I nt. J. Math. Math. Si. No. 2 (1980) 255-266 255 Vol. QUASI-CONVEX UNIVALENT FUNCTIONS K. INAYAT NOOR and D.K. THOMAS Kerman University P. O. Box 182 Kerman, Iran University College of Swansea SWANSEA SA2 8PP, Wales (Received May 9, 1979 and Revised form in June 25, 1979) In this paper, a new class of normalized univalent functions is intro- ABSTRACT. duced. n The properties of this class and its relationship with some other sub- classes of univalent functions are studied. The functions in this class are close- to-convex. 1980 MATHEMATICS SUBJECT CLASSIFICATION CODES: KEFWORDS AND PHRASES. 1 Primary OA2; Secondary 0A34 Univalent functions, Quasi-nvex, Close-to-nvex. INTRODUCTION Denote by S the class of functions f which are regular and univalent in the unit disc E and satisfy f(0) 0 and f’ (0) i. The subclasses S , and C of star- like and convex functions respectively are well known and have been extensively studied. S , and C are connected by the basic property K. I. NOOR AND D. K. THOMAS 256 C f The subclass known and K S of zf’ e S if and only if , (i.i) consisting of close-to-convex function is also well , many of the properties of can be extended to the wider class K. S The purpose of this paper is to introduce a natural analogue of the class C in terms of the property defined in (i.i). MAIN RESULTS. 2. Def. be Let f quasi-convex g’(0) E be regular in E in with 0 and f’(0) f(O) if there exists a convex function 1 such that for z e I. Then f is said to with g(0) g 0, E, .(zf’ (z))’ g’ (z)" Re (2.2) 0. > Denote the class of quasi-convex functions by Q. It is clear that that Q c when f(z) g(z), C Q so that C Q. c We show first K, so that every quasi-convex functions is univalent. THEOREM I. Let f e Q. Then, for z e E, Re and so Q c K univalent in PROOF: c S __zf..’.(z). > 0 g(z) thus, every quasi-convex function is close-to-convex and hence E. A result of Libera [4] shows that, if gular in E with s(0) t(0) Re , 0 and t e S __s.’_(z) t’(z) > 0 An immediate application of this with s(z) s and Re s(z) t(z) are functions re- E, then for z + t > O. zf’ (z) and t(z) g(z) proves QUASI-CONVEX UNIVALENT FUNCTIONS 257 Theorem i. It follows at once from the definition that Q if and f only if zf’ e K. (2.2) We can thus write C > S Q > K where the direction of the arrow indicates set inclusion. Theorem i shows that the image domain for all f e Q is close-to-convex. However a specific characterisation of the image domain for f e Q remains an open question. we state now some basic properties of quasi-convex functions which can easily We omit the proofs as they are be extended from the class of convex functions. simple extensions from the convex case. THEOREM 2. Let f e Q with f(z) + z n=2 (i) lanl < (l+r) 2- r w n If’(z) > 1/2 Then for If (z) where zl r < i, in E 2,3,..., < (l-r) <_ l+r (iv) n < i, (+/-i) (iii) a z n -< 2 r l-r f(z)+ w All inequalities are sharp, equality being attained for f0(z) Z l-z 258 K. I. NOOR AND D. K. THOMAS We now give an example of a function in Q Let Example 1 3 1 maps E i-i fl which is not convex. z fl(z) be the Koebe function; i e (-z) conformally onto the w-plane cut from- x+ f2:f2 "(z) Then f3 f3(z) f3" (0) 0 (since f3 fl[f2(z)] fl(x) is univalent), and let maps 0 and f4 f3 (z) f4 (z) f2 by f3(0) E. z f’0----- Combining all these transformations, we can write F(z) fl +_z )_ i) f The function F fl(x + xz (x) z (i- z) 2 (-Ix is close-to-convex. In fact, fl/X+ 1 .z/ z F’(z) + fl(x) (i + x--z)2 F"(z) F’ (z) (i + xz) and fl along the to z E. x l+xz self and takes the origin onto the point x. Define Let negative real axis. Then 2 E onto it- OUASI-CONVEX UNIVALENT FUNCTIONS ,,!x + x+z zF"(z) F’ (z) i +z z x+z f 1 (x+z) (i +xz) I z(, 2xz .- 1,* lz). +i (x+ z) (i +xz) ,,/x+z fl z(- xl )("(-,-z)" + (,,,z) +_ )l fl ,/" [l+xz / z i+ Le z re x ie 1 For and + re + x 2 with (i Re (I re Now, + ie ie r re is < 2 ), I )z fl ei (x+z) (l+x--z) re i8 2 d8 Re Re dq). d8 ) (i + we have correspndlng } (x+z) (.iz) (i- (x + re F"(rei) F’ re 18) (x+z) (t+xz) xre and 81 8 2 with ( flf’()} ; i+ d () (rlei@)} ’i+ rlei f-[ ----’) fl rlei d K. I. NOOR AND D. K. THOMAS 260 and Hence, for 0 1 i8 which shows that < 81 82, 02 iS)I d8 + x+ z) F_’:( re F’ (re i8) Re 1+ rle fM(rl ei i fl (rlei I )i de > - f e K. Now, l+z F’ (z) (i -x) (i +x) 3 i (l+xz) 2 l+xz + l+x l+xZ! 1 i- l-x l+x z) 3 (i Intergrating, we have for z e E, z(l + ’F(z) (i We notice that nce F maps F x in E is not, in general, starlike. between the classes Q 1-’x =z z) z) is arbitrary, we can select x ". in such does not pass through the origin in F(E), which a way that the half-line means i- x onto the w-plane cut along a half-line E the choice of the point - B=I+x, (1 + and K and Because of relationships (3) and (1) S we conclude that, in general, QUASI-CONVEX UNIVALENT FUNCTIONS 261 2 f, the function E defined in by F () f,(z) to Q d belongs 0 C. SOME GROWTH PROBLEMS 3. Clunle and Keogh [i] showed that, if C with f(z) f z + . but not to a z n and n=2 f CE) has finite area. Q with f(z) f Let THEOREM 3. n n o(I) as n and the exponent Is best pos- / We extend this result to quasl-convex functions. slble. n a Then n a o(1) as n + ,, z the index of n a z If f(E) has finite area, then n=2 n n being 5est possible. + We use a modified version of the method of Clunie and Pommerenke PROOF: [2]. By (2), we can write (zf’(z))’ where Re h(z) )0 for z and h(0) E and so with z 2 n a re 21[r n z(zf’ (z))’e-nSd8 0 21[ 21[ 1 n 1[r zg’(z)Re h (z)e _in0 d O- n21an -< 1[r J 0 n zg ’(z) h(z)e -in0 dO. 0 E, 21[ 21[ i n I 2r 0 h(z) > 0 for a Since Re zg’ (z) h(z), 0 < r < i, Cauchy’s formula gives for n > I 1 n Thus, i. 2zg’(z) Reh(z) z(zf’(z))’ i8 g’(z)h(z), [zg’(z) IRe h(z) dO + i 21[r n ’(z) h(z)e In0 dO (3.2) 262 K. I. NOOR AND D. K. THOMAS Re[z(zf,(z)),e-larg zg’(z)] from (3.1), and so inte- grating the first of the above two integrals by parts we have 1 [zg’(z) IRe hCz)dO - 1 Re z(zf’Cz)) ’e-larg z g’ (z)de 2/[ Re ----In wr 0 z(zf,(z)),e-21 Also, zg’(z) h(z) 2 n [an] zf’(z)e -i arg z g’ -< --n r Re arg z g’(z) f’(z)e -i arg g (Z)de e z g (z)) and so (3.2) and (3.3) give ’(z) do(arg Z g’(z)) 2/[ i 2r 1 n r To estimate f(E) II, is finite, where M(r,f) I I 1 2n + z 2n 0.(i) 1 -r xl(reiS.) I- f o(I) as r 1-r + i. z g’(Z)do is regular in (3.4) E and the area of as r- i 2 d Since 0 I1 (’f’ (z))’ e -2i arg say. -12 we note that, since M(r,f’) n+l 0 0 (arg z g’(z)) 2, we have OUASI-CONVEX UNIVALENT FUNCTIONS by parts gives 12 Integrating 263 F (z)e 2 12 -2i arg z n 0 g’ (Z)Re(zg’ (z))’ g’(z) d0 where z Fn(Z) t n zn+if’(z) -nfn(Z), (tf’(t))’dt 0 and z tnf ’(t)dt. f (z) 0 r IF (z) n Now rn+iM( r,f’) < + nM(r,f’) t n dt 0 2rn+iM(r, f’). < 2w Re Since (z’ (z))_’ dO g (z) 2, we have 0 12 < 1- Finally, choosing r na n o(I)i-r 8rn+iM(r,f,)_. o(i) as n i n in as r + (3.4) i as before. the estimates for I I and 12 give and Theorem 3 is proved. / An examination of the proof of Theorem 3 gives COROLLARY: Let f e Q, Then, for n > 2, E r {z: Izl r < i} and A(r) be the area of f(Er). 264 K. I. NOOR AND D. K. THOMAS n[anl 1_)% 0(i) A(I We remark that (3.5) holds for the class S problem (3.5) n , but appears still to be an open for the class K. Denote by C(r) the closed curve which is the image of f(E r) and by L (r) the C(r). length of THEOREM 4. We prove Then, for 0 < r ,e i, Let f e Q. 2/(A(r)) < Further, if A(r) < L(r) for 0 < L(r) PROOF: 2/ (A(/r)) (io < o(i) l/_r r < i, then (lo_ir)1/2 as r +. I. The left hand inequality follows at once. from the isoperlmetrlc in- equality. Since zf’ (z) F(z) f e Q, is close-to-convex. 2 2 [zf LCr) (z) ld0 [(z);d0 0 0 r < M(;),zf’) 2 d;) 0 n=l 2/(A(/r) n=l n / i-1/2 io ig_r Thus from [3,p.45] QUASI-CONVEX UNIVALENT FUNCTIONS If a(r) < 265 for 0 < r < I, then from (3.7) L(r)--" 2 [ lan Irn n=l N < 2 [ lanlr n=l N 2 [ n=l N where / 0 as N / The convex function (3.6) + [ nlanl n=N 21/2 r 2n n=l 1/2 n lanlrn +N logl#r2 Thus L(r) fl (z) 1/2 1 Io o(i) 1 "as r /i. 1 shows that the factor log_z 1/2 lOl-r in is best possible. For f e C for n f e Q, L(r) factor io it is well know that 0(i) M(r) log_r L(r) as < 2M(r). r/l. It follows from (3.6) that The question of whether the can be removed remains open. In conclusion, we remark that other results for the class C can be extended to quasi-convex functions, often with only minor alterations in the proof. objective of this paper has been to introduce the class Q, The exhibit its basic properties and give some results whose proofs are not trivial extensions from the class C. 266 K. I. NOOR AND D. K. THOMAS REFERENCES and KEOGH, F.R., "On starlike and convex schlicht functions", J. London Math.Soc., 35 (1960), 229-233. i. CLUNIE,J.G. 2. CLUNIE, J.G. and POMMERENKE, Ch., "On the coefficients of close-to-convex univalent functions", J. London Math.Soc., 41 (1966),161-165. 3. HAYMAN, W.K., ’Multivalent functions’ Cambridge, 1967 4 LIBERA, R.J "Some classes of regular Soc., 16 (1965), 755-758. univalent functions" Proc.Amer Math