CHAPTER 5 COEFFICIENT PROBLEM FOR CERTAIN CLASSES OF ANALYTIC FUNCTIONS In this chapter, we established some coefficient problems of the functions belonging to the various generalized classes of analytic functions. Besides establishing some new results, we also generalize the results proved by various authors. We also gave a solution to the well known Rogosinski Conjecture. This chapter has been divided into 6 sections : Section 5.1: Introduction and Definitions. Section 5.2: Coefficient Estimate for the class C s A, B; C , D . Section 5.3: The classes C ; A, B and C1 ; A, B . Section 5.4: The classes T ; A, B and T1 ; A, B . Section 5.5: Rogosinski Conjecture. Section 5.6: Conclusions. 5.1 Introduction and Definitions For a univalent function in the class A, it is well known that the n-th coefficient is bounded by n. The bounds for the coefficients give information about the geometric properties of these functions. For example, the bound for the second coefficient of normalized univalent functions readily yields the growth and distortion bounds for 116 univalent functions. The coefficient estimates for various classes were discussed by many authors. Now we are going to define some important classes as below: Definition 5.1.1 Let C s A, B; C , D be the class consisting of functions f z A and satisfying the condition 2 zf z 1 Cz , 1 B A 1 , 1 D C 1, z E , g z g z 1 Dz (5.1.1) where g z z bk z k S s* A, B . (5.1.2) k 2 Obviously Cs 1,1;1,1 Cs , the class of close to convex functions with respect to symmetric points introduced by Das and Singh [34] and Cs 1,1; C , D Cs C , D . To avoid repetition, throughout this chapter, we assume that 1 B A 1 , z E . Definition 5.1.2 By C ; A, B 0 , we denote the class of functions f z A with f z 0 and zf z 1 f z 0 in E 0 satisfying 1 f z zf z 1 Az , g z g z 1 Bz (5.1.3) g z z bk z k S . where k 2 In particular, (i) C ;1,1 C , the class studied by Parvatham and Srinivasan [125]. 117 (5.1.4) C 0; A, B C A, B , the subclass of close -to-star (ii) functions discussed in Chapter 4. (iii) C 1; A, B C A, B , the subclass of close -to-convex functions studied by Mehrok [97]. (iv) C 0;1,1 C , the class of close -to-star functions introduced by Reade [130]. (v) C 1;1,1 C , the class of close-to-convex functions introduced by Kaplan[76]. Definition 5.1.3 Let C1 ; A, B 0 be the class of functions f z A with f z 0 and zf z 1 f z 0 in E 0 satisfying 1 f z zf z 1 Az , h z h z 1 Bz (5.1.5) h z z d k z k K . where (5.1.6) k 2 The following observations are obvious: (i) C1 0; A, B C1 A, B , a subclass of close -to-star functions discussed in Chapter 4. (ii) C1 1; A, B C1 A, B , a subclass of close-to-convex functions discussed in Chapter 3. (iii) C1 1;1,1 C1 , a class introduced by Silverman and Telage [144]. 118 Definition 5.1.4 Let T ; A, B 0 denote the class of functions f z A which satisfying the condition f z z 2 f z 1 Az z , g S , 1 Bz gz gz (5.1.7) There follows the observations: (i) T 0; A, B C A, B . (ii) T ;1,1 T , the class discussed by Al-Abbadi and Darus [6]. (iii) T 0;1,1 C. Definition 5.1.5 Let T1 ; A, B 0 denote the class of functions f z A which satisfying the condition z f z z 2 f z 1 Az , hK . hz hz 1 Bz (5.1.8) There follows the observations: (i) T1 0; A, B C1 A, B . (ii) T1 0;1,1 C1. 5.2 Coefficient Estimate for the class C s A, B; C , D In this section, we derived the coefficient estimates for a generalized class of closeto-convex functions with respect to symmetric points. Our result is a generalization of the 119 results due to Das and Singh [34] and Janteng and Halim [73]. This section is based on the paper “Coefficient estimate for a subclass of close-to-convex functions with respect to symmetric points” published in “Tamkang Journal of Mathematics” (China) 2011, Vol. 42, No. 2, pp 217-222. To prove the main result, we need the following Lemma due to Goel and Mehrok [48] and Lemma 3.2.1 as discussed in Chapter 3. Lemma 5.2.1 Let g z z bk z k S s* A, B , then for n 1 , k 2 b2 n A B n1 A B 2 j n!2 n j 1 and b2 n 1 A B n1 A B 2 j . n!2 n j 1 Theorem 5.2.1. Let f z C s A, B; C , D , then for n 1 , a2n a2n1 C D n1 A B 2 j n!2 n (5.2.1) j 1 n1 A B 1 1 2 A B j C D . 2n 1 2n n 1!2n1 j1 (5.2.2) Proof. By definition of subordination, it follows that f z C s A, B; C , D if and only if 2 zf z 1 Cwz P z , w z U , g z g z 1 Dwz (5.2.3) where P z 1 pk z k . k 1 120 (5.2.4) As g S s* A, B , it follws that 2zg z g z g z K z for z E where (5.2.5) K z 1 d1 z d 2 z 2 d 3 z 3 ... On equating the coefficients of like powers of z in (5.2.5), we get 2b2 d1 , 2b3 d 2 , 4b4 d3 b3 d1 , 4b5 d 4 b3 d 2 , (5.2.6) (5.2.7) Continuing in this way , we have 2nb2 n d 2 n 1 b3 d 2 n 3 b5 d 2 n 5 ... b2 n 1 d1 , (5.2.8) 2nb2 n 1 d 2 n b3 d 2 n 2 b5 d 2 n 4 ... b2 n 1d 2 . (5.2.9) From (5.2.3) and (5.2.4) , we have z 2a2 z 2 3a3 z 3 ... 2na2 n z 2 n 2n 1 a2 n 1 z 2 n 1 ... z b3 z 3 b5 z 5 ... b2 n 1 z 2 n 1 b2 n 1 z 2 n 1 ... . 1 p1 z p2 z 2 ... p2 n z 2 n p2 n 1 z 2 n 1 ... . (5.2.10) On equating the coefficients in (5.2.10), we obtain 2a2 p1 3a3 p2 b3 , , 4a4 p3 b3 p1 , 5a5 p4 b3 p2 b5 , (5.2.11) (5.2.12) and so on 2na2 n p2 n 1 b3 p2 n 3 b5 p2 n 5 ... b2 n 1 p1 , 2n 1 a2 n1 p2 n b3 p2 n2 b5 p2 n4 ... b2 n1 p2 b2 n1 . 121 (5.2.13) (5.2.14) Using Lemma 3.2.1 and equation (5.2.11) , we get 2 a2 C D 3 a3 C D , A B . 2 Again applying Lemma 3.2.1 and using equations (5.2.6) and (5.2.7), we obtain from (5.2.12) 4 a4 C D A B 2 , 5 a5 2 A B 2 A B 4 C D 8 . It follows that (5.2.1) and (5.2.2) hold for n 1, 2 . We now prove (5.2.1) and (5.2.2) by induction. (5.2.13) and (5.2.14) in conjunction with Lemma 3.2.1 yield, a2n C D 1 n 1 b 2n k 1 2 k 1 (5.2.15) and a2n1 1 n1 1 b2k 1 b2n1 . C D 2n 1 k 1 (5.2.16) Again by using Lemma 3.2.1 in (5.2.9) , we have b2n1 A B n1 1 b 2n k 1 2k 1 . (5.2.17) From (5.2.16) and (5.2.17) , we obtain a 2 n 1 A B 1 n1 b . 1 C D 2k 1 2n 1 2n k 1 We assume that (5.2.1) and (5.2.2) holds for k 3,4,..., n 1 . Using Lemma 5.2.1 in (5.2.15) and (5.2.16), we obtain 122 (5.2.18) a2n C D 1 n1 A B k 1 ( A B 2 j ) . 2n k!2 k k 1 j 1 (5.2.19) and a 2 n 1 n 1 A B k 1 1 ( A B 2 j ) C D 1 k 2n 1 j 1 k 1 k!2 A B 1 n1 A B k 1 ( A B 2 j ) . 2n k!2 k k 1 j 1 (5.2.20) In order to prove (5.2.1) , it is sufficient to show that C D 1 m1 A B k 1 ( A B 2 j ) = C D m 1 ( A B 2 j ) , (m 3,4,..., n) . (5.2.21) m k 2m k 1 k!2 j 1 m!2 j 1 (5.2.21) is valid for m 3 . Let us suppose that (5.2.21) is true for all m , 3 m n 1 . Then from (5.2.19) , we have C D 1 n1 A B k 1 ( A B 2 j ) 2n k 1 k!2 k j 1 n 1 C D n 2 A B k 1 = ( A B 2 j ) 1 n 2n 1 k 1 k!2 k j 1 C D . A B n2 ( A B 2 j ) n 1!2 n1 2n j 1 = n 1 . C D n2 ( A B 2 j ) n n 1!2 n 1 j 1 C D . A B n2 ( A B 2 j ) n 1!2 n1 2n j 1 = C D n2 ( A B 2 j ) A B 2n 1 2n n 1!2 n1 j 1 123 = C D n1 ( A B 2 j ). n!2 n j 1 Thus (5.2.21) holds for m n and hence (5.2.1) follows. Now from (5.2.20) , we have a 2 n 1 A B 1 n1 A B k 1 A B 2 j . 1 C D 2n 1 2n k 1 k!2 k j 1 (5.2.22) From (5.2.21) , we have n 1 1 k 1 A B k 1 ( A B 2 j ) = k k!2 j 1 n 1 1 ( A B 2 j) n 1!2 n1 j 1 (5.2.23) From (5.2.22) and (5.2.23), we have a2n1 A B 1 n1 A B 2 j 1 C D 2n 1 2n n 1 !2n1 j1 which proves (5.2.2). Putting A C 1 and B D 1 in the above result , we get the following result: Corollary 5.2.1 Let f z be schlicht and starlike with respect to symmetric points in the unit disc E , having the form f z z ak z k , then k 2 an 1 for any natural number n . This result was proved by Das and Singh [34]. For C 1 and D 1 , we have the following result for the class C s A, B . 124 Corollary 5.2.2 Let f C s A, B , then for n 1 , a2n1 n1 1 A B 1 A B j 2 2 . 2n 1 2n n 1 !2n1 j1 Remark- This result was proved by Janteng and Halim [73], but the result given by them is not justified. 5.3 The classes C ; A, B and C1 ; A, B In this section, we obtained the coefficient estimate and integral representation formulae for the functions in the classes C ; A, B and C1 ; A, B . The results due to Parvatham and Srinivasan [125], Reade [130], Mehrok [97] and Silverman and Telage [144] follows as special cases from our results. Also some of the results proved in Chapter 3 and Chapter 4 are particular cases of these results. 5.3.1 Integral Representation Formula for the class C ; A, B Theorem 5.3.1 A function f z is in C ; A, B if and only if there exists a starlike function g z and a function Pz which is analytic and has a positive real part in E such that f z z 1 2 1 t Pt g t dt 1 z 0 Pz g z for 0. for 0. Proof. As f z C ; A, B 0, then 1 f z zf z Pz g z , 125 (5.3.1) where g z S and Pz is a function with positive real part in E. 1 2 1 If 0, then multiply (5.3.1) by z 1 and integrating with respect to z , we get 1 z f z 1 1 z 2 t Pt g t dt 0 or f z On choosing g z z 1 2 1 t Pt g t dt. 1 z 0 z 1 Az and P z in the above result, we obtain: 2 1 Bz 1 z 1 f z z 1 1 z 1 1 t 1 At 1 t 1 Bt dt 2 for 0. 0 If 0, then f z P z g z . The converse is immediate. 5.3.2 Coefficient Estimate for the class C ; A, B Theorem 5.3.2 If f z C ; A, B , then 2 (n 1) A B an n , n 2 . 2n 1 (5.3.2) Proof. As f z C ; A, B , then by definition of subordination 1 f z zf z 1 Awz 1 pk z k , wz U . g z g z 1 Bw z k 1 Using (1.2.1) and (5.1.4) in (5.3.3), we get 1 z a 2 z 2 a3 z 3 ... a n z n ... z 1 2a 2 z 3a3 z 2 ... na n z n 1 ... 126 (5.3.3) z b2 z 2 b3 z 3 ... bn z n ... 1 p1 z p 2 z 2 ... p n 1 z n 1 ... . (5.3.4) On equating the coefficients of z n in (5.3.4), we have 1 n a n p n 1 b2 p n 2 ... bn 1 p1 bn . Using the fact bn n for n 2 and Lemma 3.2.1, we get 1 n a n A B 1 2 ... n 1 n. Hence the Theorem 5.3.2. The bounds are sharp being attained for the function obtained in Theorem 5.3.1. For A 1, B 1 ,Theorem 5.3.2 gives the following result due to Parvatham and Srinivasan [125]. Corollary 5.3.1 If f z C , then an n2 , n 2. n 1 On putting 0, Theorem 5.3.2 yields Corollary 5.3.2 If f z C A, B , then (n 1) A B a n n 1 , n 2. 2 For 1, Theorem 5.3.2 gives the following result proved by Mehrok [97]. Corollary 5.3.3 If f z C A, B , then an 1 (n 1) A B , n 2. 2 On putting 0 and A 1, B 1 Theorem 5.3.2 observes the result due to Reade [130]. 127 Corollary 5.3.4 If f z C , then a n n 2 , n 2. 5.3.3 Integral Representation Formula for the class C1 ; A, B Theorem 5.3.3 A function f z is in C1 ; A, B if and only if there exists a convex function hz and a function Pz which is analytic and has a positive real part in E such that f z 1 z 2 1 t Pt ht dt 1 z 0 Pz h z for 0, for 0. Proof. The proof of this theorem is similar to Theorem 5.3.1. On choosing hz z 1 Az and P z in the above result, we obtain: 1 Bz 1 z 1 f z 1 1 z 1 1 t 1 At 0 1 t 1 Bt dt for 0. z 5.3.4 Coefficient Estimate for the class C1 ; A, B Theorem 5.3.4 If f z C1 ; A, B , then an 1 (n 1) A B , n 2. n 1 Proof. Using the technique of Theorem 5.3.2, we can easily prove this theorem. Equality holds for the functions obtained in Theorem 5.3.3. Putting 0, Theorem 5.3.4 follows: 128 Corollary 5.3.5 If f z C1 A, B , then a n 1 (n 1) A B , n 2. For 1, Theorem 5.3.4 gives the following result. Corollary 5.3.6 If f z C1 A, B , then an 1 (n 1) A B , n 2. n n For 1 and A 1, B 1 Theorem 5.3.4 yields the result of Silverman and Telage [144]. Corollary 5.3.7 If f z C1 , then 1 a n 2 , n 2. n 5.4 The classes T ; A, B and T1 ; A, B This section is concerned with the integral representation formulae and coefficient estimates for the functions in the classes T ; A, B and T1 ; A, B . The results due to Mehrok [97] and Silverman and Telage [144] are particular cases of our theorems. 5.4.1 Integral Representation Formula for the class T ; A, B Theorem 5.4.1 A function f z is in T ; A, B if and only if there exists a starlike function g z and a function Pz which is analytic and has a positive real part in E such that 129 t 1 c c1 u g u .Pu du dt c1 t 0 0 z f z where c 1 (5.4.1) 1, 0. g t f z Pt dt. t 0 z If 0 , then (5.4.2) Proof. Let f z T ; A, B 0, then zf z z 2 f z Pz , g z g z (5.4.3) where Pz is a function with positive real part in E. zf z z zf z 1 P z . g z g z (5.4.3) is equivalent to Dividing by and putting c 1 1, we get zf z z zf z c 1 c P z . g z g z (5.4.4) Multiplying (5.4.4) by z c 1 g z , we obtain cz c1 zf z z c zf z 1 c z c1 g z P z which implies zf z .z 1 c z g z Pz . c c 1 Therefore on integrating (5.4.5) with respect to z , we obtain (5.4.1). Conversely, if f z satisfies (5.4.1), then it is easy to see that f z T ; A, B . 130 (5.4.5) On choosing g z z 1 Az and Pz 2 1 Bz 1 z in (5.4.1) , we obtained the following function of T ; A, B : z f z 1 c t 0 c 1 t u c 1 Au 1 du dt , c 1 . 2 0 1 u 1 Bu 5.4.2 Coefficient Estimate for the class T ; A, B Theorem 5.4.2 Let f z T ; A, B , then 2 (n 1) A B an , n 2 . 2n 1 (5.4.6) Proof. Since f z T ; A, B , then by definition of subordination f z z 2 f z 1 Aw z z 1 pk z k , w z U . 1 Bw z gz gz k 1 (5.4.7) Using (1.2.1) and (5.1.4) in (5.4.7), we get z 1 2a 2 z 3a 3 z 2 ... na n z n 1 ... z 2 2a 2 6a 3 z ... nn 1a n z n 2 ... z b2 z 2 b3 z 3 ... bn z n ... 1 p1 z p 2 z 2 ... p n 1 z n 1 ... . (5.4.8) On equating the coefficients of z n in (5.4.8), we have n1 n a n p n 1 b2 p n 2 ... bn 1 p1 bn . Using the fact bn n for n 2 and Lemma 3.2.1, we get n1 n a n A B 1 2 ... n 1 n. This proves the Theorem 5.4.2. The bounds are sharp being attained for the function obtained in Theorem 5.4.1. For 0, Theorem 5.4.2 leads to the result of Corollary 5.3.3. 131 5.4.3 Integral Representation Formula for the class T1 ; A, B Theorem 5.4.3 A function f z is in T1 ; A, B if and only if there exists a convex function hz and a function Pz which is analytic and has a positive real part in E such that t 1 c c1 f z c1 u hu .Pu du dt t 0 0 z where c 1 1, 0. ht f z Pt dt. t 0 z If 0 , then Proof. The proof of this theorem is similar to Theorem 5.4.1. On choosing hz z 1 Az and P z in the above result, we obtain the following 1 z 1 Bz z f z 1 c t 0 c 1 t u c 1 Au 1 du dt , c 1 . 0 1 u 1 Bu 5.4.4 Coefficient Estimate for the class T1 ; A, B Theorem 5.4.4 Let f z T1 ; A, B , then an 1 (n 1) A B , n2 . nn 1 Proof. Using the technique of Theorem 5.4.2, we can easily prove this theorem. Equality holds for the functions obtained in Theorem 5.4.3. For 0, Theorem 5.4.4 leads to the result of Corollary 5.3.6. 132 5.5 Rogosinski Conjecture In this section, we established a solution to the well known conjecture due to Rogosinski. This section is based on the paper “A solution of Rogosinski Conjecture” published in “Antarctica Journal of Mathematics” (India) 2011, Vol.8, No. 1, pp 37-40. In 1916, Bieberbach [21] made the following conjecture: If f z z a k z k is analytic and univalent in the unit disc E z : z 1, then k 2 an n . Rogosinski [133] introduced the concept of subordination and conjectured: If k 2 k 2 f z z a k z k is univalent in E and g z z bk z k be analytic in E such that g z f z , then bn n . This result is known as Rogosinski Conjecture. To prove this result, we need the following result due to Rogosinski [133]: Lemma 5.5.1 (Rogosinski Inequality) n 0 n 0 If f z a n z n F z An z n , then n ak k 0 Theorem 5.5.1 If 2 n Ak . 2 k 0 k 2 k 2 f z z a k z k is univalent in E and g z z bk z k be analytic in E such that g z f z , then bn n . 133 Proof. Using Lemma 5.5.1, we have n bk 2 k 1 n ak 2 k 1 which gives b1 b2 b3 ... bn 2 2 2 2 a1 a 2 a3 ... a n . 2 2 2 2 Using Bieberbach conjecture, we have b1 b2 b3 ... bn 2 2 2 12 2 2 3 2 ... n 2 2 = nn 12n 1 . 6 (5.5.1) Similarly b1 b2 b3 ... bn 1 2 2 2 2 n 1n2n 1 . 6 (5.5.2) Subtracting (5.5.2) from (5.5.1) , we have bn 2 n2 . Hence bn n . 5.6 Conclusions The results derived in this chapter are very useful. In Section 5.2, the generalized class of close-to- convex functions with respect to symmetric points has been investigated and coefficient estimate for this class has been established. By giving particular values to the parameters A, B, C and D, one can easily obtain the results proved earlier by Das and 134 Singh [34] and Janteng and Halim [73]. Also in Sections 5.3 and 5.4, we investigated the subclasses of Alpha- close-to-star functions and obtained sharp coefficient estimate and extremal functions. The results already proved by Parvatham and Srinivasan [125], Mehrok [97], Silverman and Telage [144] and Reade [130] are particular cases of our results. In Section 5.5, a solution to the well known conjecture made by Rogosinski [133], has been established. 135