Journal of Inequalities in Pure and Applied Mathematics http://jipam.vu.edu.au/ Volume 5, Issue 3, Article 72, 2004 ON LANDAU TYPE INEQUALITIES FOR FUNCTIONS WITH HÖLDER CONTINUOUS DERIVATIVES LJ. MARANGUNIĆ AND J. PEČARIĆ D EPARTMENT OF A PPLIED M ATHEMATICS FACULTY OF E LECTRICAL E NGINEERING AND C OMPUTING U NIVERSITY OF Z AGREB U NSKA 3, Z AGREB , C ROATIA . ljubo.marangunic@fer.hr FACULTY OF T EXTILE T ECHNOLOGY U NIVERSITY OF Z AGREB P IEROTTIJEVA 6, Z AGREB C ROATIA . pecaric@element.hr Received 08 March, 2004; accepted 11 April, 2004 Communicated by N. Elezović A BSTRACT. An inequality of Landau type for functions whose derivatives satisfy Hölder’s condition is studied. Key words and phrases: Landau inequality, Hölder continuity. 2000 Mathematics Subject Classification. 26D15. 1. I NTRODUCTION S.S. Dragomir and C.I. Preda have proved the following theorem (see [1]): Theorem A. Let I be an interval in R and f : I → R locally absolutely continuous function on I. If f ∈ L∞ (I) and the derivative f 0 : I → R satisfies Hölder’s condition (1.1) |f 0 (t) − f 0 (s)| ≤ H · |t − s|α for any t, s ∈ I, 0 where H > 0 and α ∈ (0, 1] are given, then f ∈ L∞ (I) and one has the inequalities: α α+1 α 1 1 · ||f || α+1 · H α+1 2 1 + α 1 α+1 1 α+2 ||f || 1 α+1 α+1 ; 1 + if m(I) ≥ 2 α H (1.2) ||f 0 || ≤ 4·||f || H + 2α (α+1) [m(I)]α m(I) 1 α+1 α+2 1 (1 + 1 ) α+1 , if 0 < m(I) ≤ 2 α+1 ||f || H ISSN (electronic): 1443-5756 c 2004 Victoria University. All rights reserved. 079-04 α 2 L J . M ARANGUNI Ć AND J. P E ČARI Ć where || · || is the ∞-norm on the interval I, and m(I) is the length of I. In our paper we shall give an improvement of this theorem. 2. M AIN R ESULTS Theorem 2.1. Let I be an interval and f : I → R function on I satisfying conditions of Theorem A. Then f 0 ∈ L∞ (I) and the following inequlities hold: α α 1 2 1 + α1 α+1 · ||f || α+1 · H α+1 1 α+1 1 1 ||f || 1 α+1 α+1 1 + ; if m(I) ≥ 2 H α (2.1) ||f 0 || ≤ 2||f || H + α+1 [m(I)]α m(I) 1 α+1 1 1 if 0 < m(I) ≤ 2 α+1 ||fH|| 1 + α1 α+1 , where || · || is the ∞-norm on the interval I, and m(I) is the length of I. In our proof and in the subsequent discussion we use three lemmas. Lemma 2.2. Let a, b ∈ R, a < b, α ∈ (0, 1]. Then the following inequality holds: (2.2) (b − x)α+1 + (x − a)α+1 ≤ (b − a)α+1 , ∀x ∈ [a, b]. Proof. Consider the function y : [a, b] → R given by: y(x) = (b − x)α+1 + (x − a)α+1 . We observe that the unique solution of the equation y 0 (x) = (α + 1) [(x − a)α − (b − x)α ] = 0 is x0 = a+b ∈ [a, b]. The function y 0 (x) is decreasing on (a, x0 ) and increasing on (x0 , b). Thus, 2 the maximal values for y(x) are attained on the boundary of [a, b] : y(a) = y(b) = (b − a)α+1 , which proves the lemma. A generalization of the following lemma is proved in [1]: Lemma 2.3. Let A, B > 0 and α ∈ (0, 1]. Consider the function gα : (0, ∞) → R given by: A (2.3) gα (λ) = + B · λα . λ 1 A α+1 Define λ0 := αB ∈ (0, ∞). Then for λ1 ∈ (0, ∞) we have the bound A if 0 < λ1 < λ0 λ1 + B · λα1 (2.4) inf gα (λ) = α 1 α λ∈(0,λ1 ] (α + 1)α− α+1 · A α+1 · B α+1 if λ1 ≥ λ0 . Proof. We have: A + α · B · λα−1 . λ2 1 A α+1 The unique solution of the equation gα0 (λ) = 0, λ ∈ (0, ∞), is λ0 = αB ∈ (0, ∞). The function gα (λ) is decreasing on (0, λ0 ) and increasing on (λ0 , ∞). The global minimum for gα (λ) on (0, ∞) is: 1 α α α 1 αB α+1 A α+1 (2.5) gα (λ0 ) = A +B = (α + 1)α− α+1 · A α+1 · B α+1 , A αB gα0 (λ) = − J. Inequal. Pure and Appl. Math., 5(3) Art. 72, 2004 http://jipam.vu.edu.au/ O N L ANDAU T YPE I NEQUALITIES FOR F UNCTIONS. . . 3 which proves (2.4). Lemma 2.4. Let A, B > 0 and α ∈ (0, 1]. Consider the functions gα : (0, ∞) → R and hα : (0, ∞) → R defined by: gα (λ) = Aλ + B · λα (2.6) h (λ) = 2A + B λα . α λ 2α Define λ0 := A αB 1 α+1 (2.7) ∈ (0, ∞). Then for λ1 ∈ (0, ∞) we have: inf gα (λ) < λ∈(0,λ1 ] inf gα (λ) = λ∈(0,λ1 ] inf hα (λ) if 0 < λ1 < 2λ0 inf hα (λ) if λ1 ≥ 2λ0 . λ∈(0,λ1 ] λ∈(0,λ1 ] Proof. In Lemma 2.3, we found that the global minimum for gα (λ) is obtained for λ = λ0 . Similarly we find that the global minimum for hα (λ) is obtained for λ = 2λ0 , and its value is equal to the minimal value of gα (λ), i.e. hα (2λ0 ) = gα (λ0 ). The only solution of equation gα (λ) = hα (λ), λ ∈ (0, ∞), is: A λS = B(1 − 2−α ) 1 α+1 , and we can easily check that λ0 < λS < 2λ0 . Thus, for λ1 < λ0 we have gα (λ1 ) < hα (λ1 ) and inf gα (λ) < inf hα (λ), and the rest of the proof is obvious. λ∈(0,λ1 ] λ∈(0,λ1 ] Proof of Theorem 2.1. Now we start proving our theorem using the identity: Z x 0 (2.8) f (x) = f (a) + (x − a)f (a) + [f 0 (s) − f 0 (a)]ds; a, x ∈ I a or, by changing x with a and a with x: Z 0 a f (a) = f (x) + (a − x)f (x) + (2.9) [f 0 (s) − f 0 (x)]ds; a, x ∈ I. [f 0 (s) − f 0 (x)]ds; b, x ∈ I. x Analogously, we have for b ∈ I: Z 0 f (b) = f (x) + (b − x)f (x) + (2.10) b x From (2.9) and (2.10) we obtain: (2.11) b Z 0 f (b) − f (a) = (b − a)f (x) + x [f 0 (s) − f 0 (x)]ds Z x + [f 0 (s) − f 0 (x)]ds; a, b, x ∈ I a and (2.12) f (b) − f (a) 1 f (x) = − b−a b−a 0 J. Inequal. Pure and Appl. Math., 5(3) Art. 72, 2004 Z x b 1 [f (s) − f (x)]ds − b−a 0 0 x Z [f 0 (s) − f 0 (x)]ds. a http://jipam.vu.edu.au/ 4 L J . M ARANGUNI Ć AND J. P E ČARI Ć Assuming that b > a we have the inequality: Z |f (b) − f (a)| 1 b 0 0 0 (2.13) |f (x)| ≤ + |f (s) − f (x)|ds b−a b−a x Z 1 x 0 0 . + |f (s) − f (x)|ds b−a a Since f 0 is of α − H Hölder type, then: Z b Z b 0 0 α (2.14) |f (s) − f (x)|ds ≤ H · |s − x| ds x x Z =H b (s − x)α ds x = (2.15) Z a x H (b − x)α+1 ; α+1 b, x ∈ I, b > x Z x α |f (s) − f (x)|ds ≤ H · |s − x| ds Z xa =H (x − s)α ds 0 0 a H = (x − a)α+1 ; α+1 a, x ∈ I, a < x. From (2.13), (2.14) and (2.15) we deduce: (2.16) |f 0 (x)| ≤ |f (b) − f (a)| b−a H + [(b − x)α+1 + (x − a)α+1 ]; (b − a)(α + 1) a, b, x ∈ I, a < x < b. Since f ∈ L∞ (I) then |f (b) − f (a)| ≤ 2 · ||f ||. Using Lemma 2.2 we obviously get that: (2.17) |f 0 (x)| ≤ 2||f || H + (b − a)α ; b−a α+1 a, b, x ∈ I, a < x < b. Denote b − a = λ. Since a, b ∈ I, b > a, we have λ ∈ (0, m(I)), and we can analyze the right-hand side of the inequality (2.17) as a function of variable λ. Thus we obtain: (2.18) |f 0 (x)| ≤ 2||f || H α + λ = gα (λ) λ α+1 for x ∈ I and for every λ ∈ (0, m(I)). Taking the infimum over λ ∈ (0, m(I)) in (2.18), we get: |f 0 (x)| ≤ (2.19) inf λ∈(0,m(I)) gα (λ). If we take the supremum over x ∈ I in (2.19) we conclude that (2.20) sup |f 0 (x)| = ||f 0 || ≤ x∈I inf λ∈(0,m(I)) gα (λ). Making use of Lemma 2.3 we obtain the desired result (2.1). J. Inequal. Pure and Appl. Math., 5(3) Art. 72, 2004 http://jipam.vu.edu.au/ O N L ANDAU T YPE I NEQUALITIES FOR F UNCTIONS. . . 5 1 h i α+1 Remark 2.5. Denote λ0 = 2 1 + α1 ||fH|| . Comparing the results of Theorem A and Theorem 2.1 we can see that in the case of m(I) ≥ 2λ0 the estimated values for ||f 0 || in both theorems coincide. If 0 < m(I) < 2λ0 the estimated value for ||f 0 || given by (2.1) is better than the one given by (1.2). Namely, using Lemma 2.4 we have: H 4||f || H 2||f || (2.21) + [m(I)]α < + α [m(I)]α ; m(I) ∈ (0, λ0 ] m(I) α + 1 m(I) 2 (α + 1) and α α+1 α 1 H 4||f || 1 + α [m(I)]α ; m(I) ∈ [λ0 , 2λ0 ). · ||f || α+1 · H α+1 < (2.22) 2 1 + m(I) 2 (α + 1) α Remark 2.6. Let the conditions of Theorem 2.1 be fulfilled. Then a simple consequence of (2.11) is the following inequality: H |(b − a)f 0 (x) − f (b) + f (a)| ≤ (b − x)α+1 + (x − a)α+1 ; a, b, x ∈ I, a < x < b. α+1 This result is an extension of the result obtained by V.G. Avakumović and S. Aljančić in [2] (see also [3]). R EFERENCES [1] S.S. DRAGOMIR AND C.J. PREDA, Some Landau type inequalities for functions whose derivatives are Hölder continuous, RGMIA Res. Rep. Coll., 6(2) (2003), Article 3. ONLINE [http://rgmia. vu.edu.au/v6n2.html]. [2] V.G. AVAKUMOVIĆ AND S. ALJANČIĆ, Sur la meilleure limite de la dérivée d’une function assujetie à des conditions supplementaires, Acad. Serbe Sci. Publ. Inst. Math., 3 (1950), 235–242. [3] D.S. MITRINOVIĆ, J.E. PEČARIĆ AND A.M. FINK, Inequalities Involving Functions and Their Integrals and Derivatives, Kluwer Academic Publishers, Dordrecht, Boston, London, 1991. J. Inequal. Pure and Appl. Math., 5(3) Art. 72, 2004 http://jipam.vu.edu.au/