Extremal properties characterizing weakly l-valent principal functions by Dennis Evo Garoutte A thesis submitted to the Graduate Faculty in partial fulfillment of the requirements for the degree DOCTOR OF PHILOSOPHY in Mathematics of Montana State University © Copyright by Dennis Evo Garoutte (1966) Abstract: Extremal properties, characterizing weakly λ-valent principal functions F0^ and F1^ defined on planar bordered Riemann surfaces, are developed. The λ-valent principal functions are constructed by using Sario’s linear operator method. The extremal properties derived are analogous, in a rather natural way, to those of the univalent principal functions F0 and F1 of Sario. Such extremal properties characterize the λ-valent principal functions in a class much larger than the class of weakly λ-valent functions. The class consists of functions λ-valent near the border and near a fixed interior point, with the valence being arbitrary elsewhere. The principal function F0^λ is shown to be the λ-th power of the univalent function F0. The function F0^λ then has the geometric property of being a weakly λ-valent radial slit disk mapping. The functional extremized by F1^λ is slightly different than the expected one. It does reduce to the expected one, however, when the class of competing functions is taken to be a smaller class consisting only of properly normalized λ-th powers of univalent functions. , I1 1 j: EXTREMAL PROPERTIES CHARACTERIZING WEAKLY X-VALENT PRINCIPAL FUNCTIONS DENNIS EVO GAROUTTE ■! A thesis submitted to the Graduate Ihculty in partial fulfillment of the requirements for the degree i i ' DOCTOR OF PHILOSOPHY! Mathematics Approved: ad, Major Department < ? - 9 u Chairman, Examining Committee Ii ^ G r a d u a t e Dean ” I . i I MONTANA STATE! UNIVERSITY Bozeman, Montana I ! i ; December, 19 6 6 ■ I / .ii 1' 1 I I iii ACKNOWLEDGMENT The author wishes to thank his thesis advisor, Dr. Paul Nickel, ; 1 for his proposal of the thesis topic, a nd for his assistance during its solution. ' i '! i I i: ■I I - ■ , I ! I ;I sI Mi I TABLE O F ,COETEHTS Chapter Page INTRODUCTION I. 'PRELIMINARIES § I. S . 2. § II. 3« § 3* g 6. Basic Properties; of Rieinann !surfaces . . Normal Linear Operators,' on a Riemann Surface . . . . . . The Reduction Theorem i. . . . L- . .j . . ; . . . . § Approximating Surfaces . . . . Extremal Properties of:Harmbnic Functions on Approximating Surfaces . . . . . . Extremal Properties of Analytic Functions on Approximating Surfaces . . . . . . . J. Extremal Properties . Principal Harmonic 8 . Extremal Properties Principal Analytic 9* §10. X-th . . . X-th . . . The Extremal Property,of the Second X-th Principal Harmonic Function . . . . . . Extremal Properties of the;Second X-th Principal Analytic' F u n c t i o n ...........■ LITERATURE CITED ' W of; the'First Function . o f the First Function . EXTREMAL PROPERTIES OF THE SECOND X - T H PRINCIPAL HARMONIC A N D ANALYTIC FUNCTIONS 1 . . . . . . . g 4 4 6 9 11 Construction of Harmonic Functions on EXTREMAL PROPERTIES OF THE FIRST X-TH PRINCIPAL HARMONIC A N D ANALYTIC FUNCTIONS .............. § I V .1 . j. . .j........... EXTREMAL PROPERTIES OF 'HARMONIC A N D ANALYTIC FUNCTIONS ON APPROXIMATING SURFACES . . . . . • I ' :: ' 1 . § '4. III. . . . . . . . . . . . . ........... 11 12 16 19 19 23 28 28 34 36 V ABSTRACT Extremal pr operties , characterizing weakly A-valent principal functions F 0 a n d F^ defined on planar Bordered Riemann surfaces, are developed. The X-vulent principal functions are constructed by using S a r i o 1s linear operator method. The extremal properties derived are a n a logous, in a rather natural way, to those of the univalent prin c i ­ pal functions F 0 and F-] of S a r i o . Such extremal properties characterize the X-valent principal functions in a class much larger than the class of weakly X-valent functions. . The class consists of functions X-valent near the border and near a fixed interior point, with the valence being arbit­ rary elsewhere. The principal function F 0 is shown to b e the X-th power of the univalent function F 0 . The function Fg then has the geometric property of b eing a w eakly X-valent radial slit disk mapping. The functional extremized b y F^ is slightly different than the expected one. It does reduce to the expected one, however, when the class of competing functions is taken to be a smaller' class consisting only of properly normalized X-th powers of univalent f u n ctions. INTRODUCTION L. Sario has used linear operators to establish the existence of certain univalent canonical mappings of planar bordered Riemann surfaces W onto slit disks [ 6 ] . These mappings F q and F^-, called principal analytic functions, are foraied from principal harmonic functions, which themselves are constructed by applying the linear operator method of [ 5 / 2 • The singularity function s, wh i ch plays a central role in the linear operator method, is defined near the border y- as the flux function s ^ , 2 rc which is constant on y w i t h flux point surface I it is equal to log \z there, while near a fixed interior - £ |'. By exhausting the planar bordered W, one constructs the mappings Fq and F of W onto a slit plane disk, w i t h radial or circular slits, possibly degenerate, removed. In this investigation, we study the extremal properties of k-th principal functions s is taken to be Fq S^ and near F^ T to be k-th X Fq x and powers by F^ 2jtX log . jz - g | near £ . Here s^ In a manner similar to [ 6 ] the are constructed. These maps were first shown Nickel ['4,] and hence their geometric properties are completely understood. mappings. X and is constant on T with a flux of canonical maps which result when the singularity function Namely, they are weakly k-valent slit disk In a manner different from [ 4 ], it will be shown b y using derived extr'emal properties that similar property for F^ F^ is the k-th power of •F q . The is not proved here because it is needed to I i I derive a n extremal property for F i ,* Chapter I is introductory b y !nature,! and !contains important theorems and definitions used in the sequel. ; . j : ' : I ■: ■ ' _ In Claapter II, we consider a 'planar bordered Riemann surface W of infinite connectivity. border c o m p m e n t sion. T It is assumed,that The surface W is approximated by an!increasing sequence of compact containing T and are constructed. , F W £ . F W n ! , each of;finite connectivity and each : : : x the analytic functions F ; ; j :| ! , on ' x and Among a class of normalized analytic functions Fin F of maximizes the functional on log r(F) + ' Aj (F) : minimizes in 2jt X Here f nj On : 2rtX and I i- j £ e Int Wj to be held fixed in our discus­ and a point bordered Riemann surfaces W has one isolated, compact, log r(F) - A (Fy I 2L _ i V log IF ■ I ; jj represents the radius of the image :of T r(F) is the complementary logarithmic area. !i d arg F. under F and A(F) ; : ij jIn Chapter III it is then shown :b y using the, Reduction Theorem Ls in the limit for the limit (Sario [ 8 ]) that the extremal property holds function F X o ,X that F “ is the A deviation formula is. derived, from w h i c h it is established ; U' X-th po w e r of the univalent, ,function geometric properties of F is a w eakly X Fq ' i! m a y be inferred from those of X -valent radial slit, disk mapping. :r. ties of F o Fq are also derived, :i. -i . I I iiI I I : .1 of [ 6 ]. Fq . The In fact, Other extremal proper- -3.I It is shown in Chapter IV that the extremal property holds in }y. the limit for the limit function . ' S a r i o 1s Reduction Theorem could not be applied and an argument based on the convergence of certain Dirichlet integrals is used. established. One additional ,extremal property is also : , :: Such extremal properties characterize the 1-valent principal I functions in a m u c h larger class of functions. Namely, this class of functions consists of functions w h i c h are I X-valent near T and ; are no further restrictions. I ; : i . There CHAPTER I PRELIMINARIES Important definitions and theorems used throughout this discourse are listed in this chapter. g I. Basic properties of Riemann surfaces. Def i n i t i o n . A Riemann surface (bordered Riemann surface) is a connected Hausdorff topological space open sets O and a family W together with a covering of mappings cp V by satisfying the following pr o p e r t i e s : I. Each cp e $ of the complex plane 2. Every 3. If is a homeomorphism of an onto an open s e t ' (closed upper half plane). O e V O O e V is the domain for some and Og are the domains of is a directly conformal mapping of cp (o cp e $ . Cp^ and cp^ then Cp^ ° cp_J fl O g ) . We w i l l denote a bordered Riemann surface by W and W will be its interior, which is itself a Riemann surface. D e f inition. The border of a bordered Riemann surface set of points belonging to W W is that wh ich is mapped into the real axis by some q> e f . A bordered Riemann surface W is known to be orientable, and hence the border has a positive as well as a negative direction. Definition. Surface W. C C be an arbitrary Jordan curve on a Riemann The surface is called planar if whenever each a neighborhood then Let W - C N such that if R - C Def i ni t i o n . called harmonic W has is composed of two components, is composed of at least two components-. locally separates p on C then it also separates That is to say, if W. A real-valued (complex-valued) function f ° <p 1 (a n a l y t i c ) if the functions f on W is a r e , where they are defined, harmonic (analytic) in the usual sense, for all 9 e j|. Definition. "{^nj" A n exhaustion of a Riemann surface W is a sequence comPac^ bordered Riemann surfaces satisfying: 2. The boundary of W consists of a finite number of disjoint analytic Jordan c u r v e s . 3. E a c h 'component of 4. W. = The sets Let surface W u U w W - W^ is relatively non-compact. n W^ w i l l be referred to as "approximating surfaces." and v be harmonic functions on a compact bordered Riemann Then Green's formula m ay be expressed as (I) W where • B(W) is the border of .W and dux- / is the' conjugate differential. -6The left side of (l) is the mi x e d Dirichlet integral for the surface and is denoted by D^( u, v ) . G r e e n ’s formula m ay also be expressed as B(W) W For functions B ( W )} pdp* W vdu * * p, harmonic on is understood as 0. = Int W, but possibly not defined on Iim the border of an approximating surface ^ ^ pdp*. Here denotes and is homologous to . We shall be concerned w i t h an important'solution of the Dirichlet boundary value problem for a Eiemann su r f a c e . Such is called a flux function a n d ■is defined in the following manner. component of a compact bordered Riemann surface o' let be a 'curve contained in-the homologous to In w W (a = 0 ) on a let a , § of be a border In addition, W ' whi c h o and o' is Denote the region bounded b y w solve the following boundary value problem; namely, w = I s^ in W . a . on o' , function for a neighborhood of Evidently, interior Let = has flux a and w is harmonic in by 'U(a ). W( a ). A flux is defined as: (2 rtX / dw*) w. 2 % X , is constant on o', 0 on a , and harmonic W( a ) . ’ 2. Formal linear operators on a Riemann surface. Definition. if (l) G is a regular region of a n open Riemann surface G is relatively compact, (2) G and its exterior have a common boundary which is a I -dimensional analytic submanifold, a nd (5 ) all W components of W - G - 7 -: . I • are non-compact,. I N o w let ji . K be a n y compact set! contained in W and let u be a i 'j function harmonic on its complement. 'It is possible to enclose K .with a regular region 1 .1 1 ' with a positively oriented boundary P . G ' r u w i l l be denoted b y of G i ' /R d u *. The flux of ii That this is independent of the choice ■ i . :l : f is a consequence of G r e e n ’s f ormula. !As described above, P will I . 1I I:i be said to represent the ideal boundary of W,i. The border of a compact • ' — bordered Riemann surface of W, while W 1 I :. I! trivially represents the ideal boundary Uo , where P . ■ I a r e 'the border components of a n exhausting ii 1 ' ■ W n , represents the ideal boundary:of the R i e m a n n !surface Definition. W , •J I , W . M I 1. ' : " j ,: f j : set of functions harmonic on I ; W = f on a. ' ■ 1 lK : W • with fx + = S f2I = 5- LI =. I. 4. If f Ss O , then 5*. L d(Lf)* - Lf 0. ' ! a a nd . and continuous ,on L has the followihg properties: 2. I ' i! i: i , We denote b y L a n operator whose domain consists of I ' ■! I I ' the set of continuous real-valued functions; on L(f) ' : ' ; 1I i i' , , be the boundary of a Regularly embedded open set : compact complement. I. are distinguishable and its !exterior have the same boundary and this boundary is a, I -dimensional submanifold; O . if the interior and exterior o f • W. in the following s e n s e : Let Wi is a regular Iy embedded open set of the open ■ Riemann surface P °x;Lfx T : ; I ' ^ i0. I/ 11' : j i whose range is the W In addition, : ’ I -8Here p represents; the ideal boundary of 1 ;• a normal linear o p e rator. I W. , Such an operator is called ,1 1 ; . 1• ; , ; . Some important examples of normal linear operators are the principal r ' I ' ' . : operators Lq of C 2 ]. and boundary value problem in To define L we let W* Por a giyon f j on Oi , L Qf solves the w i t h k vanishing normal derivative on P (W). I I ! . i be the solution o f /the boundary value problem ' j -I , Uf 1 w i t h boundary values f on Oi and O on P . ■ Let ; "w be the harmonic / i function w h i c h has boundary values O on. a! a n d '.I on P Then L^ f is defined b y the equation L xf Observe that L f = i. Uf equals - Up f oh ; ; <a:(uf)* /Vpdw-* ) a w. ' ■ and is constant on P w i t h vanishing 1 i :•I flux t h e r e . I | i' " ' ' ■ 7' ! ' A third example is the operator L*, referred to as the operator . i associated vrith a removable centered at a point B of W I defined on the boundary A, a n d equal to Oi of A, then W 1 In general, 1 ' ; '; need not be connected. separating- W ' a. . f. b e 'the restrictions of L. I is defined on each 'i! 1 ' b y setting L f = the operators L.. I* I W . 1 . ; I -We can therefore consider i W*. ' .f''t o ' a n d with boundary x suppose that an operator Then we can define a n ew operator ■ L on W 1 : i ' : L ^ f Such an operator 11 continuous on ■ ,I into a finite number of disjoint sets : Let : :■ , L*f, is harmonic on A, ■ ■ on 0! . f i singularity. If A i is a parametric disk ; ! . and f ' is a continuous real-valued function IL I I , i i I is called the direct sum of ' .■ — C) H The following linear operator theorem (Ahlfors and Sario [ 2 ]) is used to construct harmonic functions which will play a key role in this investigation. Theorem. Let W he an open Riemann surface and L a normal linear operator defined w i t h respect to a regularly embedded open set where W 1 — on W W has compact complement. If s W1 is harmonic on Wt c w and continuous « w i t h vanishing flux, then there exists a function satisfying on W p - s .= L(p - s ) . p harmonic on The function p is uniquely determined to within a constant. That is to say, p modulo § p. s has the properties of L on W 1 The Reduction Theorem. he a n exhaustion of the open surface Let he a sequence of functions such that .f Definition. compact set. W, N i f is liarmonic (analytic) on S fn j* converges on compact sets of K c W , there exists an and such that W -j^ fn ; n W . if for every - R converges uniformly on K. T n fis a normal family on Definition. of W when every sequence contains a subsequence which converges on compact sets of W. A w e l l known sufficient condition for a family of analytic functions fn f to he normal is the f o l lowing. exists a constant M and an integer I f for eyery compact set R such that 'If ( z)I < M K, there for all -10n> N and a ll z g K, then I f ^ is a normal family. T l ' Let -) W f he an exhaustion of L nj he classes of functions w i t h domains let m and m W W and let and he functionals defined on n Reduction Theorem (Sario C 8 ]). and Hin and C : n C and C n respectively. W and W n for each In addition, . Suppose that the functionals satisfy the following conditions w i t h respect to the classes I. m(f) 2. If he true if 3. W = m Iim n c W n is replaced h y fe C n W and f e n among a l l ° k<_n and C f jTT e Wm . f e m n (fk ) f C . m The same must h y C. then There exists a function functional m then is a sequence of functions in If m (f). and if n converges uniformly to 4. m n n e C n C n an d if i f, I k converges to such that f n m^(f). minimizes the C . ■ n 5« Por f e m^(f) ^ m_^(f). 6. The family of minimizing functions mentioned in (4) is a normal family w i t h the limiting functions belonging to Then any limit function a n d the value of the minimum is f = m(f) Iim f n n C. minimizes llm Bn (I11) • m among all f e C, I (■ I CHAPTER II EXTREMAL PROPERTIES OF HARMONIC:A E D ANALYTIC FUNCTIONS ON APPROXIMATING SURFACES Consider a planar bordered Riemann surface W of infinite connecti­ vity a nd having one compact border component' Y •. In order to describe I 'I the remaining part of the boundary of W, we-: recall that such a surface can be embedded in a Riemann sphere S 2 (Ahlfors and Sario [2l ). to this embedding, it is assumed that no point of • Y 1 •. With respect is a limit point of points belonging to any other boundary components. 'Such a border compoi _ nent is called "isolated". Also, it; is assumed that the surface W - Y 2 is open in S . ■ : : — S ' belonging to the interior of W - Y We fix a point — Operations in W following. . . of compact bordered surfaces -j W ^ Y Y and £ . In this chapter of finite connectivity and each , _ .! CO Furthermore,; W jnc W ^ +1 and W = ^ W^. . n by P1, P2 , ..., Pk ^ W ca'n be exhausted b y a sequence r',' each 11 is one border component of W Evidently, Denote the remaining border components 1 i and for convenience 'set we propose in the I It is known that such a surface containing 2 I such as I n t , boundary, e t c . , are referred to S ^ 'i to develop extremal : ' :; i ^i* properties of certain harmonic and analytic functions defined on 'the approximating surfaces W . : I' g 4 . 1 : ^ Construction of h a r m o n i c .functions:on approximating surfaces. !I I Using S a r i o 1s linear operator method; described in [ 5 ] , ;the har- i monic functions p ^ , on and p. m I j will be constructed on each of the , I*, , -12- approximating regions W k log 12 -£ . The singularity function s will he taken to be j in a neighborhood of £ , identically zero in neighborhoods of the Pi , and equal to the singularity function s k function s near y . ^ ■ is that function w h i c h has flux equal to SjrX0 The singularity Uote that ■the total flux o f s is zero* L1 near y, Lq near each Pi , The direct sum of the linear operators and L ' near S , w i t h an application of the linear operator method, yields the X-th harmonic function p Q^ up to a constant. malized at £ so as to get property The function ii) of H ( X) satisfied. is n o r ­ Here, L* is the operator associated w i t h the class of harmonic functions having a removable singularity at £ . The X-th harmonic function p ^ is derived from the same direct sum of operators with the exception that near each is used Pi * The functions p ^ and p ^ possess the necessary properties required to belong to a class of harmonic functions HfiC X) which will be defined in the next section. In addition, the function p Pp ^ 0-Q = O on each P., while .on X on has its normal derivative is constant on each P i w i t h flux ■ 1 / p . dpin* = °* These functions will be seen to satisfy certain extremal pr o p e r t i e s . S 5 . Extremal properties of harmonic functions on approximating s u r f a c e s . The extremal properties that the X-th harmonic functions are -to enjoy are w i t h respect to classes of functions defined as follows: -13r D e f inition. H (^ ) n is the set of functions p ( z ) which are har- ; . ; : monic on W n -I while satisfying i ; ) p ( z ) = const. = c(p) for zey and J ^.dp* = 2 jcX , and ii.) the function! harmonic continuation to £ with h(z) = :p(z) - Xlog |z-£| has a ' ', h(;i) = O.; X The harmonjLc functions p Q^ and p ^ "belong to the class ( X )• A lemma that is essential in obtaining their extremal properties will first i be stated and proved. Lemma T_. iIf 5 is the border of a parametric disk about orientation induced b y the surface Wn , t h e n 1for every p and Pdq-X- have Proof. (I) J b P<3a* q.e i , with Hn ( X ) we Cjdp-X- Adding and subtracting the same term we have - qdp* = j^(p-q) dq-x- - q f(P" q )> • ' ! ! ■ Since the function p-q_ has a removable singularity at , i t has a single valued conjugate and, upon'using an! integration'by parts, the right hand i . I : side of (I) m a y be written as ' ■ i | 'J^5Cp -q ) dq-x- + ; (p-q )* d q . B u f this is equal to I m ^ s (P-Q) dQ - where P = p, + ip* and Q = q + iq * . The functions P and Q are multiple valued but the function P-Q is single I valued. 'I : Using property ii.) of H (:X )'together w i t h Cauchy's Integral 11 : I I : Formula this can be seen to be zero.The extremal property that theorem. "X $A ■ i possesses is .stated in the following I II " -I l}.- Theorem I . ---------functional Among all pe H f X ) nx ' (p) = 2jrXc(p) - JJ Pn I' x> ^ the function pdp* . -on maximizes the The value of the maximum is . ; ■ ■ ■ ■ • ■ The deviation of this: functional from its maximum is B 7 (p-p w V1x^ -irOn' ,,I 2jrX c'(p ^). on' and the maximizing function is u n i q u e . Proof. y By Green's formula we have X, \ X But since p and p fe-pon\ =JyuPn (I1-pOn) 4 (p-pon>* ' P '■ X ■ X SEtn ( X ) , J r (P-Po n ) 'd (P-Po n ) * = O . : ■ . i j Gr e e n ’s formula then Becomes (2) (P-Pon) = / 5n pap* + J p n Pon a P0J » - :/pn P0J ap, + p •on 7 The second term of the right hand expression-vanishes Because the normal dp ^ derivative O on ’ ■ For the same reason J g^pdp^i* = 0 . Therefore we change its sign and upon applying Green's formula obtain for the last t e r m of (2 ) the expression V y u B F 0h j ■ F ^ P on*..* Here 5 is the Border of a parametric ^disk A about s w i t h .orientation induced b y H - A . By lemma I .) the contribution of 5 to this integral is 0. Using property i . ) of H ( X ) n the part due to iy is 2jtX c (p ^ ) - 2 rtX c(p). J : T on Collecting all of these contributions we obtain the formula (3) X X : -r ■ 2jrxC(Pon) - Hfn(P-Pon) = SjtX CjXp ) - Jpii pap* . Since Ijjy (p-p \ on ' I ) -O we have the desired extremal property. is clearly Dgn (P-P o n ) 1 h ' The deviation I -15 - As for the uniqueness of the maximizing function, suppose another function p maximizes the functional. and hence p 1 - P q^ = C (constant). and p (p ~Po n ) = 0 ;But h y the normalization at £, C = O •on The function p p Then the deviation X xn possesses a n extremal property similar to that of X , hut the functional that it extremizes differs from the one originally on anticipated, wh i c h was 2 jcX c(p) + pdp-x-.l , Theorem 2. ---------- The function this m i n imum is +; J p n W '- sjpn The minimum value is I 2 m X ;c ( p ^ ) T h e (p - p wn Proof. minimizes .'the functional ; : atx o(p) 1L ( P ) among all p e H ( X ) . p ^ jrXn ap» • deviation from 1 i : ) and the minimizing function is u n i q u e . ln :: X We have h y using G r e e n ’s Theorem \ W j p - pl ) a P (P " - pJ p *• As in Theorem I the contribution of- T to- the integral is 0, and we write (4) Dgn ( p - P l n ) = X pdPil" + J p n P xb d P ln* • 2JpnpIPap>i• Since p ; . ; r J 4 V 0 on each Upon applying G r e e n ’s Theorem again, ' the third t e r m m a y he written as p d p in* " p I dpit ■I is constant on e a c h P. an d the Ifltuc the second term in (4) vanishes. - x : pUS p d p Init ■ - I :i i ; -X p I n d p *- By lemma I . the contribution of 5 to the jibove integral is 0 and, using property i.) of H ^ ( X ) , this reduces to. 2jtX-c(p) - 2rtX c (p i^) • ' I• Hence, (4) -16- becomes 27tX c ( p ^ ) (5) Since D g Bg ,+ (p-p - (P-P1^) P j a c(p) = Jg + pdp* 0 we have the desired!result. I' ; - 2jg p Indp*' The uniqueness of the ■ ■ minimizing functional follows exactly as the argument used in Theorem I . , - ■ I 6. . Extremal i properties of analytic functions■on approximating .surfaces. . ; . ! 1 The classes of functions in which the X-th analytic functions are to ! i I enjoy their extremal properties are !defined! as f ollows: , I ' ■ ; Definition. A (X) is the class of functions ----------n : . i :: such that the following properties are satisfied1: i.) ii.) If (z ) I = r(f) = F analytic on n . Iim F(Z)z/ (z-G )"X = ! I I Some interesting connections between t h e 1:classes i : are expressed in the following theorem. I Theorem 3- If F e A X-th analytic functions A * 0 - G . number of zeros = and A (X) n i l o g : lpi e H (x). F urthermore, the (pon t 1 pOniti 'ani ' A n = exptp^ c The function on W A F (X) then H (X) n i pin*) (X). Proof. F 2 aX. const, !for z e y w i t h d arg F = F has a X-th order zerb at , s iii.) belong to W Zq log If I will be harmonic on W - £ provided : : A n application of, the argument principle gives the on - s r JB(Hn) Wn as a arE F Hence, the only place where =: F = O ~k on Ir W d arg, F is at I.' + d arg F X. i ! -17- Wow F has a X-th order zero at £ G(z) with G ( ^ ) 4= 0. so -that near £ , F(z.) = I. ! Hence, the function h(z) mentioned in property ii.) O f H n ( X ) may he written as h(z!) = log ■IF I I- Xlogjz- £ | = loglG(z) J A sufficient condition for h ( z ) to ;he harmonically extendable is that ^iijP (z- £) h(z). = 0. This is satisfied and furthermore, by continuity ' i ! ' : considerations,' the extension takes the proper vdlue O since ^i^ h(z) log IG( £ ) j = 0. H Property ii.) of property i.) is readily checked. i j i ! ! : j I We prove the second part of Theorem 3 \ as a function. = (X ) is therefore satisfied and :n I gous proof for F ' ■ I X 'for F - and omit the analo- ' ' I X:, We must "see that F a = ‘exp (p A 4- i p A *) is defined on ■on on The multiple valued conjugate function p a^ x- has periods I 2jrX 'rd P on* and k has a period of 2 jtX * for each i . 1 That is to say, it O d v and we have F on : defined as.a function. . / •; ■ Here, F ' on ; is normalized b y setting h*, the conjugate ;of the continuation for X •on , _ X log I „I . ,. jz - £ j, equal to zero at zed in a similar manner. for z e Y", and of H n (X ’ )} / / in We have X on ' : £ ^ o n ^ _X The mapping F ■ is normali­ X ' = ■ exP (p ^L(z ) ) ■’o n ' - const. I T .X d !Pon-X- 7 2 jrX, . properties i i , ) a n d iii.) of Theorem 3 and F d arg F JT . X By the normalization condition iii.) of Afi( X) it , follows that G( I ) (%-£ ) = i Using property ii.) ) may'be obtained. I ■ X leads us to the following extremal properties for F on stated below in Theorems 4. and 3.1 : i. -1 8 - Theorem 4. The function F 2JtI among F e A ^ X ). log r(F) /jjj log jF I d arg F The maximum is 2jti log r ( F ^ ) :and the deviation from the maximum is D - (log Proof. uniquely maximizes the functional on Jf / f ). on We have log .|f ' X1 ■ p and according to Theorem I . this -tOn on maximizes the functional 2 jtX c(p) pdp*. - And "by Theorem 3 •, if Pn F eAn (X ), then log 2 Jt^ log r(F) - Ls |f | e log H e n c e , IrQn maximizes the functional ( I). |F j d argF among all F'e A (X ). x ' - - % ' D^Ciog If I - log| FonI ) - ,Ilt4i (log If / Fon I). I Uniqueness is argued as in Theorem I. II ii The deviation is *' ; ' 1 In a n analogous manner the following theorem expressing the extremal : , property for F may he proved. Theorem 5. --------— ; : The function F 2 jtX log r(F) among F e A ( X ) . n. minimum is D^n (log + in Jg uniquely minimizes the functional log Jf |id argF los X ' '' / fF in ^darg F The minimum is 2 jtX log r (F ) and the deviation frcm the ■ x , . in ’ If / F i n I )• : i CHAPTER III ' I EXTREMAL PROPERTIES OF THE FIR S T 1X-TKl PRINCIPAL HARMONIC A N D ANALYTIC -FUNCTIONS : Extremal properties for surfaces of infinite connectivity which ! 1 / ; generalize the results of Chapter II will he derived in this chapter, with i M the main tool u s e d being S a r i o ’s Reduction Theorem [8 ] i I T ■ Extremal properties of the first ^-th principal harmonic function. Let ¥ he a planar bordered Riemann !surface and -jw. f ; set as described in Chapter II. rx V Lemma T_. Proof. family -^F“ of a family The family Because p is normal. jPo n j X log If on I ; I I1 ■ !' I nJ ' an exhausting . I - a normal family on the surface W - 5. on it suffices to show that the A w e l l known sufficient condition for normality f >■ of analytic functions on a region is that the functions. for sufficiently large n, be uniformly bounded, .on every compact subset of 'I the region. I ;: ! Let K be an arbitrary compact set contained in ¥ and choose N large- e nough such that K c ¥^. log If JJ1I d arg F ^ on on Denote b y A ( F ^ , W y ) the quantity for n - N . ' This is a ,non-negative quantity) for consider the Dirichlet integral over the surface ! tion is induced b y ¥. (,) 0 - % • W^ ■ W y , whose orienta'f Then - W j 1= S lL n l1 = J f e 1=S ^ o n P g + A<-*L’V ■ ': - 20 - X The first integral of (l) is zero and (2) 2^X log r(F h ^ 2z(X log r(F^) -0.' Hence we have + A(?.^ W j : 6 on' , -N X, 2*X log r(P^). : The last inequality results from the extremal theorem for the function on W jj. It must be noted that the 'expression in the middle is the functional given in the extremal theorem (Theorem 4. of Chapter II). the r(F X . ' ' ' ' :: ) are bounded.for all n ^ N. An application of the argument on principle, yields is a z Thus on (z) | ^ r(F ) for z e W'!. o n ,. xi To see this, suppose there . w i t h W = F ^(zn ) such that l"W„ j > r(F . Then, there is no o on ° ° on ' X, . :• I change in the argument of the function jF (z) - w around each of the cycles o e W IF n i T and P. for every i. F ' i. Therefore, b y the argument principle, the function X I ' X (z) - w has no zeros on W , i . e .,ithe value w is'assumed b y F for on o . _ n' . o . on no z e W . n Hence,- 'FonJ ^or n - ^ is uniformly bounded on every compact subset and is therefore a normal f a m i l y on W. i We let p ^ be some limit of the.' normal family : X wi l l see that there is exactly one such p . o ip . Later, we i ■ X The function p is referred o to as the "first" X-th principal harmonic function to distinguish it from x ; the function p. wh i c h has an analogous developments 1 ;; : The Reduction Theorem w i l l be used to develop an extremal property for p ^ in the class of functions defined as follows: 0 . Definition. H(X) . I! ; is the class of functions jp : ; ;; on W - I and satisfy the properties: : ; which are harmonic -21 i.) p(z) c(p) ii.) h(z') P(z) "with const., for z e r with - 2jtX. hasia harmonic continuation 0. h(" ■£) The function X log jjz - §; dp* pO ' insure the existence of certain integrals is stated and,proved "below. Lemma 2. If m<n the inequality 7 5 pdp*' J all p e H (X). Proof. • • I Consider the surface j W The Dirichlet integral of p !! ^ 7 5 pdp* v ^n ' holds for Wm , with orientation induced "by on this surface is non-negative and we have 0 i \ - H1 (p> 'Pn - :Pm papi! In the following, Hn ' pdp* is to he understood as pdp*. Iim pdp*, the existence of the limit being guaranteed b y the monotonicity of such integrals as shown in lemma 2. f 4 haustion I i r as can be seen b y applying. G r e e n ’s Theorem. Theorem 1_. <j) (p) = 2 jcX c(pQ ) This definition is independent of the ex- 2 tcX c(p) The function - /ppap* X: p^ i uniquely maximizes the functional among all p e E ( X ) . and the deviation from this maximum is given b y , Proof. ----- 0 : D ^ (p - P q )* , ^ It w i l l be shown via the Reduction Theorem that p the functional p The maximum is maximizes the functional ;, -2xX c(p) + L pdp*. 0 minimizes Then it will follow that — The functional - § 22— satisfies - (| (p) Iim (In Cp) L the existence n of the limit "being insured by the monotonicity of the integrals as proved in lemma 2. W m a pdp*, Pn The restrictions of functions defined on surfaces belong to the class H ( X ) n' defined on W. J for n £ The same m ay be said of functions Condition 3«) is satisfied because of standard theorems on interchange of limits. ments of 4.). The extremal theorem for p ^ fulfills the requireon Lemma 2.) provides us with the monotone property of the while the requirements of 6 .) are met b y lemma I . functionals the functional Gj) is maximized by p^" o Thus with the maximum being Iim 0 (p ^) = n n on 2 rtX c(p^) . As for the deviation formula, consider the function p e Cp-P 0 ) which belongs to H ( X ) for arbitrary e. functional (3 ) ffl In (pe) n = Po + The value it gives the is 2k X . X c(p^) - 1P p O dbo* + n ° X % = - = X, a(p-pj* /p.(p-po ) ~ n wh i c h is a quadratic polynomial in e with a ^ the coefficient of the e term. The integral in the last term is precisely IH (p-p^). X N o w suppose L^C p -Pq ) = X '° Iim Ey (P“P 0 ) is finite. X n of EyCp-P 0 ) determines the finiteness of the integrals ^ P 0 d P^" * P /ppdp* and as can be seen b y using the triangle inequality for Dirichlet integrals on surfaces W - A 7 where A setting e The finiteness =1, is a parametric d i s k about consider taking limits on both sides of (3-) •" i. Upon In view of the preceding rema r k s , each term with the possible exception of an possesses a finite limit. Therefore so does a^. Denote this limit b y a. Taking ■ limits of (3 ) we obtain (^) I (Pg ) = : I (Pq ) + a .e ,! : - e 2 D_(p - p ^ ) . ■ But according to the first part of this theorem <j) (p ; e = 0. = ) has a maximum when 6 ; Therefore its derivative with respect to e evaluated at e must vanish; i.e., a e ■ - 0. = 0 The desired deviation formula results when I is substituted into (4). . .. ■ I If Dg(p-pJ":) is infinite, then the triangle inequality for Dirichlet integrals on surfaces W - A can be used to show that Ij (p) is infinite as well. ' The deviation formula holds in' the sense that both sides . J : are i n f inite. ' ■ The uniqueness of the maximizing function .'follows'exactly as in the i pr o o f of the extrbmal theorem in the; finite case. I ; . ■ ; ! uniqueness of the maximizing function is the fact I limit of the normal T x family |Pon An analytic function that of on i.e., F^ " 0 0 that p^ is the only ': ' This completes the proof of Theorem I. will be exp (p^ referred to as the "first” i : A consequence of the d e f i n e d .in an analogous manner to + i p^ i ). 0 * • The function F^ : 0 will be 1 -th principal analytic function to distin­ guish it from another principal function F^; . § 8 . Extremal Properties of the First 1-th Principal Analytic Function The extremal property of the'function p I.; '' leads immediately to the extremal property for the 1 -th principal analytic function A class of competing analytic functions is defined on W as fo l l o w s : Definition. A (l) is the class of !functions F analytic on W and -24satisfying the p r o p e r t i e s : i.) ii.) iii.) I F(z )I = I r(F) c o n s t . for !zey with / F has a X-th order zero at §. / X ( z -g)A Iim F( z ) / .Z^S The function F^ , i °i ■' ■ : = darg F = O . I 1 4 :' ' already defined a b o v e ,,helongs to the class A ( X ) r i ■; The quantity - L l o g IFj darg, F^ denoted b y A(F),:iis understood as : Iim - /n n log Jf | d arg F. ■ Theorem 2. : J :i : X uniquely maximized by the function F ” . the deviation from the maximum is ■ If F e A(X (log ) then log -5- A(F) is The maximum is 2 jrX log r(F^^) and ! Proof. , Among F e A( X ) the functional G ^ X log r(F) I FIZ^Fq I ). ;■ i If J e H :( X ). : The argument used in Theorem 4 of the preceding chapter yields the result. . : :X An important corollary characterizing F q univalent function F of [6] , i.e.;, the principal function corresponding ° to X= ' i as a 'I^th power of the ,■ I ^ w i l l now be drawn from Theorem 2. Corollary 1_. function F • o Proof. X The function F q 1 ' is the X-th power of the univalent 4 ■; : ' : It is first observed that Cf ( z )^ ; ; . belongs to A( X •)-. Because of the uniqueness of the maximizing function in Theorem remains to be seen that the deviation 2, it only Dfl(log .I (F0(z )X/ Fq (z ) * ) = For convenience, denote b y f ( z ) the analytic; function (F ( z ) ) / F ; -I O ' O w i t h a removable singularity at B .; By G r e e n ’s Theorem O(z ) i : I ; -25D- (log j fI ) ' 'yUp ' I f I darg f . log The contihution of T .to the integral is 0. ( 5) (log I f ! ) O - D - = L N ow l o g i f J da'rg f 1 Let f 131 ( z) = ( F ^ .( z ) OIil w i t h a limit "being f. ) F ( z-). Oul = I log If Idarg f . ! The family K f > i s a normal family ,• • 331J Therefore, interchanging limits and. integration, the' right hand side of (5 ) m a y he written, in the sense of iterated limits, as y™ AJft Jpn 10S I V (6) D0 (IogIfI) -ve 1»^ darB = lfi'U m J p n log |fm l, d ar g fm Considering the Dirichlet integral of log |f_| we obtain 0. over;the surface Wn^ ■' . W a = 0 " i: = /@ml08lIiB ld arg:fm " / % losliB l d ars fH : The first integral is 0 a n d we have (8) r s log | f j d arg f ^ 1 ; ’ ' :i . for every im ^ 0 ■ J *. - : n. ; Thus, taking iterated limits of (8),jwe obtain p: „ (9) 1Jjl A^JPnlog lfB l a ars Combining (6) and (9) ij fm' s ' ' 0J ' . J ’ the desired result D ^ '(logl f Ij = 0. This completes the pr o o f of Corollary I ,1 A consequence of this corollary is the fact 'that F q — is a weakly I: X -valent radial slit mapping of W into a disc in the following s e n s e : Definition: ' The m a p ping F( z ]) is called weakly I X -valent if for each w e F ( W ) , the set F ”1(w) consists of at most X points z e W, and for some w e F ( W ) the set F - (w) consist^ o f .exactly X points. A weakly. X- valent mapping F(z) of W into the point set Si is called a radial slit - 'I f • • , mapping of W into S if e a c h component1 of the set i v e 8 ; F ~ (w) contains - I i I 1 at most k-1 points z e W r is a radial slit of point. : J i '• i x Ttro corollaries stating other extremal properties of F will he ° dratm. Define a class of functions differing from A(l) in that 5' (l) which is a' subclass of A(F) ' ^i 0 for each F e A(X) and ^J-(X). : I X1■ F q1 has the property Corollary 2. 'The function fiSgl) Proof. Let F e PmXlogr(F) =; 3(x) ^ TheA(F) 'I PmXl o g r(F); + & band; 'I A(F) . ' :: The last inequality is the extremal property, of : Define a class of functions Jtj'-(x) 1 : g)- (X) only in thd condition |f (z ) I '= Corollary 3« I Proof; r > r(F ). : j ■ I | ' -N ^ 2 m X :log r( F ). -X F • ■ - 0 . :Hence the r e s u l t . !° defined on W and differing from ;; i. : c o n s t . on y is dropped. . The function F | has the extremal property oI , ,Sgfo W 1 g P ? 1 " : i Suppose there is a function G e .%|:(x) with mi^. |c(z)| Let W jhe the simply connected region !enclosed hy = G(y). Choose a function cp belonging to the' class'of univalent functions A(l) ‘ I' i; :• ; defined for W $ and consider the disk of radius ^ 1Whicli is the image of W 1 under ©. Let C be tbie circle, contained entirely in W s , having radius : ; : . r and center at 6. Let M be the maximum value of j® ( z ) | for z e C. i L . : By Cauchy's estimate for the derivative of ah analytic function we have -27-: ICp 5(o)j cp!(o) I= M / r . The normalization condition' iii.) of A(j_) implies that 3, and we have r A M. function cp ° G Belongs to Because M ^ p we have p > r( ). The 3 ( X ) and we have- a contradiction to Corollary •I : M I : ■ CHAPTER! IV . ! I ! EXTREMAL PROPERTIES OF TEE SECOND X -TH PRINCIPAL HARMONIC A N D ANALYTIC FUNCTIONS : I 1 ; I' In this chapter extremal properties of certain X-th principal functions on W w i l l he derived. § 9 • . , The extremal property of the second X-th principal harmonic function. , First, a few properties of the' functions p Q X . are developed which are needed to obtain the normality of the. family Lemma I .. If Proof. m-n, then Jg p Qm X d P om* 0. Dy lemma 2 of Chapter :3 11 'p p O m d p Om* Lemma 2. X X I 'P ^ o m ^ o m * ' !' •m ■ i, i; - X 1c (p o^ ) f The sequence is! a decreasing sequence. ' Proof. Let m - n . T :i i X Using the'extremal property for p we have : on 2>a =Cp 0^ ) SI 2«x o(po^ ) Q pO m d J o m * ! P. : i 2 * X 6 (p « ) - The last inequality follows from lemma I. Lemma j5* Proof. log ( Xi F I . The family X ^p "j" -is a normal family on W - L It suffices to show the family ;I ' large enough such that K c W ™ F on ,: • • For an arbitrary compact set ; K ! we write for Xl ■ X F_._ f" ;normal because p_q L I^ n j in ■ ; ! contained in W Using the 'extremal property on n - N I chose N ! W for -292rt)v.log r(F ^ _ in ^ 2jtX log r ( F X ) on ^ 2jtX log r ( F X Oh ). The last inequality follows from lemma 2. upon noting that log r ( F ^ ) = c CPyn )- Thus the r Clni n ^ are uniformly hounded for n ^ N. An a p p l i ­ cation of the argument principle as in lemma I . of Chapter III provides us w i t h If ^ ( z) I^ r(F ^ ) for z e K. Therefore, is uniformly hounded on the compact set K. -j? H 'for n ^ M We infer that r is a normal family. We denote h y p ^ a limit of the 'family' w i l l he referred to as the "second" was not a h Ie to he applied. . Such a function 1-th principal harmonic function. Due to the extra term of the functional orem -jp^^j- ^ of Chapter II, the Reduction The Namely, it is not known whether the I functional is an increasing one or not. \ A sequence of lemmas involving convergence of Dirichlet integrals is needed to obtain the extremal theorem _ X for P i Lemma 4. -------- For the functions Sr Proof. X n According to C 4 ] ■exp (Pi + i p^*) X log |f I= X p . ^ in K n " we 2% ) have the relation = the function and F i [ 6] , corresponding to the case p X= ( F 1Cz) ) ^ F i (z) where is the univalent principal function of I. X Thus we have p -"- L i k ewise, for the functions p^ = we have log p^ I Fi I = = X n The concern for these properties is manifested in the following equation: (I) Ds n (p^ - 'P^ ) X 2 D; (Pj Pm)' , 'i -3q~ The restriction because /„ d p * Pi J of p to W ■ belongs to the class d arg i F^; consequence of the univalence of F1 :0 j for,"each . p> of [ 6 ] This is a B y 'ldmma 2. of [6] , i.e., the extremal theorem for p . , w e have in (2) (l> where - P,-) is. the functional and the family "j P i n T The functional 9(p) Theorem Iim cP (p Iim (P1 Tl % n j!i Hl - ) P111) cPvi (p' ) , <? J p 1 ) 2^ c(p) + pdpx- . satisfies the conditions:of the Reduction Theorem. is defined as'lim^p (p) = . cP (p ) . ILa 0. 'and we have via the Reduction Taking limits of , (2) we obtain Equation (l) then yields the r e s u l t . - Lemma 5 . If p e H(l) and- . (p) , then lira _ n Proof. is finite, where A 'ij\.' X ( p ! - p ^ p) 1■ . !1 j parametric disk about = is a 0. According to the Cauchy-Schwarz inequality for the mixed • I . :' ]! Dirichlet integral we have (DW - A (pin - pi ' p> )2 n : % n has the limit 0. ' i - A (pin " Us i n g lemma 4. and noting the finiteness of co m p l e t e . The functional ^(p) 'A • . n _ A (p) ' the right hand side Therefore, the left side must also, and the proof is ■ . . ': ; I ;! ;; -31 If A is a parametric disk about 1 : I ' we have the following l e m m a . ' : ■i :' 1 : , is finite, :then .Iim Dr ' An . ■1 In .W ■ - A in ; 1 .; , n exists for each pi e H ( X ) , and is equal to j(p . p). .Lemma 6. If --------W - A B- „ (p) .' Proof. : ' I ~ 1 i . n) . The linearity of the inixed Pirichlet integral furnishes us wi t h (5) X-A (p“ n ’ p) " dH n - A (i,x i > p) “ I : X ni; By lemma 5-) the right hand side of ;(3) tends to i O . complete if the sequence jD^ _ ^ ( p ^ b )• The proof will be p)j-possesses a l i m i t . It will be n , ' : I established that this sequence is Cauchy. The Cauchy-Schwarz inequality i supplies us w i t h I M X m -W n A < P) - dS m - W (Px )-IdB -W (P) n ! m ■ :n The existence of Iim B g (p) for each p ei H(X); implies that each of the n . terms on the right hand side of (4) n an be made ,small for sufficiently i I large m and n. :! Thus the desired sequence is Cadchy, a n d in view of previous remarks the result is established. i I ' 1 A corollary of this lemma is Ilater Used in establishing a deviation I fo r m u l a . Corollary 1_. Xx D1^(P-P1 ) • : If Dg _ ^ (p) ' Ii is finite,Hthen Iim D 5 ( p - p ^ ) W "in' n. -32- Proof. disk A It suffices to prove 'the stj.ted .relation-with a parametric about §' removed from each; of the surfaces W and W. i I ■ i! n the quadratic character of the Dirichlet integral, we m a y write (P - Vm ) = D5 _ & (p) 2 I DW - A (P' ?3.n? n .;. ; <5) X - A ;i We consider taking limits of (5 )• side is limit : Due to nM - A n ' ■ The limit of the first term on the right D g _ ^ ( p ) , while according to ;lemma 6.; t h e :middle term has the / 1 \ (p, P 1 ) . -2 D t-t a W - A : :I The triangle inequality for Dirichlet integrals is used to write (G) 1% n - A DW - A n X jI - % X n X l- - A dW Xn- X The last inequality follows from the fact that the Dirichlet integral is an ■ increasing function of the region. " %• lemma the right hand side of (6) approaches 0 w i t h increasing n. . H e n c e , the term DT-T 'A (p possesses , w - A in I ' I n ■ the limit D,-^ A (p ). After taking limits of (5 ) and using the quadratic IrV ~ ZX I ; X I character of D g Cp-Pi ), the result is established. H o w it can be seen that the f u n c t i o n a l ' Cf) has a limit on n. =■ 2jtXc(p) +jfp pdp*:: - P i^ dp* The second term'of the functional possesses a limit because of the monotone property for buch integrals as proved in lemma 2 of Chapter III. If Dr. A (p) is: finite, then lemma 6. furnishes us w i t h ■! -33-: the existence of Iim B g •_ ^ ( p ^ / p ) , and this implies the existence ■ n“ ; \ of the limit of the last term of the functional. ,Lemma 6. also furnishes X p^ dp* . us with the value of the limit, that being VJhen D g A (p) is infinite, the triangle inequality,:as in equation’(6) of Corollary I. with p in the role of p^ i * n (p ) and we have that Iim n ^ 2 jcX c(p 3 1Ir (p) n by ^(p) xKp) Iim p from this value is ' uniquely ,minimizes the functional -f L - . p^. dp-x- 2 jcX c(p^) ' ' I and the deviation 2- ; 1 : p X furnishes us with i (p) i (p). it follows that p ' 1 2j^ pdp* j:- . X Dr. (p - p ). ¥ I Therefore we define a Li The m i n imum value is I 2JtX c(p^) 1 The extremal property!of ^ dH (P - P 1^) ■ n X ' 2jtX c(pi n ) Is , V (p). 11. 2aX c(p) among all p e H(X),. + is infinite as'well. i Theorem 1_. The function Proof. d i m DT -=(p ^ - p) ■n Vy in -By equation (5 ) of Chapter, II infinite as well. n e w functional t h e r e , implies that ■ . for e a c h p G iH(X). . Passing to the limit we obtain Using the factb that ' 'p^ V (p ) = 2jtX c.(p ) „ Hence = Xp^ and.^ dp^x- = 0, the minimum property of As for the deviation recall the equation (5 ). from Chapter II, namely tl/ n (p) n == ■ Dg X (p-Pi n ) n ■ in + X 2 jcX c(p i^). • in By iCorollary ■I . to lemma 6. the : : ; . . ■ . ) -34limit of the term is f i n i t e . % Hence, if (p - p ^ ) ^_^(p ) is (p - p^) providing' D jj _ A (p) ' is finite, we obtain the desired deviation formula after passage to the limit of the above expression for ^n (p)• ^W-A Dg. i® infinite, then so'is (p - p ^ ) . If The use of the triangle inequality for' Dirichlet integrals obtains this fact. Therefore, the devia­ tion formula holds in the.sense that both sides are infinite. i 10. Extremal properties of the second X-th principal analytic function. We define the function . normalized so as to get h-x-( £) nuation of p ^ - log Jz - I |. and for e a c h Fe A(X) as = exp (p^ + i p^-x-) 0. Here, h-x-( I ) The function we have IogjFj where p \ is the harmonic conti belongs to the class belonging to is H(x). A(l) These remarks together w i t h Theorem I establish, in a manner analogous to the proof of Theorem 2 of Chapter'III, the following theorem. Theorem 2. The X-th principal function F^ uniquely minimizes the A(X). The minimum' value is functional 23tX log r(F) + f IogjFj among a l l functions F 2 jtX log r(F^) d argF belonging to the class and the deviation from this value is It should be remarked that disk mapping. - 2^ F^ This is a consequence of univalent function F (IogjFzZF^I)'. is a weakly X-valent circular slit F ^ being a X-th power of the w h i c h is proved in [ 4] . This could not be proved X here since the fact was used to derive the extremal property of F^ . _35_. A corollary exhibiting another extremal property of drawn from Theorem 2. Denote hy B(F) will he the quantity I % Define a class o f •functions (l) which is to he a subclass of w i t h the additional restriction that Corollary T_. Proof. 2aX log Let r(F) The function min r(F) 3(x) - F e F = B(F) X ^ A n (X) 0. has the extremal property r(F^). i 3 (x). Then 2itX log r(F) +. B(F) ^ 2 jtX log r ( F ^ ) . The last inequality is the extremal property of F^ . It is a n open question whether F^ Hence the result. has a min-max extremal property similar to that of Corollary 2. of Chapter III. A proof similar to the pr o o f there wo u l d he complete if it could he established that the function <P° G in £} belongs to the class (X) with m|x | G(z) 3 (X). | = Here G r < r(F^) is a function assumed to he . I I; ! I I I LITERATURE CITED I ! ; I .1 : ; : i I' LITERATURE CITED [ I ] L. A l i l f o r s Complex A n a lysis', McGraw-Hiil Book- Company, Inc., (1952). | : j ' I -I C 2 ] L. A h l f o r s , and L. S a r i o , Riemann Surf a c e s , Princeton University ■ Press, (i9 6 0 ). ; ; [ 3 ] P. E i c k e I, On extremal properties for annular r a d i a l ,and circular slit mappings of bordered Riemann surfaces, Pacific J. M a t h . , 1 1 , 1487-1503. , i [ 4 ] P. Eic k e I, A note on principal functions and multipI y -valent canonical m a p p i n g s . Pacific J. M a t h . , (to a p p e a r ) . L 5 ] ■ ;i . L. S a r i o , A linear operator -method on arbitrary Riemann surfaces, Trans. A m e r . Math. Soc., J2 (l952), 281-295. [ 6 ] ,L. S a r i o , Strong and weak boundary components, J. Analyse M a t h . , 5, (1956-57), 389-398.; " I -' ! ' [ 7 3 E. S a r i o , Capacity of a boundary and a 'boundary component, Ann. of Math., 52, TT95 ii-T, 135-144. [ 8 ] L. S a r i o , Extremal probl e ms'and' harmonic interpolation on open Riemann surfaces, T r a n s . A m e r . Math. Soc., 79, (3955), 362-377. i I i I,' I MONTANA STATE UNIVERSITY LIBRARIES CO 111I 11(111IIIIII 7(32 10C 1C>40* O ■ cop.2 ^ Garoutte, D. E. Extremal properties character: .Z ing weakly I -valent principal functions " AND ADONtCOa q n (2 67/9 ' ^