General Mathematics Vol. 17, No. 4 (2009), 29–44 Hankel determinant for p-valently starlike and convex functions of order α Toshio Hayami, Shigeyoshi Owa Abstract For p-valently starlike and convex functions f (z) in the open unit disk U, the upper bounds of the functional |ap+2 − µa2p+1 |, defined by using the second Hankel determinant H2 (n) due to J. W. Noonan and D. K. Thomas (Trans. Amer. Math. Soc. 223(2) (1976), 337-346), are discussed. 2000 Mathematics Subject Classification: Primary 30C45 Key words and phrases: Hankel determinant, p-valently starlike function, p-valently convex function. 1 Introduction Let Ap denote the class of functions f (z) of the form p f (z) = z + ∞ X an z n (p ∈ N = {1, 2, 3, · · · }) n=p+1 29 30 Toshio Hayami, Shigeyoshi Owa which are analytic in the open unit disk U = {z ∈ C : |z| < 1}. Furthermore, let P denote the class of functions p(z) of the form p(z) = 1 + ∞ X ck z k k=1 which are analytic in U and satisfy Re p(z) > 0 (z ∈ U). Then we say that p(z) ∈ P is the Carathéodory function (cf. [1]). If f (z) ∈ Ap satisfies the following condition µ 0 ¶ zf (z) Re > α (z ∈ U) f (z) for some α (0 5 α < p), then f (z) is said to be p-valently starlike of order α in U. We denote by Sp∗ (α) the subclass of Ap consisting of functions f (z) which are p-valently starlike of order α in U. Similarly, we say that f (z) belongs to the class Kp (α) of p-valently convex functions of order α in U if f (z) ∈ Ap satisfies the following inequality µ ¶ zf 00 (z) > α (z ∈ U) Re 1 + 0 f (z) for some α (0 5 α < p). As usual, in the present investigation, we write Sp∗ = Sp∗ (0), Kp = Kp (0), S ∗ (α) = S1∗ (α) and K(α) = K1 (α). Hankel determinant for p-valently starlike and convex functions . . . Remark 1. 31 For a function f (z) ∈ Ap , it follows that zf 0 (z) ∈ Sp∗ (α) p f (z) ∈ Kp (α) if and only if and Z f (z) ∈ Sp∗ (α) z if and only if 0 pf (ζ) dζ ∈ Kp (α). ζ Example 1. f (z) = zp ∈ Sp∗ (α) (1 − z)2(p−α) and f (z) = z p 2 F1 (2(p − α), p; p + 1; z) ∈ Kp (α) where 2 F1 (a, b; c; z) represents the hypergeometric function. In [7], Noonan and an an+1 Hq (n) = det .. . an+q−1 Thomas stated the q–th Hankel determinant as an+1 · · · an+q−1 an+2 · · · an+q (n, q ∈ N = {1, 2, 3, · · · }). .. .. .. . . . an+q · · · an+2q−2 This determinant is discussed by several authors. For example, we can know that the Fekete and Szegö functional |a3 − a22 | = |H2 (1)| and they consider the further generalized functional |a3 − µa22 |, where µ is some real number (see, [2]). Moreover, we also know that the functional |a2 a4 − a23 | is equivalent to |H2 (2)|. Janteng, Halim and Darus [4] have shown the following theorems. 32 Toshio Hayami, Shigeyoshi Owa Theorem 1. Let f (z) ∈ S ∗ . Then |a2 a4 − a23 | 5 1. Equality is attained for functions f (z) = z = z + 2z 2 + 3z 3 + 4z 4 + · · · (1 − z)2 and f (z) = Theorem 2. z = z + z3 + z5 + z7 + · · · . 1 − z2 Let f (z) ∈ K. Then 1 |a2 a4 − a23 | 5 . 8 The present paper is motivated by these results and the purpose of this investigation is to find the upper bounds of the generalized functional |ap+2 −µa2p+1 |, defined by the second Hankel determinant, for functions f (z) in the class Sp∗ (α) and Kp (α), respectively. 2 Preliminary results In order to discuss our problems, we need some lemmas. The following lemma can be found in [1] or [8]. Lemma 1. If a function p(z) = 1 + ∞ P ck z k ∈ P, then k=1 |ck | 5 2 (k = 1, 2, 3, · · · ). Hankel determinant for p-valently starlike and convex functions . . . 33 The result is sharp for ∞ X 1+z =1+ 2z k . p(z) = 1−z k=1 Using the above, we derive Lemma 2. If a function p(z) = p + ∞ P ck z k satisfies the following in- k=1 equality Re p(z) > α (z ∈ U) for some α (0 5 α < p), then (1) |ck | 5 2(p − α) (k = 1, 2, 3, · · · ). The result is sharp for ∞ X p + (p − 2α)z p(z) = 2(p − α)z k . =p+ 1−z k=1 ∞ P p(z) − α ck k = 1+ z . Noting that q(z) ∈ P p−α k=1 p − α and using Lemma 1, we see that Proof. Let q(z) = ¯ ¯ ¯ ck ¯ ¯ ¯ ¯ p − α ¯ 5 2 (k = 1, 2, 3, · · · ) which implies |ck | 5 2(p − α) (k = 1, 2, 3, · · · ). 34 Toshio Hayami, Shigeyoshi Owa Lemma 3. The power series for p(z) = 1 + ∞ P ck z k converges in U to a k=1 function in P if and only if the Toeplitz determinants ¯ ¯ ¯ ¯ ¯ 2 c2 · · · cn ¯ c1 ¯ ¯ ¯ ¯ ¯ c−1 2 c1 · · · cn−1 ¯ ¯ ¯ ¯ ¯ Dn = ¯ c−2 c−1 2 · · · cn−2 ¯ (n = 1, 2, 3, · · · ), ¯ ¯ ¯ .. .. .. .. ¯ .. . ¯ . . . . ¯ ¯ ¯ ¯ ¯ ¯ c−n c−n+1 c−n+2 · · · 2 ¯ where c−k = ck , are all non-negative. They are strictly positive except for m P p(z) = ρk p0 (eitk z), ρk > 0, tk real and tk 6= tj for k 6= j, where p0 (z) = k=1 1+z ; in this case Dn > 0 for n < m − 1 and Dn = 0 for n = m. 1−z This necessary and sufficient condition is due to Carathéodory and Toeplitz, and it can be found in [3]. And then, Libera and Zl/otkiewicz [5] (see, also [6]) have given the following result by using this lemma with n = 2, 3. Lemma 4. If a function p(z) ∈ P, then the representations 2c = c2 + (4 − c2 )ζ 2 1 1 4c = c3 + 2(4 − c2 )c ζ − (4 − c2 )c ζ 2 + 2(4 − c2 )(1 − |ζ|2 )η 3 1 1 1 1 1 1 for some complex numbers ζ and η (|ζ| 5 1, |η| 5 1), are obtained. By virtue of Lemma 4, we have Hankel determinant for p-valently starlike and convex functions . . . Lemma 5. If a function p(z) = p+ ck z k satisfies Re p(z) > α (z ∈ U) k=1 for some α (0 5 α < p), then (2) ∞ P 35 2(p − α)c2 = c21 + {4(p − α)2 − c21 }ζ 4(p − α)2 c3 = c31 + 2{4(p − α)2 − c21 }c1 ζ − {4(p − α)2 − c21 }c1 ζ 2 +2(p − α){4(p − α)2 − c21 }(1 − |ζ|2 )η for some complex numbers ζ and η (|ζ| 5 1, |η| 5 1). ∞ P p(z) − α ck k Proof. Since q(z) = =1+ z ∈ P, replacing c2 and p−α k=1 p − α c2 c3 c3 by and in Lemma 4, respectively, we immediately have the p−α p−α relations of the lemma. We also need the next remark. Remark 2. If f (z) ∈ Sp∗ (α), then there exists a function p(z) = p + ∞ P ck z k such that Re p(z) > α (z ∈ U) and k=1 zf 0 (z) = f (z)p(z) which implies that p+ ∞ X nan z n−p = p + ∞ X n=p+1 n=p+1 à n X ! al cn−l z n−p l=p where ap = 1 and c0 = p. Therefore, we have the follwing relation (3) (n − p)an = n−1 X l=p al cn−l (n = p + 1). 36 3 Toshio Hayami, Shigeyoshi Owa Main results In this section, we begin with the upper bound of |ap+2 −µa2p+1 | for p-valently starlike functions of order α below. Theorem 3. If a function f (z) ∈ Sp∗ (α) (0 5 α < p), then µ ¶ 1 (p − α) {(2(p − α) + 1) − 4(p − α)µ} µ5 2 µ ¶ 1 p+1−α 2 |ap+2 −µap+1 | 5 p−α 5µ5 2 2(p − α) µ ¶ p+1−α µ= (p − α) {4(p − α)µ − (2(p − α) + 1)} 2(p − α) with equality for zp 2(p−α) (1 − z) f (z) = zp (1 − z 2 )p−α µ µ p+1−α 1 µ 5 or µ = 2 2(p − α) 1 p+1−α 5µ5 2 2(p − α) ¶ ¶ . Proof. If f (z) ∈ Sp∗ (α), then we have the equation (3) which means c2 + c21 that ap+1 = c1 and ap+2 = . Thus, by the inequality (1) and the 2 representation (2), we can suppose that c1 = c (0 5 c 5 2(p − α)) without Hankel determinant for p-valently starlike and convex functions . . . 37 loss of generality and we derive ¯ ¯ ¯ c2 + c2 ¯ 2 2 |ap+2 − µap+1 | = ¯¯ − µc ¯¯ 2 ¯ ¯ c2 + {4(p − α)2 − c2 }ζ ¯¯ 1 ¯¯ 2 (1 − 2µ)c + = ¯ 2¯ 2(p − α) = 1 |{2(p − α) − 4(p − α)µ + 1}c2 + {4(p − α)2 − c2 }ζ| 4(p − α) ≡ A(ζ). Applying the triangle inequality, we deduce £ ¤ 1 |(2(p − α) + 1) − 4(p − α)µ| c2 + {4(p − α)2 − c2 } 4(p − α) µ ¶ 1 2(p − α) + 1 2 2 [2(p − α)(1 − 2µ)c + 4(p − α) ] µ5 4(p − α) 4(p − α) A(ζ) 5 = 5 µ ¶ 1 2(p − α) + 1 2 2 [2{2(p − α)µ − (p + 1 − α)}c + 4(p − α) ] µ = 4(p − α) 4(p − α) ¶ µ 1 (p − α) {(2(p − α) + 1) − 4(p − α)µ} µ 5 , c = 2(p − α) 2 µ ¶ 1 2(p − α) + 1 p−α 5µ5 ,c = 0 2 4(p − α) µ ¶ 2(p − α) + 1 p + 1 − α p−α 5µ5 ,c = 0 4(p − α) 2(p − α) ¶ µ p+1−α , c = 2(p − α) . µ= (p − α) {4(p − α)µ − (2(p − α) + 1)} 2(p − α) 38 Toshio Hayami, Shigeyoshi Owa Equality is attained for functions f (z) ∈ Sp∗ (α) defined by zf 0 (z) p + (p − 2α)z = p(z) = f (z) 1−z for the case c1 = c = 2(p − α), ζ = 1 and c2 = 2(p − α), or zf 0 (z) p + (p − 2α)z 2 = p(z) = f (z) 1 − z2 for the case c1 = c = 0, ζ = 1 and c2 = 2(p − α). Taking α = 0 or p = 1 in Theorem 3, we obtain the following corollaries, respectively. If a function f (z) ∈ Sp∗ , then Corollary 1. µ ¶ 1 p {(2p + 1) − 4pµ} µ5 2 µ ¶ 1 p+1 2 p 5µ5 |ap+2 − µap+1 | 5 2 2p µ ¶ p+1 µ= p {4pµ − (2p + 1)} 2p with equality for f (z) = zp 2p (1 − z) µ µ zp (1 − z 2 )p 1 p+1 µ 5 or µ = 2 2p 1 p+1 5µ5 2 2p ¶ . ¶ Hankel determinant for p-valently starlike and convex functions . . . If a function f (z) ∈ S ∗ (α), then ¶ µ 1 µ5 (1 − α) {(3 − 2α) − 4(1 − α)µ} 2 µ ¶ 1 2−α 2 |a3 − µa2 | 5 1−α 5µ5 2 2(1 − α) ¶ µ 2 − α µ= (1 − α) {4(1 − α)µ − (3 − 2α)} 2(1 − α) with equality for µ ¶ z 1 2 − α µ 5 or µ = 2(1−α) 2 2(1 − α) (1 − z) f (z) = ¶ µ z 1 2−α 5µ5 . (1 − z 2 )1−α 2 2(1 − α) Corollary 2. Also, by Corollary 1 and Corollary 2, we readily know Corollary 3. If a function f (z) ∈ S ∗ , 3 − 4µ 2 |a3 − µa2 | 5 1 4µ − 3 with equality for f (z) = z 2 (1 − z) µ µ z 1 − z2 then µ ¶ 1 µ5 2 µ ¶ 1 5µ51 2 (µ = 1) 1 µ 5 or µ = 1 2 ¶ 1 5µ51 . 2 ¶ 39 40 Toshio Hayami, Shigeyoshi Owa Next, in consideration of Remark 1, we derive the upper bounds of |ap+2 − µa2p+1 | for p-valently convex functions. Theorem 4. If a function f (z) ∈ Kp (α) (0 5 α < p), then |ap+2 − µa2p+1 | 5 µ ¶ 2 2 p(p − α) {(2(p − α) + 1)(p + 1) − 4(p − α)p(p + 2)µ} (p + 1) µ5 (p + 1)2 (p + 2) 2p(p + 2) µ ¶ p(p − α) (p + 1)2 (p + 1)2 (p + 1 − α) 5µ5 p+2 2p(p + 2) 2p(p + 2)(p − α) p(p − α) {4(p − α)p(p + 2)µ − (2(p − α) + 1)(p + 1)2 } (p + 1)2 (p + 2) µ ¶ (p + 1)2 (p + 1 − α) µ= 2p(p + 2)(p − α) with equality for µ ¶ 2 2 (p + 1) (p + 1) (p + 1 − α) z p 2 F1 (2(p − α), p; p + 1; z) µ 5 or µ = 2p(p + 2) 2p(p + 2)(p − α) f (z) = ¶ µ ³p p 2´ (p + 1)2 (p + 1)2 (p + 1 − α) p 5µ5 . z 2 F1 , p − α; 1 + ; z 2 2 2p(p + 2) 2p(p + 2)(p − α) Proof. Noting that f (z) ∈ Kp (α) if and only if Hankel determinant for p-valently starlike and convex functions . . . 41 ∞ P zf 0 (z) n = zp + an z n ∈ Sp∗ (α) and using Theorem 3, we see that p n=p+1 p (p − α) {(2(p − α) + 1) − 4(p − α)ν} ¯ ¯ 2 ¯ ¯p + 2 (p + 1) 2 ¯ ¯ p−α ¯ p ap+2 − ν p2 ap+1 ¯ 5 (p − α) {4(p − α)ν − (2(p − α) + 1)} , ¯ ¯ 2 ¯ ¯ (p + 1) 2 that is, that ¯¯ap+2 − νap+1 ¯¯ 5 p(p + 2) µ ¶ p(p − α) {(2(p − α) + 1) − 4(p − α)ν} 1 ν5 p+2 2 µ ¶ p(p − α) 1 p+1−α 5ν5 p+2 2 2(p − α) µ ¶ p(p − α) {4(p − α)ν − (2(p − α) + 1)} p+1−α ν= . p+2 2(p − α) Now, putting (p + 1)2 ν = µ, the proof of the theorem is completed. p(p + 2) When α = 0 or p = 1 in Theorem 4, the following three corollaries are obtained. 42 Toshio Hayami, Shigeyoshi Owa Corollary 4. If a function f (z) ∈ Kp , then 2 µ ¶ 2 2 2 (p + 1) p {(2p + 1)(p + 1) − 4p (p + 2)µ} µ5 (p + 1)2 (p + 2) 2p(p + 2) µ ¶ p2 (p + 1)2 (p + 1)3 2 |ap+2 −µap+1 | 5 5µ5 2 p+2 2p(p + 2) 2p (p + 2) µ ¶ p2 {4p2 (p + 2)µ − (2p + 1)(p + 1)2 } (p + 1)3 µ= 2 (p + 1)2 (p + 2) 2p (p + 2) with equality for z p 2 F1 (2p, p; p + 1; z) f (z) = ³p p ´ z p 2 F1 , p; 1 + ; z 2 2 2 µ µ (p + 1)2 (p + 1)3 µ5 or µ = 2 2p(p + 2) 2p (p + 2) (p + 1)2 (p + 1)3 5µ5 2 2p(p + 2) 2p (p + 2) ¶ ¶ . If a function f (z) ∈ K(α), then µ ¶ (1 − α) 2 {(3 − 2α) − 3(1 − α)µ} µ5 3 3 µ ¶ 2 1−α 2(2 − α) 2 |a3 − µa2 | 5 5µ5 3 3 3(1 − α) µ ¶ (1 − α) 2(2 − α) {3(1 − α)µ − (3 − 2α)} µ= 3 3(1 − α) Corollary 5. with equality for µ ¶ µ ¶ 1 2(2 − α) 2 1 − (1 − z)2α−1 and log µ 5 or µ = 2α − 1 1−z 3 3(1 − α) f (z) = ¶ µ µ ¶ 3 2 2(2 − α) 1 2 , 1 − α; ; z 5µ5 . z 2 F1 2 2 3 3(1 − α) Hankel determinant for p-valently starlike and convex functions . . . Corollary 6. If a function f (z) ∈ K, 1−µ 1 2 |a3 − µa2 | 5 3 µ−1 43 then ¶ µ 2 µ5 3 µ µ 2 4 5µ5 3 3 4 µ= 3 ¶ ¶ with equality for f (z) = z 1−z µ ¶ 1 + z 1 log 2 1−z µ µ 2 4 µ 5 or µ = 3 3 2 4 5µ5 3 3 ¶ ¶ . References [1] P. L. Duren, Univalent Functions, Springer-Verlag, New York, Berlin, Heidelberg, Tokyo, 1983. [2] M. Fekete and G. Szegö, Eine Bemerkung uber ungerade schlichte Funktionen, J. London Math. Soc. 8(1933), 85-89. [3] U. Grenander and G. Szegö, Toeplitz Forms and their Applications, Univ. of California Press, Berkeley and Los Angeles, (1958). [4] A. Janteng, S. A. Halim and M. Darus, Hankel determinant for starlike and convex functions, Int. Journal of Math. Anal. 1(2007), 619-625. 44 Toshio Hayami, Shigeyoshi Owa [5] R. J. Libera and E. J. Zl/otkiewicz, Early coefficients of the inverse of a regular convex function, Proc. Amer. Math. Soc. 85(1982), 225-230. [6] R. J. Libera and E. J. Zl/otkiewicz, Coefficient bounds for the inverse of a function with derivative in P, Proc. Amer. Math. Soc. 87(1983), 251-257. [7] J. W. Noonan and D. K. Thomas, On the second Hankel determinant of areally mean p-valent functions, Trans. Amer. Math. Soc. 223(2) (1976), 337-346. [8] Ch. Pommerenke, Univalent Functions, Vandenhoeck and Ruprecht, Göttingen, (1975). Toshio Hayami Kinki University Department of Mathematics Higashi-Osaka, Osaka 577-8502, Japan e-mail: ha ya to112@hotmail.com Shigeyoshi Owa Kinki University Department of Mathematics Higashi-Osaka, Osaka 577-8502, Japan e-mail: owa@math.kindai.ac.jp