The AM - GM Inequality states that the arithmetic mean of a set of numbers $a_1,a_2...a_n$, will always be greater than or equal to the geometric mean of the same set.

More formally,

$\sum_{i=1}^n \frac{a_i}{n} \geq \sqrt[n]{\prod_{i=1}^n{a_i}}$

This has many applications, 1 of them specifically being to prove the Cauchy-Schwarz Inequality, which states that for a set $a_1,a_2...a_n$, and a set $b_1,b_2...b_n$, that the sum of the squares of each set multiplied together is greater than or equal to the square of the dot product of the two sets.

$\sum_{i=1}^n{a_i^2} \space \times \space \sum_{i=1}^n{b_i^2} \space\space\geq\space\sum_{i=1}^n{a_ib_i}$

Within my textbook, there is a proof for the CS inequality that applies the AM-GM inequality, and I am not completely sure what was done to achieve their result.

Below is the proof.

Let $A =\sqrt{\sum_{i=1}^n{a_i^2}} \space$ and $B =\sqrt{\sum_{i=1}^n{b_i^2}} \space$
Applying the AM-GM inequality, we have
$\sum_{i=1}^n{\frac{a_ib_i}{AB}} \leq \sum_{i=1}^n{\frac{1}{2}(\frac{a_i^2}{A^2}+\frac{b_i^2}{B^2})} = 1$
By cross multiplying...

The proof then continues into basic algebra to prove the inequality. My question, is where that crazy sigma equation came from.

  • $\begingroup$ Welcome to MSE. Please, be more careful choosing the tags. This is not a question of Measure Theory. $\endgroup$
    – jjagmath
    Apr 7, 2021 at 0:16
  • $\begingroup$ I apologize. I saw the word sigma when I searched Algebra and clicked. $\endgroup$
    – Smartsav10
    Apr 7, 2021 at 0:18
  • $\begingroup$ The LHS should be $\sum_{i=1}^n \frac{a_i b_i}{AB}$. Is that what is confusing you? $\endgroup$
    – jjagmath
    Apr 7, 2021 at 0:20
  • $\begingroup$ @jjagmath Thank you, but unfortunately no. I fixed it in the question. $\endgroup$
    – Smartsav10
    Apr 7, 2021 at 0:22

1 Answer 1


The mean in this case is between two terms: $a_i^2/A^2$ and $b_i^2/B^2$. If you write the AM-GM inequality for these terms for any given $i\in\{1,2,3...,n\}$, you get $\sqrt{ a_i^2/A^2 \times b_i^2/B^2 }\leq \frac{1}{2}(a_i^2/A^2 + b_i^2/B^2) $.

Summing both sides of this inequality across all $i$ preserves the inequality, and the resulting right hand side simplifies to one, giving the desired crazy sigma equation.

  • $\begingroup$ So are you saying that the terms are $\frac{a_1^2}{A^2}, \frac{a_2^2}{A^2}...\frac{a_n^2}{A^2}$. Also including the $b$ counterparts? $\endgroup$
    – Smartsav10
    Apr 7, 2021 at 0:28
  • $\begingroup$ For any particular $i\in \{1,2,3,...\}$, there are only two relevant terms involved in computing the means: $a^2_i/A^2$ and $b^2_i/B^2$. Write the AM-GM inequality for the mean of these two terms. Summing both sides of this inequality across all $i$ preserves the inequality. Does that make sense? $\endgroup$ Apr 7, 2021 at 0:31
  • $\begingroup$ Oh!!!!! I just realized that when I tried it out. It is multiple AMGM inequalities summed together. $\endgroup$
    – Smartsav10
    Apr 7, 2021 at 0:36
  • $\begingroup$ Essentially, yes-added more detail to my explanation $\endgroup$ Apr 7, 2021 at 0:37
  • $\begingroup$ Thank you so much! $\endgroup$
    – Smartsav10
    Apr 7, 2021 at 0:37

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.