Holder Ineq

Embed Size (px)

Citation preview

  • 7/27/2019 Holder Ineq

    1/3

    International Mathematical Forum, 4, 2009, no. 40, 1993 - 1995

    A Note on Holders Inequality

    Marwan S. Abualrub

    Department of Mathematics, University of JordanP.O. Box 11942, Amman, Jordan

    [email protected]

    W. T. Sulaiman

    Department of Mathematics, Mosul University

    Mosul, Iraq

    Abstract

    In this note an easy method has been introduced to prove Holders

    Inequality.

    Mathematics Subject Classification code: 26D15

    Keywords: Holder Inequality, Rogers Inequality, Cauchy Inequality, Minkowski

    Inequality

    1 Introduction

    In 1888 Rogers proved the Inequality we will be considering in this Note whichhas been proved later in 1889 by Holder and named after him even thoughsome Mathematicians prefer to call it Rogers Inequality.

    In 1994 Hovenier [2] proved sharpening Cauchys Inequality; and in 1995

    Abramovich, Mond, and Pecaric [1] generalized the result of Hovenier toHolders Inequality. Finally, it is vital to mention that Holders Inequalityis used to prove Minkowskis Inequality.

    In this Note we will give an easier proof of Holders Inequality.

  • 7/27/2019 Holder Ineq

    2/3

    1994 M. S. Abualrub and W. T. Sulaiman

    2 Preliminary Notes

    We will first consider the following Cauchys Inequality.

    2.1 Cauchys InequalityLet a = (a1, a2,...,an) and b = (b1, b2,...,bn) are complex vectors. In addi-

    tion, let 1 p ; where1

    p+ 1

    q= 1, then

    n

    j=1

    ajbj

    (n

    j=1

    |aj|p)

    1

    p (n

    j=1

    |bj|q)

    1

    q (1)

    In (1) equality holds if and only if the sequences {|aj|p} and {|bj |

    q} forj = 1, 2,...,n are proportional and arg(ajbj) is independent of j.

    3 Main Results

    These are the main results of the paper.

    Lemma 3.1 If a 0, b 1, s + t = 1, s 0, andt 0. Then

    (1 a)s(1 b)t 1 asbt (2)

    Proof: We will prove the above Lemma by two Methods.Method 1.Consider F(a) = (1 a)s(1 b)t + asbt ; where b is fixed. ThenF

    (a) = 0 implies that ( 1a 1)1s = (1

    b 1)t.

    Which gives that a = b.Now[F

    (a)]a=b = s(s 1)[(1 a)1 + a1] 0.

    This means that F attains its maximum at a = b, and in turns we have[F(a)]a=b = 1 a + 1 = 1.Hence F(a) 1 which gives inequality (2).

    Method 2.

    Consider f(x) = xs1x1

    ; where x 0, s 1, s + t = 1, s 0, and t 0.Observe that f is decreasing with limx1 f(x) = s.Therefore f(x) s.

  • 7/27/2019 Holder Ineq

    3/3

    A note on Holders inequality 1995

    Let x = ab

    ; and a b. (if a b, take x = ba

    ). Thereforeasbt as + bt.Now, replace a and b by 1 a and 1 b respectively we obtain(1 a)s(1 b)t s(1 a) + t(1 b)

    = 1 sa tb

    1 asbt.

    We will now proof Holders Inequality.

    3.2 Holders Inequality

    n1j=1

    ajbj

    (n1j=1

    |aj|p)

    1

    p (n1j=1

    |bj |q)

    1

    q (3)

    Proof:

    We want to prove that the above inequality (3) is true for j = 1, 2, ...,n.Using the above Lemma we havenj=1 ajbj

    n1j=1 ajbj

    + |anbn| (

    n1j=1 |aj|

    p)1

    p (n1

    j=1 |bj |q)

    1

    q + |an| |bn|

    = (n

    j=1 |aj |p |an|

    p)1

    p (n

    j=1 |bj |q |bn|

    q)1

    q + |an| |bn|

    = (n

    j=1 |aj |p)

    1

    p (n

    j=1 |bj |q)

    1

    q

    1 |an|

    pnj=1

    |aj |p

    1p

    1 |bn|qn

    j=1|bj|

    q

    1q

    + |an| |bn|

    (n

    j=1 |aj|p)

    1p (n

    j=1 |bj |q)

    1q

    1 |an||bn|nj=1

    |aj |p

    1pn

    j=1|bj |

    q

    1q

    + |an| |bn|

    = (n

    j=1 |aj |p)

    1

    p (n

    j=1 |bj |q)

    1

    q

    References

    [1] Abramovich, S., Mond, B. and Pecaric, J.E., Sharpening Holders Inequal-ity, J. Math. Anal. Appl. 196 (1995), 1131 - 1134.

    [2] Hovenier, J.W., Sharpening Cauchys Inequality, J. Math. Anal. Appl. 186(1994), 156 - 160.

    Received: March, 2009