I present here a detailed proof of the equality of the Schwarz inequality given a condition that two vectors are linearly dependent.

Let $p$ be an inner product of an inner product space $V$. Then $|p(\alpha,\beta)|=\| \alpha \| \|\beta\| $ if and only if $\{\alpha,\beta\}$ is linearly dependent.


1 Answer 1



$(\Longleftarrow):$ Suppose that $\{\alpha,\beta\}$ is linearly dependent. Without loss of generality, assume that $\beta \neq 0$. (Since if $\beta=0$, the equation $|p(\alpha,\beta)|=\| \alpha \| \|\beta\| $ is trivially satisfied,i.e,$|p(\alpha,0)|=0=\| \alpha \| \|0\| $ ) Then $\alpha =c\beta$ for some scalar $c$. Thus \begin{align} |p(\alpha,\beta)|&=|p(c\beta,\beta)| \\ &= |c p(\beta,\beta)| \\ &=|c||p(\beta,\beta)| \\ &=|c|\|\beta\|^2 \\ &= (|c|\|\beta\|)\|\beta\| \\ &=||c\beta\|\|\beta\| \\ &=\|\alpha\|\|\beta\|. \end{align}

$(\implies$): Let $|p(\alpha,\beta)|=\| \alpha \| \|\beta\| $. If $\beta=0$, then $\{\alpha,\beta\}$ is linearly dependent. Suppose that $\beta \neq 0$.
Let $$\gamma=\alpha - \frac{p(\alpha,\beta)}{p(\beta,\beta)}\cdot \beta$$ Then $\gamma$ is orthogonal to $\beta$ (This is easy to verify). Thus, \begin{align} |p(\alpha,\beta)|^2&=|p(\gamma+\frac{p(\alpha,\beta)}{p(\beta,\beta)}\cdot \beta,\beta)|^2 \\ &=\|\gamma+\frac{p(\alpha,\beta)}{p(\beta,\beta)}\cdot \beta\|^2\|\beta\|^2 \\ &=\|\gamma\|^2+\left|\frac{p(\alpha,\beta)}{p(\beta,\beta)}\right|^2\cdot \|\beta\|^2\|\beta\|^2 ,\mbox{by Pythagorean Theorem}\\ &=\|\gamma\|^2+\|\alpha\|^2\|\beta\|^2 \end{align} This implies that $\|\gamma\|^2=0$ so that $\gamma=0$. Hence $$\alpha = \frac{p(\alpha,\beta)}{p(\beta,\beta)}\cdot \beta$$ Therefore, $\{\alpha,\beta\}$ is linearly dependent.

  • 1
    $\begingroup$ Minor nit: In ($\Longleftarrow$) you also assume (without loss of generality) that $\beta \neq 0$ when you say that $\alpha = c \beta$ for some scalar $c$. // In the first part of your proof $c$ is a scalar and in the second part $c$ is a vector, maybe use $\gamma$ instead of $c$ in the second part? $\endgroup$
    – Martin
    Mar 14, 2013 at 17:40
  • $\begingroup$ I get your point Martin. I'll make an edit. Thanks much. $\endgroup$ Mar 14, 2013 at 17:46
  • 1
    $\begingroup$ Thanks for the edits, nice job, +1! If I were in an overly pedantic mood, I'd ask you to add a short explanation why you don't lose generality by making the assumption $\beta \neq 0$ in the first part of the argument; And maybe you could say that you use that $\gamma$ is perpendicular to $\beta$ in the third displayed equation of your argument (Pythagorean theorem). $\endgroup$
    – Martin
    Mar 14, 2013 at 17:52
  • $\begingroup$ Thank you very much Martin, I am in an overly pedantic mood so I reflected all your suggestions. LOL, thanks... $\endgroup$ Mar 14, 2013 at 18:05
  • $\begingroup$ :) You're welcome. I missed a typo: it should be $\lVert \gamma \rVert^2 \lVert \beta \rVert^2$ instead of just $\lVert \gamma \rVert^2$ twice. $\endgroup$
    – Martin
    Mar 15, 2013 at 7:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.