Scribd
Upload
335 views

Reed Simon2

Uploaded by

VALENTINA RAMÍREZ CIFUENTES
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
335 views

Reed Simon2

Uploaded by

VALENTINA RAMÍREZ CIFUENTES
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 370
METHODS OF MODERN MATHEMATICAL PHYSICS WW: FOURIER ANALYSIS, SELF-ADJOINTNESS MICHAEL REED BARRY SIMON Department of Mathematics Departments of Mathematics Bike Trina nd Pri Princeton Universty ACADEMIC PRESS, INC. Harcourt Brace Jovanovich, Publishers San Diego New York Berkeloy Boston London Sydney Tokyo Toronto Copvarcir © 1975, av Acapemic PREss, INC. PERNASSION IN WRITING FROM THE PUBLISHER.” ACADEMIC PRESS, INC. 1250 Sixth Avenue. Son Diego, California 92101 United Kingdom Edition published by ACADEMIC PRESS, INC. (LONDON) LTD, 14728 Oral Road, Loodon NWY Library of Congress Cataloging in Publication Data Reed, Michael ‘Methods of modern mathematical physies, Includes bibliographical references CONTENTS: ¥. 1, Functional analysi.~v.2, Four Jer analysis, seladjointness.— 1. "Mathematical physics. 1, Simon, Bary, joint author. I. Tile acz0.k37 1972 s30.1's 75-182650 ISBN 0-12-$85002-6(v.2) ‘AMS (MOS) 1970 Subject Clasifications: 42-02, 47-02 To our parents Helen and Gerald Reed Minnie and Hy Simon Preface This volume continues our series of texts devoted to functional analysis ‘methods in mathematical physics. In Volume I we announced a table of contents for Volume II. However, in the preparation of the material it became clear that we would be unable to treat the subject matter in sufficient depth in one volume. Thus, the volume contains Chapters IX and X; we expect that @ third volume will appear in the near future containing the rest of the material announced as “Analysis of Operators.” We hope to continue this series with an additional volume on algebraic methods. It gives us pleasure to thank many individuals: E. Nelson for a critical reading of Chapter X; W. Beckner, H. Kalf, R. S. Phillips, and A. S. Wightman for critically reading one or more sections. Numerous other colleagues for contributing valuable suggestions. F. Armstrong for typing most of the preliminary manuscript. J. Hagadorn, R. Israel, and R. Wolpert for helping us with the proof- reading. ‘Academic Press for its aid and patience; the National Science and Alfred P, Sloan Foundations for financial support. Jackie and Martha for their encouragement and understanding, ‘Mike ReeD Barry SIMON June 1975 Introduction A functional analyst i an analyst, frst and foremost, and nota degenerate species of topoogit E. Hille Most texts in functional analysis suffer from a serious defect that is shared to an extent by Volume T of Methods of Modern Mathematical Physics. ‘Namely, the subject is presented as an abstract, elegant corpus generally divorced from applications. Consequently, the students who learn from these texts are ignorant of the fact that almost all deep ideas in functional analysis have their immediate roots in “applications,” either to classical areas of analysis such as harmonic analysis or partial differential equations, or to another science, primarily physics. For example, it was classical electro magnetic potential theory that motivated Fredholm's work on integral equations and thereby the work of Hilbert, Schmidt, Weyl, and Riesz on the abstractions of Hilbert space and compact operator theory. And it was the impetus of quantum mechanics that led von Neumann to his development ‘of unbounded operators and later to his work on operator algebras. More deleterious than historical ignorance is the fact that students are too often misled into believing that the most profitable directions for research in functional analysis are the abstract ones. In our opinion, exactly the opposite is true. We do not mean to imply that abstraction has no role to play. Indeed, it has the critical role of taking an idea from a concrete situation and, by eliminating the extraneous notions, making the idea more easily understood as well as applicable to a broader range of x INTRODUCTION situations. But it is the study of specific applications and the consequent ‘generalizations that have been the more important, rather than the considera- tion of abstract questions about abstract objects for their own sake. This volume contains a mixture of abstract results and applications, while the next contains mainly applications. The intention is to offer the readers of the whole series a properly balanced view. We hope that this volume will serve several purposes: to provide an introduction for graduate students not previously acquainted with the ‘material, to serve as a reference for mathematical physicists already working in the field, and to provide an introduction to various advanced topics which are difficult to understand in the literature. Not all the techniques and applications are treated in the same depth. In general, we give a very thorough discussion of the mathematical techniques and applications in quantum mechanics, but provide only an introduction to the problems arising in quantum field theory, classical mechanics, and partial differential equations. Finally, some of the material developed in this volume will not find application until Volume III. For all these reasons, this volume contains a great variety of subject matter, To help the reader select which material is important for him, we have provided a “Reader's Guide” at the end of each chapter. As in Volume 1, each chapter contains a section of notes. The notes give references to the literature and sometimes extend the discussion in the text, Historical comments are always limited by the knowledge and prejudices of authors, but in mathematics that arises directly from applications, the problem of assigning credit is especially difficult. Typically, the history is in two stages: first a specific method (typically difficult, computational, and sometimes nonrigorous) is developed to handle a small class of problems. Later itis recognized that the method contains ideas which can be used to teat other problems, so the study of the method itself becomes important. ‘The ideas are then abstracted, studied on the abstract level, and the techniques systematized. With the newly developed machinery the original problem becomes an easy special case. In such a situation, it is often not ‘completely clear how many of the mathematical ideas were already contained in the original work. Further, how one assigns credit may depend on whether one first learned the technique in the old computational way or in the new easier but more abstract way. In such situations, we hope that the reader will treat the notes as an introduction to the literature and not as a judgment of the historical value of the contributions in the papers sited Each chapter ends with a set of problems. As in Volume I, parts of proofs are occasionally left to the problems to encourage the reader to Introduction xi participate in the development of the mathematics. Problems that fill gaps in the text are marked with a dagger. Difficult problems are marked with ‘an asterisk. We strongly urge students to work the problems since that is the best way to learn mathematics Contents Preface Introduction Contents of Other Volumes ‘THE FOURIER TRANSFORM The Fourier transform on (R*) and $'(R*), convolutions The range of the Fourter transform: Classical spaces ‘The range of the Fourier transform: Analytcity If Estimates pendix Abstract interpolation Fundamental solutions of partial differential equations with constant coefficients 6. Elliptic regularity 7. The free Hamiltonian for nonrelativistic quantum ‘mechanics 8. The Garding-Wightman axioms Appendix Lorentz invariant measures 9. Restriction to submanifolds 10. Products of distributions, wave front sets, and oscillatory integrals Notes Problems Reader's Guide wtaenn xii xv 15 a 32 45 9 54 on 2 16 87 108 120 133 xiv CONTENTS X: SELF-ADJOINTNESS AND THE EXISTENCE OF DYNAMICS 1. Extensions of symmetric operators 135 Appendix Motion on a half-line, limit point-limit circle methods 146 2. Perturbations of self-adjoint operators 162 3. Positivity and self-adjointness I: Quadratic forms 176 4. Positivity and self-adjointness 11: Pointwise positivity 182 5. The commutator theorem 191 6. Analytic vectors 200 7. Free quantum fields 207 Appendix. The Weyl relations for the free field 231 8. Semigroups and their generators 235 9. Hypercontractive semigroups 258 10. Graph Limits 268 HL. The Feynman-Kac formula 274 12, Time-dependent Hamiltonians 282 13, Classical nonlinear wave equations 293 14, The Hilbert space approach to classical mechanics 313 Notes 318 Problems 338 Reader's Guide 349 List of Symbols 353 Index 355 Contents of Other Volumes Volume I: Functional Analysis I u a Ww v vi vu vu Preliminaries Hilbert Spaces Banach Spaces Topological Spaces Locally Convex Spaces Bounded Operators The Spectral Theorem Unbounded Operators Volume Ill: Analysis of Operators XI XI XI Perturbations of Point Spectra Scattering Theory Spectral Analysis Later Volumes xIV xv XVI XV XVII XIX XX Group Representations Commutative Banach Algebras Convex Sets The GNS Construction Von Neumann Algebras Applications to Quantum Field Theory Applications to Statistical Mechanics IX: The Fourier Transform We have therefore the equation of condition Fx) = [da Q cos ax Uf we substituted for Q any faction of fy and conducted the integration from 4 = 0 10 4-= &, Wwe shoul find a function of x: it required to solve the imverse probiem, that is t0 Say. to ‘ascertain what faction of, after being substituted for Q. gives as a result the function F(x), ‘remarkable problem whose solution demands attentive examination. Joseph Fourier (XA The Fourier transform on (ft?) and (R),convolutions ‘The Fourier transform is an important tool of both classical and modern analysis. We begin by defining it, and the inverse transform, on (R"), the Schwartz space of C* functions of rapid decrease. Definition Suppose fe “(R"). The Fourier transform of fis the function Feiven by [erry ax ay where x+4 = Diss x4. The inverse Fourier transform of /, denoted by J. is the function F@)= em hf (x) dx 1 orl, We will occasionally write f = Ff. 21% THE FOURIER TRANSFORM Since every function in Schwartz space is in L'(R"), the above integrals make sense. Many authors begin by discussing the Fourier transform on LR), We start with Schwartz space for two reasons: First, the Fourier transform isa one-to-one map of Schwartz space onto itself (Theorem IX.1). ‘This makes it particularly easy to talk about the inverse Fourier transform, which of course turns out to be the inverse map. That is, on Schwartz space, it is possible to deal with the transform and the inverse transform on an equal footing. Though this is also true for the Fourier transform on E(R*) (see Theorem IX.6}, itis not possible to define the Fourier transform on E(B") directly by the integral formula since [2(R*) functions may not be in L!(R"};a limiting procedure must be used. Secondly, once we know that the Fourier transform is a one-to-one, bounded map of 7(R") onto SIR"), we can easily extend it to (R*). Its this extension that is funda~ mental to the applications in Sections 5, 6, and 8. We will use the standard multi-index notation. A multi-index eh ooo ad is an n-tuple of nonnegative integers. The collection of all multi-indices will be denoted by I", . The symbols |a|, x*, D*, and x? are defined as follows: lel = Yaw pera on P= Bei Oe veda In preparation for the proof that ~ and ~ are inverses, we prove: Lemma The maps ~ and ~ are continuous linear transformations of (Bt) into A(R"), Furthermore, if « and f are multi-indices, then ((aypD7\(2) = By) (xy Proof The map ~ is clearly linear. Since (@eD7\2) = ar [Be mPe 00) a 1 "al. F (Dye *)(~ ix pf (x) dx. Tap le PHBE) de 1X4 The Fot transform on (R") and Re), convolutions 3 We conctude that Ula = UPTV Sea [ IDE] dv < 2 so takes (R*) into ¥(R*), a i k is large enough, f (I + x?) lees pa Sa Owe a ‘we have also proven (IX.1). Furthermore, dx < @ so that 5 gpa (J 1+ 1 *aspupte + xy Dg inPen Using Leibnitz’s rule we easily conclude that there exist multi-indices 2,. B, and constants c, so that WhesZolben Thus, ” is bounded and by Theorem V.4 is therefore continuous. The proof for ~ is the same. We are now ready to prove the Fourier inversion theorem. The proof we give uses the original idea of Fourier. ‘Theorem IX.4 (Fourier inversion theorem) The Fourier transform is a linear bicontinuous bijection from $/(R") onto (R"). Its inverse map is the inverse Fourier transform, ic. f= f= Proof We will prove that f= f. The proof that that ” is surjective and j= implies that ~ is injective. Since ~ and“ are continuous maps of £(R") into (R"), it is sufficient to prove that f= f for f contained in the dense set C5(R"). Let C, be the cube of volume (2/2) centered at the origin in R*. Choose ¢ small enough so that the support of fis contained in C,. Let K, = (ke Reach k/ne is an integer) Then Fe) = Z.deare * Adenres* is just the Fourier series of f which converges uniformly in C, to fsince fis continuously differentiable (Theorem 11.8). Thus 105) = 5, OEY aap x2) ‘THE FOURIER TRANSFORM Since Bis the disjoint union of the cubes of volume (nc)* centered about the points in K,., the right-hand side of (IX.2) is just a Riemann sum for the integral ofthe function f(kje" */(2n)"?, By the lemma, f(k)e* *e F(R"), so the Riemann sums converge to the integral. Thus f= f. § Corollary Suppose fe (R"). Then [ rO0P dx = [IPQ ak J J Proof This is really a corollary of the proof rather than the statement of ‘Theorem IX.1. Iff has compact support, then for ¢ small enough, Fle) = 5 berets, spidered* Since {(}e)"?e"'*}x¢ x, is an orthonormal basis for Z(C,), Jp FO0P ax = f 1 G)P ax = E,ldertet ss rope = Frey mee" f POOP de This proves the corollary for fe C8. Since ~ and ||: |} are continuous on ¥ and C8 is dense, the result holds for all of . Example 1 We compute the Fourier transform of f(x) = where a > 0. ere 9(R) ortega dy fa)= DXA The Fourier transform on 7(m) and (2), ¢0nv ‘The next to last step follows from the Cauchy integral formula and the exponential decrease of e~*" along lines parallel to the x axis, We now define the Fourier transform on (Rt). Definition Let T.< #'(R*). Then the Fourier transform of T; denoted by T, is the tempered distribution defined by T(9) = T(@). ‘Suppose that h, ye Y(R*), then by the polarization identity and the corollary to Theorem IX.1 we have (h, @) = (f, 6). Substituting we obtain To) = | alx)o(s) dx = | a(x\Olx) dx = 7,6) = Flo) where T; and 7, are the distributions corresponding to the functions § and g respectively. This shows that the Fourier transform on (Rt) extends the transform we previously defined on $(R"). Theorem 1X.2__ The Fourier transform is a one-to-one linear bijection from #'(R") to.(R*) which is the unique weakly continuous extension of the Fourier transform on ¥(R) Proof If, %9,then by Theorem IX.1, @, % 6,80 T(é,) + T() for each T in #(R"). Thus T(g,) > T(9), which shows that 7 is a continuous linear functional on $(R*). Furthermore, if TT, then TT because T,(G) ~ 16) implies 1,(9) + 7(p). Thus Ts 7 is weakly continuous. The remaining properties of ~ follow immediately from the correspond- ing statements on $(R") (See Problem 19 in Chapter V). § Example 2 We compute the Fourier transform of the derivative of the delta function at be ®: 5(e) = 56) aan feats as) =f (“ee ax So, the Fourier transform of 4; is the function ixe~*/,/2n. ‘THE FOURIER TRANSFORM ‘We now introduce a new operation on functions. Definitions Suppose that fg ¢.(R"). Then the convolution of fand g, denoted by f« g, is the function (f+ 0) =f $0 — sale) dx ‘The convolution arises in many circumstances (we have already used it in discussing closed operators in Section VIIL.1). In Section 4 we use interpola- tion theorems to prove I? estimates on the convolution fg in terms of ‘fand g, In this section we concentrate on the properties of the convolution ‘as a map from (R*) x S°(R) to S(R). Using these properties we show that the convolution can be extended to a map from ¥(R") x 7(R) to O%,, the polynomially bounded C* functions. Convolutions frequently ‘occur when one uses the Fourier transform because the Fourier transform takes products into convolutions (Theorem IX.3b and Theorem IX.4c). Theorem 1X.3 (a) For each fe ¥(R"), g+f+g is a continuous map of ¥(R") into (R”). (b) = 2n)""fs g and Fog = (2a. (©) Forfa h in A(R") f+ g=9sSand f+ (gh) = (f +g) +h Proof From the polarization identity and the corollary to Theorem IX.1 we find that (9, ¥) = (6, 8) for @, ve ¥(R"). Letting ye R" be xed, we apply this identity to e”-J(x) and g obtaining (€”"%, g) = (@ F4). But (e”Fay= fe Fxlw) ax and — (ec *f.9) (er fe" stirs FG) ae a =f, Fo ~ Malay da which proves that fg = (2n)""?f» g. Using the inverse Fourier transform this formula may be stated as (ny"Fi=se9 1X1 The Fourier transform on /(f") and.(R"),convolutions 7 This shows that convolution is the composition of the inverse Fourier transform, multiplication by (2x)°F, and the Fourier transform. It follows that convolution is continuous. The statements in (c) follow trivially from (b). In order to extend the map C,:g~+/+g to ", we look for a continuous map C,: 4 + sothat C; | Y= Cy. We then define C; to be convolution ony. Definition Suppose that fe (Bt), T¢.4'(R*) and let T(x) denote the function, f(—x), Then, the convolution of T and f, denoted T « fis the distribution in "(R") given by (T+ sXe) =T(7 +9) for all pe F(R") The fact that g+7s g is a continuous transformation guarantees that T + fe. ¥'(R"), The following theorem summarizes the properties of this extended convolution Let f, denote the function /,(x) = f(x ~ y) and , the function f(y ~ x). When fis given by a large expression (--), we will sometimes write (-:)” rather than (7). Theorem IX.4 For each fe ¥(R") the map T+T +f is a weakly continuous map of #(R*) into .#"(R") which extends the convolution on (R"), Furthermore, (2) T+ fis a polynomially bounded C* function, ie. T + fe Ojy. In fact, (7 + f(y) = TUR) and DAT + f)=(D°T)« f= T+ DY (1X3) (b) (TS) =T+(fe9) () TF =Qayp7t Proof Since T ~+ T + fis defined as the adjoint of a bounded map from 7 to F, it is automatically weakly continuous. The fact that it extends the convolution on # is just a change of variables. The statements (IX.3), (b), and (c), all follow immediately from the corresponding statements for T €.F and the facts that is weakly dense in” and that ¥. D®. multiplication by f and convolution are all weakly continuous on 8 1X: THE FOURIER TRANSFORM It remains to prove the first part of (a). Since Te (BR) it follows from the regularity theorem (Theorem V.10) that there is a bounded continuous function k, a positive integer r, and a multi-index f so that TU) = JM + PY (DENY ~ x) dx Since D'fe ¥, T(],) is an infinitely diferentiable function of y. The change of variables t = y — x shows that ITH) s thle [+ 2 Y1DN ~~} dx = thle f (1+ 0 PYIDF (OL de from which it follows casily that y-+T(J,) is polynomially bounded. A similar proof works for the derivatives of yr T(7,). Thus T(7) € Of, Suppose that a distribution Se #'(R") is given by a polynomially bounded continuous function s. Then, using Fubini’s theorem we find that for ge F(R) (S+ Ao) = S(7+ 0) =f si{J7ee— yoo) ay) a =| ({seteras}oores = (910) so $ « f= S(J,). By the regularity theorem T = D*S for some such S. Thus by (1X3) Taf =(DS)*f This completes the proof. 1X2 The range of the Fourler transform: Classical spaces 9 Theorem 1X5 Let Te¥(R") and fe ¥(R"). Then fT e Oy and Fk) = ny"? T(fe-*"*), In particular, if T has compact support and ye F(R’) is identically one on a neighborhood of the support of T, then T(k) = (2m)? Te“#"*) Proof By Theorem IX.4c and the Fourier inversion formula we have FF = Qn) "fet Thus fT 0% and FFW) = Ory FH) = Qayer(e* sf) H ‘We remark that one can also define the convolution of a distribution T €9'(R") with an fe 9(R") by (T « f)(y) = TU). A proof similar to the proof of Theorem 1X.4 shows that T * fis a (not necessarily polynomially bounded) C* function and that (IX.3) holds. ‘We have already introduced the term “approximate identity” in Section VIILI; we now define it formally. Definition Let j(x) be a positive C* function whose support lies in the sphere of radius one about the origin in Rt" and which satisfies { (x) dx = 1 The sequence of functions j(x) = €°*j(x/e) is called an approximate identity. Proposition Suppose T € 7"(R*) and let j,(x) be an approximate iden- tity. Then T + j,> Tweakly as +0. Proof Ihe ¥(R%), then (T + j,\(o) = Tj. + 9) 0 it is suficient to show that j, L149. To do this it is sufficient to show that (2n)*3j,6 % 6. Since j,(2) = jled) and j(0) = (2n)-"?, it follows that (22)"3),(x) converges to 1 uniformly on compact sets and is uniformly bounded. Similarly, Yj, converges uniformly to zero. We conclude that (2n)"3), 40. 1X2. The range of the Fourier transform: Classical spaces We have defined the Fourier transform on (R") and '(@"), In this section, Section IX.3, and Section IX9, we investigate the range of the Fourier transform when it is restricted to various subsets of °"(R"). These 40 IX: THE FOURIER TRANSFORM, ‘questions are natural and have historical interest, but more important, characterizing the range of the Fourier transform is very useful. One is often able to obtain information about the Fourier transform of a function and one would like to know what this says about the function itself. We begin with two theorems which follow easily from the work that we have already done in Section IX.1. ‘Theorem IX.6 (the Plancherel theorem) ‘The Fourier transform extends uniquely to a unitary map of (0) onto 12(R"). The inverse transform extends uniquely to its adjoint Proof The corollary to Theorem IX.1 states that if fe7(R*), then Mla = [7 lz- Since ¥[F] =, # is a surjective isometry on 1°(R"). Theorem IX.7 (the Riemann-Lebesgue lemma) The Fourier transform extends uniquely to a bounded map from L'(*) into Cg(R"), the continuous functions vanishing at oo. Proof For fe. #(R"), we know that j¢ (R*) and thus 7 ¢ C,(R"). The estimate ies Qa" 1S Is is trivial. The Fourier transform is thus a bounded linear map from a dense set of L(R") into C,(R"). By the BLT. theorem, ~ extends uniquely to a bounded linear transformation of 1!(R*) into C,(R*). We remark that the Fourier transform takes L!(R") into, but not onto C,.(R") (Problem 16). ‘A simple argument with test functions shows that the extended transform on L(R*) and 2(R') is the restriction of the transform on "(R"), but it is useful to have an explicit integral representation. For f ¢ L'(R"), this is easy since we can find f,¢4(R*) so that | f — fall, 0. Then, for each 4, Fa) tim (7,(4)) sin . = aapel, eo Fee fale) as} 1X2 The range of the Fourier transtorm: sical spaces tt So, the Fourier transform of a function in L'(R") is given by the usual formula Next, suppose fe 2(R*) and let _ fl [xlsR ald= ER ‘Then xz feLi(R*) and xz, f—“+f, so by the Plancherel theorem aI —picth For xq f we have the usual formula; thus Fay Lim (2x) my, oe f(x) dx where by “Lim.” we mean the limit in the L?-norm. Sometimes we will dispense with |x| LR") and L'(R") > 1"(R") and in both cases " is a bounded operator. It is exactly in situations like this that one can use the interpolation theorems which we will prove in the ‘Appendix to Section 4 Theorem IX.8 (Hausdorff-Young inequality) Suppose 1 Titi) = TG) = Gay u(UiP) = Gay | Lica duce) Wh ‘On each compact subset of R*, j(x) converges uniformly to (2n)"*? as £70, So the pemeasure of any compact set is less than (2n}"*|f]., so nis finite We now come to the interesting point. Since y is finite, its Fourier transform is a continuous function of positive type. Since 4 and f must coincide ae, we have proven: Proposition A bounded function of weak positive type is equal almost everywhere to a continuous function of positive type. 1X3. The range of the Fourier transform: Analyticity In this section we investigate the connection between the decay properties of a function or distribution at infinity and the analyticity properties of its Fourier transform. The most extreme form of decay at infinity is to have compact support. We will prove the Paley-Wiener and Schwartz theorems which characterize explicitly the Fourier transforms of C* functions and distributions with compact support. We then state two theorems relating ‘exponential decay to analyticity properties of the Fourier transform, We close the section by characterizing the Fourier transforms of tempered distributions whose supports lie in symmetric cones. There are many other theorems of this genre; some of them are discussed in the Notes. Suppose that fe Cg (R"). Then for all C=

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy