1. Limited time only! Sign up for a free 30min personal tutor trial with Chegg Tutors
    Dismiss Notice
Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Homework Help: Inner Product Sum Inequality

  1. Nov 20, 2012 #1
    1. The problem statement, all variables and given/known data
    Let V be a real inner product space, and let v1, v2, ... , vk be a set of orthonormal vectors.
    Ʃ (from j=1 to k)|<x,vj><y,vj>| ≤ ||x|| ||y||

    When is there equality?

    2. Relevant equations

    3. The attempt at a solution

    I've tried using the two inequalities given to us in lectures, namely Cauchy-Schwarz Inequality which states

    |<v,w>| ≤ ||v|| ||w||

    But surely, using this inequality, we get Ʃ (from j=1 to k)|<x,vj><y,vj>| ≤ k(||x|| ||v|| ||y|| ||v|| = k( ||x|| ||y||) since the v are orthonormal!

    I understand this is an inequality, and so obviously the inequality above is a better approximation than the one I've just shown, but I'm not sure where to go.

    The other inequality is Bessel's Inequality which states

    ||v||2 ≥ Ʃ|<v, ei>|2 if ei is a set of orthonormal elements.

    Last edited: Nov 20, 2012
  2. jcsd
  3. Nov 20, 2012 #2
    The space [itex]\mathbb{R}^k[/itex] is an inner product space for the usual inner product. What does the Cauchy-Schwarz inequality say in this special case [itex]\mathbb{R}^k[/itex]??
  4. Nov 21, 2012 #3
    Ok, so according to Wikipedia (I haven't been taught this in lectures), the Cuachy-Schwarz inequality over ℝn is:

    (Ʃ xiyi)2 ≤ Ʃxi2 Ʃyi2

    Do I replace multiplication with inner products? I've tried that, but I must be doing something wrong.

    (Ʃ <x,vj>)2 ≤ Ʃ<x,x> Ʃ<vj,vj> = k||x||2Ʃ<vj,vj> = k||x||2 x k since ||vj||=1 ?
    But then where should I go from here, if here is where I should be?
  5. Nov 21, 2012 #4
    The version of Cauchy-Schwarz that is most standard is actually

    [tex]\sum_{k=1}^n |\alpha_k\beta_k|\leq \sqrt{\sum_{k=1}^n |\alpha_k|^2}\sqrt{\sum_{k=1}^n |\beta_k|^2}[/tex]

    Now, apply this on your original problem

    [tex]\sum_{k=1}^n |<x,v_k><y,v_k>|[/tex]
  6. Nov 21, 2012 #5
    Thank you so much! I think I have it, although it seems very easy, which always seems suspicious to me in maths. Here goes:

    Ʃ |<x,vj><y,vj>| ≤ √(Ʃ<x,vj>2) √(Ʃ<xy,vj>2)

    Then by Bessel's Inequality

    √Ʃ<x,vj>2√Ʃ<xy,vj>2 ≤ √||x||2 √||y||2

    So Ʃ |<x,vj><y,vj>| ≤ ||x|| ||y|| as required!
    Last edited: Nov 21, 2012
  7. Nov 21, 2012 #6
    That's it!!
  8. Nov 21, 2012 #7
    Thank you very much! You've made my night!
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook