Hello all! I was recently reading a proof of the AM-GM inequality by George Polya in my book titled "The Cauchy-Schwarz Master Class" by professor J. Michael Steele. I think it's an extraordinary, beautiful proof, but there's one step that I don't quite understand. I'll post a relatively quick proof, pointing out my area of confusion when I get there.

So, we want to prove that for a sequence of reals \(a_k\) with \(1\leq k\leq n\) and a sequence of reals \(p_k\) with \(1\leq k \leq n\) such that \(\sum_{k=1}^{n} p_k = 1\), that

\[a_1^{p_1}a_2^{p_2}a_3^{p_3}...a_n^{p_n}\leq a_1p_1+a_2p_2+a_3p_3+...+a_np_n\]

Polya's proof begins with the observation that \(1+x\leq e^x\) which can easily be seen graphically with equality occurring only at \(x=0\). If we make the change of variables \(x\mapsto x-1\), our initial observation becomes \(x\leq e^{x-1}\). Applying this to our sequence \(a_k\) we get \[a_k\leq e^{a_k -1}\]

\[a_k^{p_k}\leq e^{a_kp_k - p_k}\] now, we can see that \[G=\prod_{k=1}^{n} a_k^{p_k} \leq \exp \left(\sum_{k=1}^{n} a_kp_k - \sum_{k=1}^{n} p_k\right)\] \[G=\prod_{k=1}^{n} a_k^{p_k} \leq \exp \left(\sum_{k=1}^{n} a_kp_k - 1\right)\] however, we can also see from one of our initial observation that \[A=\sum_{k=1}^{n} a_kp_k \leq \exp \left(\sum_{k=1}^{n} a_kp_k -1\right)\] which is the same bound we just found for \(G\). So we could say

\[A, G \leq \exp \left (\sum_{k=1}^{n} a_kp_k -1 \right )\] So we have related \(A\) and \(G\) by inequality, but we have not separated them. Now we look closer at the case where \(A\) and \(G\) are equal to the expression on the right. The idea of "normalization" then comes to mind. Normalization was discussed earlier in the book, but \(\color{LimeGreen}{\text{this is precisely where I got lost.}}\) What Steele does is the following:

Define a new sequence \(\alpha_k\) with \(1\leq k \leq n\) where we have \[ \alpha_k = \frac{a_k}{A}\] where we have \[A=a_1p_1+a_2p_2+a_3p_3+...+a_np_n\]

Applying our earlier bound for \(G\) for our new variables \(\alpha_k\), we get \[\prod_{k=1}^{n} \alpha_k^{p_k} \leq \exp \left(\sum_{k=1}^{n} \alpha_kp_k-1\right)\] \[\prod_{k=1}^{n} \left(\frac{a_k}{A}\right)^{p_k} \leq \exp \left(\sum_{k=1}^{n} \frac{a_k}{A}p_k-1\right)\] Now, Steele implies that \[\sum_{k=1}^{n} \frac{a_k}{A}p_k = 1\] and it follows that \[\exp \left(\sum_{k=1}^{n} \frac{a_k}{A}p_k-1\right) = 1\]

Now from here, showing that the AM-GM inequality holds is easy. The only place that brings me confusion is the purpose of introducing and defining the sequence \(\alpha_k\) in the way that Steele does. If someone could help me out with this, it would be greatly appreciated! Consider this a recommendation for "The Cauchy-Schwarz Master Class," it's a fantastic book and you'd all love it!

## Comments

Sort by:

TopNewestUpdate: a few minutes later, while proofreading the post for errors, I looked at what I was confused with and I have seen the light! I now understand why the sequence was introduced.

\[\sum_{k=1}^{n} \frac{a_k}{A}p_k = \frac{a_1p_1+a_2p_2+a_3p_3+...+a_np_n}{A} = \frac{A}{A} = 1\]

Seems so obvious now, but I guess that's how it always goes. I will leave the post and proof up because I think it's a great proof for the more general version of the AM-GM Inequality. Also, I feel like this is a prime example of how sometimes, one just needs to take a step back and look at problems from a fresh perspective. – Ryan Tamburrino · 2 years, 3 months ago

Log in to reply

– Jake Lai · 2 years, 3 months ago

A bit of a confusing point for me too, but fantastic proof nonetheless.Log in to reply

– Calvin Lin Staff · 2 years, 3 months ago

This is great! Can you add this to the AM-GM wiki? Thanks!Log in to reply

– Ryan Tamburrino · 2 years, 3 months ago

Sure, I'd love to!Log in to reply