4k 2 τ C(, M)( ξ 2 ( + 4τ 2 ) 2k τ + ε(5 + ξ 2 ( + 4τ 2 ) 2k 2 ξ )e 2 (+4τ 2 ) 4k 2 ) (3.4.6) for some constant C(, M). Taking square on both sides of (3.4.6), and using elementary inequality (a + b + c) 2 4a 2 + 4b 2 + 2c 2, then choose + 4τ 2 = 4k2 +( E 2 )2, we obtain ξ 2 ĉ 2 (ξ) C(, M)[ 2k 2 + ( E 2 ) ξ 2 4k 2 + ( E 2 )2 ξ + ε( k2 + E2 4 )2 ] C(, M)[ 2k 2 + ( E 2 ) ρ 2 4k 2 + ( E 2 )2 2ρ + ε( k2 + E2 4 )2 ] (3.4.7) for < ξ e < ρ and ξ 3 < ρ. 26

33 Now, writing f(ξ e ) = ĉ 2 (ξ e, ξ 3 ) = ( c(x)e i(x eξ e +x 3 ξ 3 ) ) 2, one can check that f(ξ R 3 e ) is an entire function if we extend ξ e = r + ir 2 onto C. Assume Γ = {0 < r <, r 2 = 0}, = { r i < 2, i =, 2}, and Γ 2 = { < r < 2, r 2 = 0} in C. Applying the corollary.2.2 [4] for analytic function f(ξ e ), ξ e C, we obtain the inequality sup Γ f(ξ e ) C(sup f(ξ e ) ) λ (sup f(ξ e ) ) λ (3.4.8) Γ 2 for some 0 < λ <, and constant C. The bound of sup Γ2 f(ξ e ) is given by (3.4.7). Also by the definition of f, we have sup f(ξ e ) Me 2. Therefore, (3.4.8) implies, for ξ e < and ξ 3 < ρ ĉ 2 (ξ) sup Γ C(, M)[ f(ξ e ) CMe 2 (sup f(ξ e ) ) λ Γ 2 (2k 2 + ( E 2 )2 + 4) λ + ρ 2λ (4k 2 + ( E 2 )2 2ρ 2 ) λ + ελ (5 + 2k 2 + E2 4 )2λ ]. (3.4.9) Adding the bounds of (3.4.7) and (3.4.9) gives us the bound for ĉ 2 in the domain ξ < ρ ĉ 2 (ξ) C(, M)[ (2k 2 + ( E 2 )2 + 4) + ρ 2λ λ (4k 2 + ( E 2 )2 2ρ 2 ) λ + ε λ (5 + 2k 2 + E2 4 )2λ ], where we have dropped all terms in (3.4.7) due to their lower orders. (3.4.20) Now we take integral on both sides of (3.4.20) over ξ < ρ. Let ξ = r and choose ρ = (E + k) λ 2λ+3. It follows that ξ <ρ ρ ĉ 2 r 2 dr (ξ) C(, M)[ (2k 2 + ( E 2 )2 + 4) + λ + ρ 0 0 ρ 0 ρ 2λ r 2 dr (4k 2 + ( E 2 )2 2ρ 2 ) λ ε λ (5 + 2k 2 + E2 4 )2λ r 2 dr] C(, M)[ (2k 2 + ( E 2 )2 + 4) λ + ρ 2λ+3 (4k 2 + ( E 2 )2 2ρ 2 ) λ + ελ (5 + 2k 2 + E2 4 )2λ ρ 3 ] (E + k) λ C(, M)[ (4k 2 + ( E 2 )2 ) + λ ελ (E + k) 8λ 2 +5λ 2λ+3 ], ρ 3 (3.4.2) where the first term on right side gets absorbed by the second one due to its lower order. 27

34 Finally, in the domain ξ > ρ, we use properties of Fourier transform, ĉ 2 (ξ) ( + ξ 2 ) ĉ 2 (ξ)dξ = ρ< ξ + ρ 2 R + ρ 3 2 c 2 () () 4 + ρ M 2 4M 2 =. 2 + (E + k) 2λ 2λ+3 (3.4.22) Adding inequalities (3.4.2) and (3.4.22), then using Parseval s identity, we obtain the bound for c (0) (). Note that the first term in the bound (3.4.2) is lower order comparing to the bound (3.4.22) due to 0 < λ <, hence it gets absorbed. Therefore c c 2 2 (0) () C(, M)ελ (E + k) 8λ2 +5λ 2λ+3 + C(, M) (E + k) 2λ 2λ+3 (3.4.23) for some constant C(, M). Let λ = 8λ2 +5λ and λ 2λ+3 2 = 3λ, then (3.4.23) implies (4.0.8). 2λ+3 This proves Theorem

35 CHAPTER 4 THE LINEARIZED INVERSE PROBLEM FOR THE SCHRÖDINGER EQUATION In the view of equation (2..), we notice that, since u depends on c, the inverse problem of solving c is non-linear and not convex. Thus a good approach is to consider linearized inverse problem. Such approach is useful from two point of view. First, it often occurs that a coefficient of a differential equation differs little from a known function. The problem linearized in the neighbourhood of this function may give satisfactory results for application. On the other hand, the algorithm solving full, non-linearized inverse problem would be too difficult and time consuming in real life. Therefore, for numerical computation, it is sufficient to use linearize inverse problem for the Schrödinger equation. The following proof justifies the linearization by assuming that c is a small perturbation. Let = {x R 3 0 < x i < π, i =, 2, 3}. Consider the Schrödinger equation u k 2 u + cu = 0 in, (4.0.) with the boundary data ν u = g H 2 ( ). (4.0.2) Introduce a function u 0 which satisfies u 0 k 2 u 0 = 0 in, ν u 0 = g on. By [26], pp , (4.0.3) u 0 () () < C g ( )( ), (4.0.4) 2 where C is a constant depending on k and. Let u 2 = u u 0. Subtracting (4.0.3) from (4.0.) results in u 2 k 2 u 2 = cu in, ν u 2 = 0 on. (4.0.5) 29

36 We are interested in comparing u 2 with the function u which solves the linearised Schrödinger equation u k 2 u = cu 0 in, ν u = 0 on. (4.0.6) If c < δ, then u () () Cδ g ( )( ) (4.0.7) 2 due to [26], pp and (4.0.4). Now, subtracting (4.0.6) from (4.0.5) gives (u 2 u ) k 2 (u 2 u ) + c(u 2 u ) = cu in, ν (u 2 u ) = 0 on. (4.0.8) By [26], pp and (4.0.7), we have u 2 u () () Cδ 2 g ( 2 )( ). This justifies the linearisation (replacement of u 2 by u ). From now on, we will use the linearised Schrödinger equation (4.0.6). Suppose that k 2 is not eigenvalue for (4.0.6). Define the Neumann-to-Dirichlet map Λ : g u on, (4.0.9) which is a bounded operator from L 2 ( ) to L 2 ( ) due to (4.0.7). Theorem Let c () M 0, c () M (4.0.0) and ε = Λ, E = lnε, E > 0, then c 2 (0) () 4 3 k4 π 4 ε(e + k) 3 + 4M 2 (E + k), (4.0.) 2 where M = M 0 + M. The bound in right hand side of (4.0.) suggests that there is a optimal choice of k for given ε. Indeed, by taking derivative, we found that, at the approximate minimum point 30

37 k 0 = ( 2M 2 20π 4 ε ) 9 E, the bound 4 3 k4 0π 4 ε(e + k 0 ) 3 + 4M 2 (E+k 0 ) 2 < C(, M)ε 4 9, which is Hölder continuous in ε. Thus the bound decreases in the zone k (0, k 0 ) and shows increasing stability phenomenon there. If ( 2M 2 20π 4 ε ) 9 Hölder stability still holds in this case. < E, then ε > α > 0 for some constant α, and We need some lemmas to prove Theorem First introduce a function v that solves the equations v k 2 v = 0 in, ν v = g on. (4.0.2) The following lemma follows from the simple application of Green s theorem. Lemma For solutions u, v H () satisfying (4.0.6),(4.0.2), we have cu 0 v = (Λ( ν u 0 )) ν v. (4.0.3) Lemma Let ξ R 3, ( + 4τ 2 ) 4k2 ξ 2, then there are exponential solutions u 0 = e iζ() x, v = e iζ(2) x (4.0.4) to the equations u 0 k 2 u 0 = 0, v k 2 v = 0 in (4.0.5) with ζ() + ζ(2) = ξ, ζ() ζ() = ζ(2) ζ(2) = k 2 Imζ() = Imζ(2) = ξ 2 ( 4 + τ (4.0.6) 2 ) k 2 in addition, ν u 0 (0) ( ) 2kπe Imζ(), ν v (0) ( ) 2kπe Imζ(2) (4.0.7) Proof: Suppose ξ R 3, ξ 0. We define a coordinate system (x e, x 2e, x 3e ) e in the following way. e() = (ξ 2 +ξ 2 2) 2 (ξ, ξ 2, 0), e(3) = (0, 0, ), e(2) is chosen to form an orthonormal basis 3

38 e(), e(2), e(3) in R 3. Denote ξ e = ξ 2 + ξ 2 2. Now, let ζ() = ( ξ e 2 τξ 3, i( ξ 2 ( 4 + τ 2 ) k 2 ) 2, ξ τξ e) e, (4.0.8) ζ(2) = ( ξ e 2 + τξ 3, i( ξ 2 ( 4 + τ 2 ) k 2 ) 2, ξ 3 2 τξ e) e, (4.0.9) where τ is a positive real number. Then one can check that (4.0.5)(4.0.6) are satisfied for the above choices of ζ(), ζ(2). (4.0.7) follows by direct computation of L 2 norm of ν u 0, ν v in the domain = {x R 3 0 < x i < π, i =, 2, 3}. Now, we are ready to prove Theorem 4.0.4, which gives the stability estimate for potential c with full data. Proof (Theorem 4.0.4): For u 0, v defined in (4.0.4), substituting them into identity (4.0.3) and using Cauchy Schwarz inequality results in c(x)e iξx < ε ν u 0 (0) ν v (0) < 2εk 2 π 2 ξ e 2 (+4τ) 2 4k 2, (4.0.20) where ε = Λ, and the second inequality is due to (4.0.7). Assume c(x) = 0 outside of, by the definition of Fourier transform, the left side of (4.0.20) implies c(x)e iξx = c(x)e iξx = ĉ (ξ). R 3 (4.0.2) Let + 4τ 2 = 4k2 +( E 2 )2, where E = lnε. Then (4.0.20)(4.0.2) gives ξ 2 ĉ (ξ) < 2k 2 π 2 ε. (4.0.22) Taking square on both sides of (4.0.22), then integral it over ξ < ρ. Let ξ = r, we obtain ξ <ρ ĉ 2 (ξ) < ρ 0 4k 4 π 4 εr 2 dr < 4 3 k4 π 4 ερ 3. (4.0.23) To bound the value of integral in the domain ξ > ρ, we use properties of Fourier transform, ρ< ξ ĉ 2 (ξ) ( + ξ 2 ) ĉ 2 (ξ)dξ = + ρ 2 R + ρ 3 2 c 2 () () 4 + ρ M 2. (4.0.24) 2 32

39 Adding inequalities (4.0.23) and (4.0.24), let ρ = (E + k), then using Parseval s identity, we obtain the bound for c (0) (), c 2 (0) () 4 3 k4 π 4 ε(e + k) 3 + 4M 2 (E + k). (4.0.25) 2 The above inequality proves Theorem NUMERICAL SIMULATION potential, For numerical simulation, we will choose the trigonometric representation for the c(x) = 5 n,n 2,n 3 = C n n 2 n 3 cos(n x )cos(n 2 x 2 )cos(n 3 x 3 ), (4..) where n, n 2, n 3 are integers from to 5. The goal is to recover all coefficients C n n 2 n 3. Let u 0 = e iζ() x, v = e iζ(2) x with ζ() = ( ξ e 2, (k2 ξ 2 4 ) 2, ξ 3 2 ) e, ζ(2) = ( ξ e 2, (k2 ξ 2 4 ) 2, ξ 3 2 ) e. (4..2) Substituting above u 0, v, and (4..) into the equation (4.0.3),with help of (4.0.9), results in 5 C n n 2 n 3 e iξ x cos(n x )cos(n 2 x 2 )cos(n 3 x 3 ) = u (x) ν (e iζ(2) x ), (4..3) n,n 2,n 3 = for u is the solution to (4.0.6) and = {x R 3 0 < x i < π, i =, 2, 3}. To solve all C n n 2 n 3 uniquely, it is necessary to select vector ξ multiple times to form a system of equations. Let a n n 2 n 3 (ξ) = e iξ x cos(n x )cos(n 2 x 2 )cos(n 3 x 3 ), b(k, ξ) = u ν (e iζ(2) x ), (4..4) then the system of equations generated by (4..3) will be a (ξ ) a 2 (ξ ) a 555 (ξ ) a (ξ 2 ) a 2 (ξ 2 ) a 555 (ξ 2 ) a (ξ 25 ) a 2 (ξ 25 ) a 555 (ξ 25 ) C C 2 C 555 = b(k, ξ ) b(k, ξ 2 ), (4..5). b(k, ξ 25 ) 33

40 where ξ i, i =, denotes the multiple choices of ξ. The function u contained in b(k, ξ) is the data in practical situation. Here we will design it by solving the direct problem (4.0.6). For this purpose, choose c in the first equation of (4.0.6) as c(x) = 2 cos(5x )cos(2x 2 )cos(4x 3 ) + cos(4x )cos(5x 2 )cos(2x 3 ). (4..6) The solution u is given by the Fourier series u (x) = f m m 2 m 3 cos(m x )cos(m 2 x 2 )cos(m 3 x 3 ) (4..7) m,m 2,m 3 =0 with f m m 2 m 3 = Cm m 2 m 3 f cu 0cos(m x )cos(m 2 x 2 )cos(m 3 x 3 ) (m 2 + m m 2 3) + k 2, (4..8) where C m m 2 m 3 f represents the constant for normalizing the orthogonal basis {cos(m i x i )}. Their values are given as follows C m m 2 m 3 f = π 3 if m = m 2 = m 3 = 0 C m m 2 m 3 f = 4 π 3 if m i = 0, m j m k 0 C m m 2 m 3 f = 2 π 3 if m i 0, m j = m k = 0 C m m 2 m 3 f = 8 π 3 if m m 2 m 3 0, where i, j, k {, 2, 3} and i, j, k do not take the same value. By computing right side of (4..8) carefully, we find that f m m 2 m 3 = {C m m 2 m 3 f [ 2 P (ζ (), m, 5)P (ζ (2), m 2, 2)P (ζ (3), m 3, 4) + P (ζ (), m, 5)P (ζ (2), m 2, 2)P (ζ (3), m 3, 4) + P (ζ (), m, 4)P (ζ (2), m 2, 5)P (ζ (3), m 3, 2) + P (ζ (), m, 4)P (ζ (2), m 2, 5)P (ζ (3), m 3, 2)]}/ (m 2 + m m 2 3) + k 2, where P (x, y, z) is a function of three variables defined as P (x, y, z) = 4 (ei(x+y+z)π i(x + y + z) + ei(x y+z)π i(x y + z) + ei(x+y z)π i(x + y z) + ei(x y z)π i(x y z) ), and ζ i () is the ith component of vector ζ(). 34

41 Now, to find C n n 2 n 3, we solve the system of equations (4..5) by finding the inverse of the matrix, then compare the result with the designed solution (4..6). The prediction of the theory suggests that there is an increasing stability zone and a optimal choice of k. Thus we will confirm this prediction by solving (4..5) multiple times for different k. The following algorithm is used for computation. Algorithm input: k, ξ i, for i =, , f m m 2 m 3. output: C n n 2 n 3, for n, n 2, n 3 =, A = zeros(25); for i = : 25 j = ; for n = : 5 for n 2 = : 5 for n 3 = : 5 A(i, j) = eiξi x cos(n x )cos(n 2 x 2 )cos(n 3 x 3 ); j = j + end end end end b = zeros(, 25); for i = : 25 h = 0; for m = 0 : N, (N is the truncation number for Fourier series (4..7) ) for m 2 = 0 : N for m 3 = 0 : N h = h + f m m 2 m 3 cos(m x )cos(m 2 x 2 )cos(m 3 x 3 ) ν (e iζi (2) x ); 35

42 (computation of b(k, ξ i )) end end end b(i) = h; end c = A/b; In the above algorithm, the choice of ξ i is quite crucial. They need to be selected so that the condition number of the first matrix in (4..6) is as small as possible. A good choice is given by the following code i = ; for n = : 5 for n 2 = : 5 for n 3 = : 5 ξ i = (( ) i+ ( + n 2 ), ( ) i ( + n 2 2), ( ) i+ ( + n 2 3)); i = i + end end end, more precisely, in the above code, for i =, 2..5, n =, n 2 = 2, n 3 =, 2..5, for i = 6, 7..0, n =, n 2 = 2, n 3 =, 2..5, and so forth. The condition number roughly equals 200 for the above ξ i. The following figures illustrate the outcome of the algorithm. 36

On uniqueness in the inverse conductivity problem with local data

On uniqueness in the inverse conductivity problem with local data On uniqueness in the inverse conductivity problem with local data Victor Isakov June 21, 2006 1 Introduction The inverse condictivity problem with many boundary measurements consists of recovery of conductivity

More information

Complex geometrical optics solutions for Lipschitz conductivities

Complex geometrical optics solutions for Lipschitz conductivities Rev. Mat. Iberoamericana 19 (2003), 57 72 Complex geometrical optics solutions for Lipschitz conductivities Lassi Päivärinta, Alexander Panchenko and Gunther Uhlmann Abstract We prove the existence of

More information

Follow links Class Use and other Permissions. For more information, send to:

Follow links Class Use and other Permissions. For more information, send  to: COPYRIGHT NOTICE: Kari Astala, Tadeusz Iwaniec & Gaven Martin: Elliptic Partial Differential Equations and Quasiconformal Mappings in the Plane is published by Princeton University Press and copyrighted,

More information

Outline of Fourier Series: Math 201B

Outline of Fourier Series: Math 201B Outline of Fourier Series: Math 201B February 24, 2011 1 Functions and convolutions 1.1 Periodic functions Periodic functions. Let = R/(2πZ) denote the circle, or onedimensional torus. A function f : C

More information

Stability and instability in inverse problems

Stability and instability in inverse problems Stability and instability in inverse problems Mikhail I. Isaev supervisor: Roman G. Novikov Centre de Mathématiques Appliquées, École Polytechnique November 27, 2013. Plan of the presentation The Gel fand

More information

Traces, extensions and co-normal derivatives for elliptic systems on Lipschitz domains

Traces, extensions and co-normal derivatives for elliptic systems on Lipschitz domains Traces, extensions and co-normal derivatives for elliptic systems on Lipschitz domains Sergey E. Mikhailov Brunel University West London, Department of Mathematics, Uxbridge, UB8 3PH, UK J. Math. Analysis

More information

SHARP BOUNDARY TRACE INEQUALITIES. 1. Introduction

SHARP BOUNDARY TRACE INEQUALITIES. 1. Introduction SHARP BOUNDARY TRACE INEQUALITIES GILES AUCHMUTY Abstract. This paper describes sharp inequalities for the trace of Sobolev functions on the boundary of a bounded region R N. The inequalities bound (semi-)norms

More information

Analysis Finite and Infinite Sets The Real Numbers The Cantor Set

Analysis Finite and Infinite Sets The Real Numbers The Cantor Set Analysis Finite and Infinite Sets Definition. An initial segment is {n N n n 0 }. Definition. A finite set can be put into one-to-one correspondence with an initial segment. The empty set is also considered

More information

Sobolev Spaces 27 PART II. Review of Sobolev Spaces

Sobolev Spaces 27 PART II. Review of Sobolev Spaces Sobolev Spaces 27 PART II Review of Sobolev Spaces Sobolev Spaces 28 SOBOLEV SPACES WEAK DERIVATIVES I Given R d, define a multi index α as an ordered collection of integers α = (α 1,...,α d ), such that

More information

Reminder Notes for the Course on Distribution Theory

Reminder Notes for the Course on Distribution Theory Reminder Notes for the Course on Distribution Theory T. C. Dorlas Dublin Institute for Advanced Studies School of Theoretical Physics 10 Burlington Road, Dublin 4, Ireland. Email: dorlas@stp.dias.ie March

More information

SPECTRAL PROPERTIES OF THE LAPLACIAN ON BOUNDED DOMAINS

SPECTRAL PROPERTIES OF THE LAPLACIAN ON BOUNDED DOMAINS SPECTRAL PROPERTIES OF THE LAPLACIAN ON BOUNDED DOMAINS TSOGTGEREL GANTUMUR Abstract. After establishing discrete spectra for a large class of elliptic operators, we present some fundamental spectral properties

More information

Reconstructing conductivities with boundary corrected D-bar method

Reconstructing conductivities with boundary corrected D-bar method Reconstructing conductivities with boundary corrected D-bar method Janne Tamminen June 24, 2011 Short introduction to EIT The Boundary correction procedure The D-bar method Simulation of measurement data,

More information

Vectors in Function Spaces

Vectors in Function Spaces Jim Lambers MAT 66 Spring Semester 15-16 Lecture 18 Notes These notes correspond to Section 6.3 in the text. Vectors in Function Spaces We begin with some necessary terminology. A vector space V, also

More information

Sobolev spaces. May 18

Sobolev spaces. May 18 Sobolev spaces May 18 2015 1 Weak derivatives The purpose of these notes is to give a very basic introduction to Sobolev spaces. More extensive treatments can e.g. be found in the classical references

More information

Dangerous and Illegal Operations in Calculus Do we avoid differentiating discontinuous functions because it s impossible, unwise, or simply out of

Dangerous and Illegal Operations in Calculus Do we avoid differentiating discontinuous functions because it s impossible, unwise, or simply out of Dangerous and Illegal Operations in Calculus Do we avoid differentiating discontinuous functions because it s impossible, unwise, or simply out of ignorance and fear? Despite the risks, many natural phenomena

More information

Fact Sheet Functional Analysis

Fact Sheet Functional Analysis Fact Sheet Functional Analysis Literature: Hackbusch, W.: Theorie und Numerik elliptischer Differentialgleichungen. Teubner, 986. Knabner, P., Angermann, L.: Numerik partieller Differentialgleichungen.

More information

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.

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. Sep. 1 9 Intuitively, the solution u to the Poisson equation S chauder Theory u = f 1 should have better regularity than the right hand side f. In particular one expects u to be twice more differentiable

More information

The continuity method

The continuity method The continuity method The method of continuity is used in conjunction with a priori estimates to prove the existence of suitably regular solutions to elliptic partial differential equations. One crucial

More information

be the set of complex valued 2π-periodic functions f on R such that

be the set of complex valued 2π-periodic functions f on R such that . Fourier series. Definition.. Given a real number P, we say a complex valued function f on R is P -periodic if f(x + P ) f(x) for all x R. We let be the set of complex valued -periodic functions f on

More information

CRACK DETECTION IN A THREE DIMENSIONAL BODY. A Thesis by. Steven F. Redpath. BS, Kansas University, May Submitted to the

CRACK DETECTION IN A THREE DIMENSIONAL BODY. A Thesis by. Steven F. Redpath. BS, Kansas University, May Submitted to the CRACK DETECTION IN A THREE DIMENSIONAL BODY A Thesis by Steven F. Redpath BS, Kansas University, May 1986 Submitted to the Department of Mathematics and Statistics, College of Liberal Arts and Sciences

More information

Lectures on. Sobolev Spaces. S. Kesavan The Institute of Mathematical Sciences, Chennai.

Lectures on. Sobolev Spaces. S. Kesavan The Institute of Mathematical Sciences, Chennai. Lectures on Sobolev Spaces S. Kesavan The Institute of Mathematical Sciences, Chennai. e-mail: kesh@imsc.res.in 2 1 Distributions In this section we will, very briefly, recall concepts from the theory

More information

TD 1: Hilbert Spaces and Applications

TD 1: Hilbert Spaces and Applications Université Paris-Dauphine Functional Analysis and PDEs Master MMD-MA 2017/2018 Generalities TD 1: Hilbert Spaces and Applications Exercise 1 (Generalized Parallelogram law). Let (H,, ) be a Hilbert space.

More information

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

TD M1 EDP 2018 no 2 Elliptic equations: regularity, maximum principle TD M EDP 08 no Elliptic equations: regularity, maximum principle Estimates in the sup-norm I Let be an open bounded subset of R d of class C. Let A = (a ij ) be a symmetric matrix of functions of class

More information

EXPOSITORY NOTES ON DISTRIBUTION THEORY, FALL 2018

EXPOSITORY NOTES ON DISTRIBUTION THEORY, FALL 2018 EXPOSITORY NOTES ON DISTRIBUTION THEORY, FALL 2018 While these notes are under construction, I expect there will be many typos. The main reference for this is volume 1 of Hörmander, The analysis of liner

More information

The Dirichlet s P rinciple. In this lecture we discuss an alternative formulation of the Dirichlet problem for the Laplace equation:

The Dirichlet s P rinciple. In this lecture we discuss an alternative formulation of the Dirichlet problem for the Laplace equation: Oct. 1 The Dirichlet s P rinciple In this lecture we discuss an alternative formulation of the Dirichlet problem for the Laplace equation: 1. Dirichlet s Principle. u = in, u = g on. ( 1 ) If we multiply

More information

Lecture Notes 1: Vector spaces

Lecture Notes 1: Vector spaces Optimization-based data analysis Fall 2017 Lecture Notes 1: Vector spaces In this chapter we review certain basic concepts of linear algebra, highlighting their application to signal processing. 1 Vector

More information

Sobolev Spaces. Chapter 10

Sobolev Spaces. Chapter 10 Chapter 1 Sobolev Spaces We now define spaces H 1,p (R n ), known as Sobolev spaces. For u to belong to H 1,p (R n ), we require that u L p (R n ) and that u have weak derivatives of first order in L p

More information

Proceedings of the 5th International Conference on Inverse Problems in Engineering: Theory and Practice, Cambridge, UK, 11-15th July 2005

Proceedings of the 5th International Conference on Inverse Problems in Engineering: Theory and Practice, Cambridge, UK, 11-15th July 2005 Proceedings of the 5th International Conference on Inverse Problems in Engineering: Theory and Practice, Cambridge, UK, 11-15th July 2005 SOME INVERSE SCATTERING PROBLEMS FOR TWO-DIMENSIONAL SCHRÖDINGER

More information

Inverse Gravimetry Problem

Inverse Gravimetry Problem Inverse Gravimetry Problem Victor Isakov September 21, 2010 Department of M athematics and Statistics W ichita State U niversity W ichita, KS 67260 0033, U.S.A. e mail : victor.isakov@wichita.edu 1 Formulation.

More information

Partial Differential Equations

Partial Differential Equations Part II Partial Differential Equations Year 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2015 Paper 4, Section II 29E Partial Differential Equations 72 (a) Show that the Cauchy problem for u(x,

More information

Finite-dimensional spaces. C n is the space of n-tuples x = (x 1,..., x n ) of complex numbers. It is a Hilbert space with the inner product

Finite-dimensional spaces. C n is the space of n-tuples x = (x 1,..., x n ) of complex numbers. It is a Hilbert space with the inner product Chapter 4 Hilbert Spaces 4.1 Inner Product Spaces Inner Product Space. A complex vector space E is called an inner product space (or a pre-hilbert space, or a unitary space) if there is a mapping (, )

More information

Laplace s Equation. Chapter Mean Value Formulas

Laplace s Equation. Chapter Mean Value Formulas Chapter 1 Laplace s Equation Let be an open set in R n. A function u C 2 () is called harmonic in if it satisfies Laplace s equation n (1.1) u := D ii u = 0 in. i=1 A function u C 2 () is called subharmonic

More information

Regularity for Poisson Equation

Regularity for Poisson Equation Regularity for Poisson Equation OcMountain Daylight Time. 4, 20 Intuitively, the solution u to the Poisson equation u= f () should have better regularity than the right hand side f. In particular one expects

More information

COMPLEX SPHERICAL WAVES AND INVERSE PROBLEMS IN UNBOUNDED DOMAINS

COMPLEX SPHERICAL WAVES AND INVERSE PROBLEMS IN UNBOUNDED DOMAINS COMPLEX SPHERICAL WAVES AND INVERSE PROBLEMS IN UNBOUNDED DOMAINS MIKKO SALO AND JENN-NAN WANG Abstract. This work is motivated by the inverse conductivity problem of identifying an embedded object in

More information

Decouplings and applications

Decouplings and applications April 27, 2018 Let Ξ be a collection of frequency points ξ on some curved, compact manifold S of diameter 1 in R n (e.g. the unit sphere S n 1 ) Let B R = B(c, R) be a ball with radius R 1. Let also a

More information

Hyperbolic inverse problems and exact controllability

Hyperbolic inverse problems and exact controllability Hyperbolic inverse problems and exact controllability Lauri Oksanen University College London An inverse initial source problem Let M R n be a compact domain with smooth strictly convex boundary, and let

More information

A CALDERÓN PROBLEM WITH FREQUENCY-DIFFERENTIAL DATA IN DISPERSIVE MEDIA. has a unique solution. The Dirichlet-to-Neumann map

A CALDERÓN PROBLEM WITH FREQUENCY-DIFFERENTIAL DATA IN DISPERSIVE MEDIA. has a unique solution. The Dirichlet-to-Neumann map A CALDERÓN PROBLEM WITH FREQUENCY-DIFFERENTIAL DATA IN DISPERSIVE MEDIA SUNGWHAN KIM AND ALEXANDRU TAMASAN ABSTRACT. We consider the problem of identifying a complex valued coefficient γ(x, ω) in the conductivity

More information

LECTURE 1: SOURCES OF ERRORS MATHEMATICAL TOOLS A PRIORI ERROR ESTIMATES. Sergey Korotov,

LECTURE 1: SOURCES OF ERRORS MATHEMATICAL TOOLS A PRIORI ERROR ESTIMATES. Sergey Korotov, LECTURE 1: SOURCES OF ERRORS MATHEMATICAL TOOLS A PRIORI ERROR ESTIMATES Sergey Korotov, Institute of Mathematics Helsinki University of Technology, Finland Academy of Finland 1 Main Problem in Mathematical

More information

Math212a1413 The Lebesgue integral.

Math212a1413 The Lebesgue integral. Math212a1413 The Lebesgue integral. October 28, 2014 Simple functions. In what follows, (X, F, m) is a space with a σ-field of sets, and m a measure on F. The purpose of today s lecture is to develop the

More information

Chapter 7: Bounded Operators in Hilbert Spaces

Chapter 7: Bounded Operators in Hilbert Spaces Chapter 7: Bounded Operators in Hilbert Spaces I-Liang Chern Department of Applied Mathematics National Chiao Tung University and Department of Mathematics National Taiwan University Fall, 2013 1 / 84

More information

Course Description for Real Analysis, Math 156

Course Description for Real Analysis, Math 156 Course Description for Real Analysis, Math 156 In this class, we will study elliptic PDE, Fourier analysis, and dispersive PDE. Here is a quick summary of the topics will study study. They re described

More information

Appendix A Functional Analysis

Appendix A Functional Analysis Appendix A Functional Analysis A.1 Metric Spaces, Banach Spaces, and Hilbert Spaces Definition A.1. Metric space. Let X be a set. A map d : X X R is called metric on X if for all x,y,z X it is i) d(x,y)

More information

Boundary problems for fractional Laplacians

Boundary problems for fractional Laplacians Boundary problems for fractional Laplacians Gerd Grubb Copenhagen University Spectral Theory Workshop University of Kent April 14 17, 2014 Introduction The fractional Laplacian ( ) a, 0 < a < 1, has attracted

More information

7: FOURIER SERIES STEVEN HEILMAN

7: FOURIER SERIES STEVEN HEILMAN 7: FOURIER SERIES STEVE HEILMA Contents 1. Review 1 2. Introduction 1 3. Periodic Functions 2 4. Inner Products on Periodic Functions 3 5. Trigonometric Polynomials 5 6. Periodic Convolutions 7 7. Fourier

More information

A local estimate from Radon transform and stability of Inverse EIT with partial data

A local estimate from Radon transform and stability of Inverse EIT with partial data A local estimate from Radon transform and stability of Inverse EIT with partial data Alberto Ruiz Universidad Autónoma de Madrid U. California, Irvine.June 2012 w/ P. Caro (U. Helsinki) and D. Dos Santos

More information

Numerical Solutions to Partial Differential Equations

Numerical Solutions to Partial Differential Equations Numerical Solutions to Partial Differential Equations Zhiping Li LMAM and School of Mathematical Sciences Peking University Sobolev Embedding Theorems Embedding Operators and the Sobolev Embedding Theorem

More information

Hilbert Spaces. Hilbert space is a vector space with some extra structure. We start with formal (axiomatic) definition of a vector space.

Hilbert Spaces. Hilbert space is a vector space with some extra structure. We start with formal (axiomatic) definition of a vector space. Hilbert Spaces Hilbert space is a vector space with some extra structure. We start with formal (axiomatic) definition of a vector space. Vector Space. Vector space, ν, over the field of complex numbers,

More information

Global Maxwellians over All Space and Their Relation to Conserved Quantites of Classical Kinetic Equations

Global Maxwellians over All Space and Their Relation to Conserved Quantites of Classical Kinetic Equations Global Maxwellians over All Space and Their Relation to Conserved Quantites of Classical Kinetic Equations C. David Levermore Department of Mathematics and Institute for Physical Science and Technology

More information

Fourier Transform & Sobolev Spaces

Fourier Transform & Sobolev Spaces Fourier Transform & Sobolev Spaces Michael Reiter, Arthur Schuster Summer Term 2008 Abstract We introduce the concept of weak derivative that allows us to define new interesting Hilbert spaces the Sobolev

More information

Recall that any inner product space V has an associated norm defined by

Recall that any inner product space V has an associated norm defined by Hilbert Spaces Recall that any inner product space V has an associated norm defined by v = v v. Thus an inner product space can be viewed as a special kind of normed vector space. In particular every inner

More information

Short note on compact operators - Monday 24 th March, Sylvester Eriksson-Bique

Short note on compact operators - Monday 24 th March, Sylvester Eriksson-Bique Short note on compact operators - Monday 24 th March, 2014 Sylvester Eriksson-Bique 1 Introduction In this note I will give a short outline about the structure theory of compact operators. I restrict attention

More information

Eigenvalues and Eigenfunctions of the Laplacian

Eigenvalues and Eigenfunctions of the Laplacian The Waterloo Mathematics Review 23 Eigenvalues and Eigenfunctions of the Laplacian Mihai Nica University of Waterloo mcnica@uwaterloo.ca Abstract: The problem of determining the eigenvalues and eigenvectors

More information

RANDOM PROPERTIES BENOIT PAUSADER

RANDOM PROPERTIES BENOIT PAUSADER RANDOM PROPERTIES BENOIT PAUSADER. Quasilinear problems In general, one consider the following trichotomy for nonlinear PDEs: A semilinear problem is a problem where the highest-order terms appears linearly

More information

Second Order Elliptic PDE

Second Order Elliptic PDE Second Order Elliptic PDE T. Muthukumar tmk@iitk.ac.in December 16, 2014 Contents 1 A Quick Introduction to PDE 1 2 Classification of Second Order PDE 3 3 Linear Second Order Elliptic Operators 4 4 Periodic

More information

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

Some lecture notes for Math 6050E: PDEs, Fall 2016 Some lecture notes for Math 65E: PDEs, Fall 216 Tianling Jin December 1, 216 1 Variational methods We discuss an example of the use of variational methods in obtaining existence of solutions. Theorem 1.1.

More information

HARMONIC ANALYSIS TERENCE TAO

HARMONIC ANALYSIS TERENCE TAO HARMONIC ANALYSIS TERENCE TAO Analysis in general tends to revolve around the study of general classes of functions (often real-valued or complex-valued) and operators (which take one or more functions

More information

Topics in Fourier analysis - Lecture 2.

Topics in Fourier analysis - Lecture 2. Topics in Fourier analysis - Lecture 2. Akos Magyar 1 Infinite Fourier series. In this section we develop the basic theory of Fourier series of periodic functions of one variable, but only to the extent

More information

2. Function spaces and approximation

2. Function spaces and approximation 2.1 2. Function spaces and approximation 2.1. The space of test functions. Notation and prerequisites are collected in Appendix A. Let Ω be an open subset of R n. The space C0 (Ω), consisting of the C

More information

SYMMETRY RESULTS FOR PERTURBED PROBLEMS AND RELATED QUESTIONS. Massimo Grosi Filomena Pacella S. L. Yadava. 1. Introduction

SYMMETRY RESULTS FOR PERTURBED PROBLEMS AND RELATED QUESTIONS. Massimo Grosi Filomena Pacella S. L. Yadava. 1. Introduction Topological Methods in Nonlinear Analysis Journal of the Juliusz Schauder Center Volume 21, 2003, 211 226 SYMMETRY RESULTS FOR PERTURBED PROBLEMS AND RELATED QUESTIONS Massimo Grosi Filomena Pacella S.

More information

Exercise 1. Let f be a nonnegative measurable function. Show that. where ϕ is taken over all simple functions with ϕ f. k 1.

Exercise 1. Let f be a nonnegative measurable function. Show that. where ϕ is taken over all simple functions with ϕ f. k 1. Real Variables, Fall 2014 Problem set 3 Solution suggestions xercise 1. Let f be a nonnegative measurable function. Show that f = sup ϕ, where ϕ is taken over all simple functions with ϕ f. For each n

More information

2 Infinite products and existence of compactly supported φ

2 Infinite products and existence of compactly supported φ 415 Wavelets 1 Infinite products and existence of compactly supported φ Infinite products.1 Infinite products a n need to be defined via limits. But we do not simply say that a n = lim a n N whenever the

More information

i=1 α i. Given an m-times continuously

i=1 α i. Given an m-times continuously 1 Fundamentals 1.1 Classification and characteristics Let Ω R d, d N, d 2, be an open set and α = (α 1,, α d ) T N d 0, N 0 := N {0}, a multiindex with α := d i=1 α i. Given an m-times continuously differentiable

More information

Point estimates for Green s matrix to boundary value problems for second order elliptic systems in a polyhedral cone

Point estimates for Green s matrix to boundary value problems for second order elliptic systems in a polyhedral cone Maz ya, V. G., Roßmann, J.: Estimates for Green s matrix 1 ZAMM Z. angew. Math. Mech. 00 2004 0, 1 30 Maz ya, V. G.; Roßmann, J. Point estimates for Green s matrix to boundary value problems for second

More information

Tools from Lebesgue integration

Tools from Lebesgue integration Tools from Lebesgue integration E.P. van den Ban Fall 2005 Introduction In these notes we describe some of the basic tools from the theory of Lebesgue integration. Definitions and results will be given

More information

The uniform uncertainty principle and compressed sensing Harmonic analysis and related topics, Seville December 5, 2008

The uniform uncertainty principle and compressed sensing Harmonic analysis and related topics, Seville December 5, 2008 The uniform uncertainty principle and compressed sensing Harmonic analysis and related topics, Seville December 5, 2008 Emmanuel Candés (Caltech), Terence Tao (UCLA) 1 Uncertainty principles A basic principle

More information

Calderón s inverse problem in 2D Electrical Impedance Tomography

Calderón s inverse problem in 2D Electrical Impedance Tomography Calderón s inverse problem in 2D Electrical Impedance Tomography Kari Astala (University of Helsinki) Joint work with: Matti Lassas, Lassi Päivärinta, Samuli Siltanen, Jennifer Mueller and Alan Perämäki

More information

Integral Representation Formula, Boundary Integral Operators and Calderón projection

Integral Representation Formula, Boundary Integral Operators and Calderón projection Integral Representation Formula, Boundary Integral Operators and Calderón projection Seminar BEM on Wave Scattering Franziska Weber ETH Zürich October 22, 2010 Outline Integral Representation Formula Newton

More information

Exercises to Applied Functional Analysis

Exercises to Applied Functional Analysis Exercises to Applied Functional Analysis Exercises to Lecture 1 Here are some exercises about metric spaces. Some of the solutions can be found in my own additional lecture notes on Blackboard, as the

More information

Notes on Sobolev Spaces A. Visintin a.a

Notes on Sobolev Spaces A. Visintin a.a Notes on Sobolev Spaces A. Visintin a.a. 2017-18 Contents: 1. Hölder spaces. 2. Regularity of Euclidean domains. 3. Sobolev spaces of positive integer order. 4. Sobolev spaces of real integer order. 5.

More information

Fourier Series. ,..., e ixn ). Conversely, each 2π-periodic function φ : R n C induces a unique φ : T n C for which φ(e ix 1

Fourier Series. ,..., e ixn ). Conversely, each 2π-periodic function φ : R n C induces a unique φ : T n C for which φ(e ix 1 Fourier Series Let {e j : 1 j n} be the standard basis in R n. We say f : R n C is π-periodic in each variable if f(x + πe j ) = f(x) x R n, 1 j n. We can identify π-periodic functions with functions on

More information

Functional Analysis. Franck Sueur Metric spaces Definitions Completeness Compactness Separability...

Functional Analysis. Franck Sueur Metric spaces Definitions Completeness Compactness Separability... Functional Analysis Franck Sueur 2018-2019 Contents 1 Metric spaces 1 1.1 Definitions........................................ 1 1.2 Completeness...................................... 3 1.3 Compactness......................................

More information

Existence and Uniqueness

Existence and Uniqueness Chapter 3 Existence and Uniqueness An intellect which at a certain moment would know all forces that set nature in motion, and all positions of all items of which nature is composed, if this intellect

More information

Topological properties of Z p and Q p and Euclidean models

Topological properties of Z p and Q p and Euclidean models Topological properties of Z p and Q p and Euclidean models Samuel Trautwein, Esther Röder, Giorgio Barozzi November 3, 20 Topology of Q p vs Topology of R Both R and Q p are normed fields and complete

More information

A COUNTEREXAMPLE TO AN ENDPOINT BILINEAR STRICHARTZ INEQUALITY TERENCE TAO. t L x (R R2 ) f L 2 x (R2 )

A COUNTEREXAMPLE TO AN ENDPOINT BILINEAR STRICHARTZ INEQUALITY TERENCE TAO. t L x (R R2 ) f L 2 x (R2 ) Electronic Journal of Differential Equations, Vol. 2006(2006), No. 5, pp. 6. ISSN: 072-669. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu ftp ejde.math.txstate.edu (login: ftp) A COUNTEREXAMPLE

More information

Banach space valued Cauchy-Riemann equations with totally real boundary conditions corrected

Banach space valued Cauchy-Riemann equations with totally real boundary conditions corrected Banach space valued Cauchy-Riemann equations with totally real boundary conditions corrected Katrin Wehrheim August 18, 2010 Abstract The main purpose of this paper is to give a general regularity result

More information

Hilbert spaces. 1. Cauchy-Schwarz-Bunyakowsky inequality

Hilbert spaces. 1. Cauchy-Schwarz-Bunyakowsky inequality (October 29, 2016) Hilbert spaces Paul Garrett garrett@math.umn.edu http://www.math.umn.edu/ garrett/ [This document is http://www.math.umn.edu/ garrett/m/fun/notes 2016-17/03 hsp.pdf] Hilbert spaces are

More information

Friedrich symmetric systems

Friedrich symmetric systems viii CHAPTER 8 Friedrich symmetric systems In this chapter, we describe a theory due to Friedrich [13] for positive symmetric systems, which gives the existence and uniqueness of weak solutions of boundary

More information

Fourier Series. 1. Review of Linear Algebra

Fourier Series. 1. Review of Linear Algebra Fourier Series In this section we give a short introduction to Fourier Analysis. If you are interested in Fourier analysis and would like to know more detail, I highly recommend the following book: Fourier

More information

Topics in Harmonic Analysis Lecture 6: Pseudodifferential calculus and almost orthogonality

Topics in Harmonic Analysis Lecture 6: Pseudodifferential calculus and almost orthogonality Topics in Harmonic Analysis Lecture 6: Pseudodifferential calculus and almost orthogonality Po-Lam Yung The Chinese University of Hong Kong Introduction While multiplier operators are very useful in studying

More information

Decoupling course outline Decoupling theory is a recent development in Fourier analysis with applications in partial differential equations and

Decoupling course outline Decoupling theory is a recent development in Fourier analysis with applications in partial differential equations and Decoupling course outline Decoupling theory is a recent development in Fourier analysis with applications in partial differential equations and analytic number theory. It studies the interference patterns

More information

4 Hilbert spaces. The proof of the Hilbert basis theorem is not mathematics, it is theology. Camille Jordan

4 Hilbert spaces. The proof of the Hilbert basis theorem is not mathematics, it is theology. Camille Jordan The proof of the Hilbert basis theorem is not mathematics, it is theology. Camille Jordan Wir müssen wissen, wir werden wissen. David Hilbert We now continue to study a special class of Banach spaces,

More information

A survey on l 2 decoupling

A survey on l 2 decoupling A survey on l 2 decoupling Po-Lam Yung 1 The Chinese University of Hong Kong January 31, 2018 1 Research partially supported by HKRGC grant 14313716, and by CUHK direct grants 4053220, 4441563 Introduction

More information

CHAPTER VIII HILBERT SPACES

CHAPTER VIII HILBERT SPACES CHAPTER VIII HILBERT SPACES DEFINITION Let X and Y be two complex vector spaces. A map T : X Y is called a conjugate-linear transformation if it is a reallinear transformation from X into Y, and if T (λx)

More information

Eigenvalues and eigenfunctions of the Laplacian. Andrew Hassell

Eigenvalues and eigenfunctions of the Laplacian. Andrew Hassell Eigenvalues and eigenfunctions of the Laplacian Andrew Hassell 1 2 The setting In this talk I will consider the Laplace operator,, on various geometric spaces M. Here, M will be either a bounded Euclidean

More information

Introduction to Real Analysis Alternative Chapter 1

Introduction to Real Analysis Alternative Chapter 1 Christopher Heil Introduction to Real Analysis Alternative Chapter 1 A Primer on Norms and Banach Spaces Last Updated: March 10, 2018 c 2018 by Christopher Heil Chapter 1 A Primer on Norms and Banach Spaces

More information

PAPER 5 ANALYSIS OF PARTIAL DIFFERENTIAL EQUATIONS

PAPER 5 ANALYSIS OF PARTIAL DIFFERENTIAL EQUATIONS MATHEMATICAL TRIPOS Part III Tuesday, 2 June, 2015 9:00 am to 12:00 pm PAPER 5 ANALYSIS OF PARTIAL DIFFERENTIAL EQUATIONS There are THREE questions in total. Attempt all THREE questions. The questions

More information

AN INVERSE PROBLEM FOR THE WAVE EQUATION WITH A TIME DEPENDENT COEFFICIENT

AN INVERSE PROBLEM FOR THE WAVE EQUATION WITH A TIME DEPENDENT COEFFICIENT AN INVERSE PROBLEM FOR THE WAVE EQUATION WITH A TIME DEPENDENT COEFFICIENT Rakesh Department of Mathematics University of Delaware Newark, DE 19716 A.G.Ramm Department of Mathematics Kansas State University

More information

Eigenvalues, random walks and Ramanujan graphs

Eigenvalues, random walks and Ramanujan graphs Eigenvalues, random walks and Ramanujan graphs David Ellis 1 The Expander Mixing lemma We have seen that a bounded-degree graph is a good edge-expander if and only if if has large spectral gap If G = (V,

More information

are Banach algebras. f(x)g(x) max Example 7.4. Similarly, A = L and A = l with the pointwise multiplication

are Banach algebras. f(x)g(x) max Example 7.4. Similarly, A = L and A = l with the pointwise multiplication 7. Banach algebras Definition 7.1. A is called a Banach algebra (with unit) if: (1) A is a Banach space; (2) There is a multiplication A A A that has the following properties: (xy)z = x(yz), (x + y)z =

More information

An introduction to some aspects of functional analysis

An introduction to some aspects of functional analysis An introduction to some aspects of functional analysis Stephen Semmes Rice University Abstract These informal notes deal with some very basic objects in functional analysis, including norms and seminorms

More information

Hölder regularity estimation by Hart Smith and Curvelet transforms

Hölder regularity estimation by Hart Smith and Curvelet transforms Hölder regularity estimation by Hart Smith and Curvelet transforms Jouni Sampo Lappeenranta University Of Technology Department of Mathematics and Physics Finland 18th September 2007 This research is done

More information

Microlocal analysis and inverse problems Lecture 3 : Carleman estimates

Microlocal analysis and inverse problems Lecture 3 : Carleman estimates Microlocal analysis and inverse problems ecture 3 : Carleman estimates David Dos Santos Ferreira AGA Université de Paris 13 Monday May 16 Instituto de Ciencias Matemáticas, Madrid David Dos Santos Ferreira

More information

2 A Model, Harmonic Map, Problem

2 A Model, Harmonic Map, Problem ELLIPTIC SYSTEMS JOHN E. HUTCHINSON Department of Mathematics School of Mathematical Sciences, A.N.U. 1 Introduction Elliptic equations model the behaviour of scalar quantities u, such as temperature or

More information

An introduction to Birkhoff normal form

An introduction to Birkhoff normal form An introduction to Birkhoff normal form Dario Bambusi Dipartimento di Matematica, Universitá di Milano via Saldini 50, 0133 Milano (Italy) 19.11.14 1 Introduction The aim of this note is to present an

More information

B Elements of Complex Analysis

B Elements of Complex Analysis Fourier Transform Methods in Finance By Umberto Cherubini Giovanni Della Lunga Sabrina Mulinacci Pietro Rossi Copyright 21 John Wiley & Sons Ltd B Elements of Complex Analysis B.1 COMPLEX NUMBERS The purpose

More information

Analysis-3 lecture schemes

Analysis-3 lecture schemes Analysis-3 lecture schemes (with Homeworks) 1 Csörgő István November, 2015 1 A jegyzet az ELTE Informatikai Kar 2015. évi Jegyzetpályázatának támogatásával készült Contents 1. Lesson 1 4 1.1. The Space

More information

THEOREMS, ETC., FOR MATH 515

THEOREMS, ETC., FOR MATH 515 THEOREMS, ETC., FOR MATH 515 Proposition 1 (=comment on page 17). If A is an algebra, then any finite union or finite intersection of sets in A is also in A. Proposition 2 (=Proposition 1.1). For every

More information

Sobolev Spaces. Chapter Hölder spaces

Sobolev Spaces. Chapter Hölder spaces Chapter 2 Sobolev Spaces Sobolev spaces turn out often to be the proper setting in which to apply ideas of functional analysis to get information concerning partial differential equations. Here, we collect

More information

INDEX. Bolzano-Weierstrass theorem, for sequences, boundary points, bounded functions, 142 bounded sets, 42 43

INDEX. Bolzano-Weierstrass theorem, for sequences, boundary points, bounded functions, 142 bounded sets, 42 43 INDEX Abel s identity, 131 Abel s test, 131 132 Abel s theorem, 463 464 absolute convergence, 113 114 implication of conditional convergence, 114 absolute value, 7 reverse triangle inequality, 9 triangle

More information

Chapter 3: Baire category and open mapping theorems

Chapter 3: Baire category and open mapping theorems MA3421 2016 17 Chapter 3: Baire category and open mapping theorems A number of the major results rely on completeness via the Baire category theorem. 3.1 The Baire category theorem 3.1.1 Definition. A

More information