Scribd Logo
Upload

Welcome to Scribd!

  • 18 views

    A Stability Theorem

    Uploaded by

    ykw kcckc
    This article presents a simplified proof of Walter Rudin's 1968 stability theorem about zeros of holomorphic functions in a polydisc. The theorem states that if a continuous map φ from the open unit disc U to the polydisc Un carries the boundary torus Tn into itself, and the index of each component φj is positive, then any holomorphic function f that is non-zero on the boundary union image set must be non-zero everywhere in Un. The new proof is shorter and relies only on properties of the index or winding number of a plane curve around the origin.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    A Stability Theorem

    Uploaded by

    ykw kcckc
    18 views4 pages
    This article presents a simplified proof of Walter Rudin's 1968 stability theorem about zeros of holomorphic functions in a polydisc. The theorem states that if a continuous map φ from the open unit disc U to the polydisc Un carries the boundary torus Tn into itself, and the index of each component φj is positive, then any holomorphic function f that is non-zero on the boundary union image set must be non-zero everywhere in Un. The new proof is shorter and relies only on properties of the index or winding number of a plane curve around the origin.

    Original Description:

    Original Title

    A stability theorem

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This article presents a simplified proof of Walter Rudin's 1968 stability theorem about zeros of holomorphic functions in a polydisc. The theorem states that if a continuous map φ from the open unit disc U to the polydisc Un carries the boundary torus Tn into itself, and the index of each component φj is positive, then any holomorphic function f that is non-zero on the boundary union image set must be non-zero everywhere in Un. The new proof is shorter and relies only on properties of the index or winding number of a plane curve around the origin.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    18 views4 pages

    A Stability Theorem

    Uploaded by

    ykw kcckc
    This article presents a simplified proof of Walter Rudin's 1968 stability theorem about zeros of holomorphic functions in a polydisc. The theorem states that if a continuous map φ from the open unit disc U to the polydisc Un carries the boundary torus Tn into itself, and the index of each component φj is positive, then any holomorphic function f that is non-zero on the boundary union image set must be non-zero everywhere in Un. The new proof is shorter and relies only on properties of the index or winding number of a plane curve around the origin.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 4

    The American Mathematical Monthly

    ISSN: 0002-9890 (Print) 1930-0972 (Online) Journal homepage: https://www.tandfonline.com/loi/uamm20

    A Stability Theorem

    Walter Rudin

    To cite this article: Walter Rudin (1999) A Stability Theorem, The American Mathematical Monthly,
    106:8, 768-770, DOI: 10.1080/00029890.1999.12005117

    To link to this article: https://doi.org/10.1080/00029890.1999.12005117

    Published online: 23 Apr 2018.

    Submit your article to this journal

    Article views: 15

    View related articles

    Full Terms & Conditions of access and use can be found at


    https://www.tandfonline.com/action/journalInformation?journalCode=uamm20
    Notice that E{X 3} = 0 since X is symmetric. This extends the first property of
    Lanczos' derivative.
    The second property of Lanczos' derivative generalizes to Lu in a similar
    fashion. Assume that both fk(x 0 ) and f{(x 0 ) exist. Then
    Luf(x 0 ) = E{X 2 duxf(x 0 ): X> 0} + E{X 2 duxf(x 0 ): X< 0}
    = E{X 2 fk(x 0 ) + X 2o(1): X> 0} + E{X 2 f{(x 0 ) + X 2o(1): X< 0}
    = E{X 2 fk(x 0 ): X> 0} + E{X 2 f{(x 0 ): X< 0} + o(1)
    = fk(x 0 )E{X 2 : X> 0} + f{(x 0 )E{X 2 : X< 0} + o(1)
    fk( Xo) + !{( Xo)
    = + o(1).
    2
    In the last step, we have applied the symmetry condition and E{X 2} = 1. The
    roles of fk and f{ are seen clearly from these five lines.
    Finally, notice that: (1) If f(x) is Lipschitz continuous at x 0 with L as its
    Lipschitz constant, then 1Luf(x 0 )i ~ L; (2) The random variable involved can be
    replaced by any suitable distribution with a compact support, since we have not
    used the positivity condition.

    REFERENCE

    1. C. W. Groetsch, Lanczos' generalized derivative, Amer. Math. Monthly 105 (1998) 320-326.

    Computational and Applied Mathematics, UCLA, 7354 Math Sciences Building, Los Angeles, CA 90095
    jhshen@math. uc/a.edu

    A Stability Theorem
    Walter Rudin

    In 1968 I proved a theorem (stated below) about zeros of holomorphic functions in


    a polydisc [2, p. 87] which was later, in [1], referred to, much to my surprise, as a
    "cornerstone" of rr.ultivariable stability theory. The authors of [1] pointed out,
    quite correctly, that my proof used quite a bit of homotopy theory, and they
    proceeded to prove the theorem by a sequence of more elementary steps. The
    present note contains an even easier proof, which is also much shorter, and which
    relies only on very simple properties of the index (or winding number) of a plane
    curve around the origin.
    The following notation will be used. C is the complex plane, C* = C \ {0} is the
    set of all nonzero complex numbers, U and U are the open and closed unit discs in
    C, respectively, and T is the unit circle. For n ~ 1,
    en= C X··· XC, un = U X··· XU, Tn = T X··· X T;
    each of these cartesian products has n factors. The torus Tn is the so-called
    distinguished boundary of un; it is a small (n-dimensional) part of the whole
    (2n -!)-dimensional boundary of the polydisc un.
    A(Un) is the class of all continuous f: un ~ C that are holomorphic in un.

    768 NOTES [Monthly 106


    If now f: [0, 27T] ~ C* is continuous and f(27T) = f(O) (so that f([O, 27T]) is a
    closed curve in C*) then there exists a continuous real-valued function a on
    [0, 27T] such that
    f(O)=If(O)Iexp{27Tia(O)} (0:::;0:::;27T). (1)
    Since f(27T) = f(O), a(27T) - a(O) is an integer (positive, negative, or 0). This is
    the index of r:
    Indf = a(27T)- a(O). (2)
    Note that Ind f is independent of the particular choice of a.)
    We need the following properties of the index.
    (I) Suppose (s, 0) ~ fs(O) is a continuous map from [0, 1] X [0,27T] into C*,
    and f,(27T) = f/0) for all s. Then Ind rs is the same for all s.
    The reason is simply that Ind fs is a continuous function of s. Being integer-
    valued, this function is constant on the connected set [0, 1].
    (II) If G : U ~ C* is continuous and if we define G IT( 0) = G(ei 8 XO :::; 0 :::; 27T)
    then Ind GIT = 0.
    To deduce this from (I) put rs(O) = G(sei 8 ) and note that rl = GIT, fo is the
    constant G(O).
    (III) If h E A(U) and h(T) c C* then Ind hI r is equal to the number of zeros of h
    in U.
    This is the classical "argument principle" of complex analysis.

    Theorem. Suppose <I> = (cp~> ... , cpn) is a continuous map of U into un that carries T
    into T n, such that
    Ind cp1IT > 0 for 1 :::; j :::; n (3)
    Put K = <I>(U). Then

    f(P u K) = f(Un) ( 4)
    for every f E A(Un).

    Proof: Assume f(z)-=!= 0 for every z E P u K. We show that f(z) -=!= 0 for every
    z E un. This implies the theorem, and shows why the term "stability" was used in
    this connection.
    Fix a= (al, ... ' an) E un. Let IndcpjiT =mi. There exist cj E c such that
    cji=a1 (1:::;j:::;n). (5)
    Since m1 > 0, lc1 1:::; 1. Define

    (6)

    for A E U. Then h E A(U), h(T) c C*, h(O) = f(a). Hence f(a)-=!= 0 follows from

    Ind hiT= 0 (7)


    because of (III).
    Since (f o <I> XU) = f(K) c C*, by assumption, (II) shows that
    lnd f o <I> IT= 0. (8)

    October 1999] NOTES 769


    There are continuous real-valued functions aj, {3j such that

    on [0, 27T ]. Note that

    (9)

    Define

    and let 'l's: [0,27T] ~ Tn be the map whose /h component is exp{27Ti1'j,s(O)}.


    Then

    'l's{27T) = 'l's(O) {0 ~ S ~ 1), {11)

    'l's(T) c Tn, hence f('l's(T)) c C*, and now (I) shows that

    Ind f o '1'1 = Ind f o '1'0 • (12)

    Since f o <I> IT = f o '1'1 and hIT = f o '1'0 , (12) and (8) imply (7).
    •
    Remarks. (i) The simplest example of a <I> as in the theorem is <I>( A) = (A, A, ... , A).
    Then K is a disc (2-dimensional), dim (P u K) = n, whereas dim un = 2n.
    (ii) It is not necessary for <I> to map U into the interior un of un. For example,
    when n = 2,

    9
    ·n ((2rei ,0) (0 ~ r ~ 1/2)
    <l>(re'") =
    ( e; 8 , 2r- 1) (1/2 ~ r ~ 1)

    will do nicely.
    (iii) The hypott:!_esis "mj > 0 for all j" cannot be omitted. To see this, take
    n = 2, <I>(A) =(A, A). Then m 1 = 1, m 2 = -1. If f(z, w) = 1 + 4zw, then 1/1 ~ 1
    on T 2 U <I>(U) but f(t, - t) = 0.
    For another example, take <I>(A) =(A, 1), so that m 1 = 1, m 2 = 0, and put
    f(z, w) = 2w - 1. Then 1!1 ~ 1 on T 2 U <I>(U) but f(z, 1/2) = 0 for all z.
    However, the hypothesis "mj > 0 for all j" can b~ replaced by "m. < 0 for
    all j" because the theorem can then be applied to <I>( A) in place of <I>( A).

    REFERENCES

    1. Ph. Delsarte, Y. Genin, Y. Kamp, A simple proof of Rudin's multivariable stability theorem, Philips
    Research Laboratory Report R412, Brussels, Dec. 1979.
    2. W. Rudin, Function Theory in Polydiscs, Benjamin, 1969.

    University of Wisconsin-Madison, Madison, Wisconsin 53706


    recep@math. wisc.edu

    770 NOTES [Monthly 106

    You might also like