Finally posted the paper on AoPS here!

It is on a new inequality I discovered while writing a Proofathon problem a while back.

Enjoy, and please give feedback!

Finally posted the paper on AoPS here!

It is on a new inequality I discovered while writing a Proofathon problem a while back.

Enjoy, and please give feedback!

No vote yet

1 vote

×

Problem Loading...

Note Loading...

Set Loading...

## Comments

Sort by:

TopNewestThat you're amazing might be sort of an understatement! – Satvik Golechha · 2 years, 1 month ago

Log in to reply

This is a very interesting read. I especially like the examples which you have chosen to illustrate the technique, as they are otherwise non-trivial. The lemma used in your proof is reminiscent to that of the Rearrangement Inequality.

Can you add it to the Reverse Rearrangement Inequality Wiki page?

P.S. I think it could have a much better name. What you need to do, is find a bunch of "hard" inequalities of which this approach simplifies it, and someday we can call it Daniel's Lemma. I've definitely seen this approach used in several olympiad problems. The 2 variable case is often overlooked by simply expanding terms and canceling, while the 3 variable case starts to show it's usefulness. – Calvin Lin Staff · 2 years, 1 month ago

Log in to reply

Currently, I'm thinking of doing something based on the corollaries, which seem to give the problem maker relative freedom, as we just need to plug in any arbitrary increasing/decreasing function in some domain (which will be specified in the problem) in order to create a problem. However, in most cases the application is pretty obvious. I'm not sure if it is also obvious using other inequalities though. I also want to use the "any permutation" condition to my advantage, as not every inequality allows that.

As an example of an inequality trivial by Reverse Rearrangement: \[\prod_{cyc}\big((y+1)^2+(x+y)(x-y)\big)\ge \big((x+1)(y+1)(z+1)\big)^2\] – Daniel Liu · 2 years, 1 month ago

Log in to reply

– Daniel Liu · 2 years, 1 month ago

I created the Wiki page. I also went along and created the Holder's Inequality page too (for basic holders)Log in to reply

– Calvin Lin Staff · 2 years, 1 month ago

Thanks. If you add more examples to it, I can add it to the featured Wiki list. I will try and add a section on motivation / explanation too.Log in to reply

– Daniel Liu · 2 years ago

Well, now there are three examples. Sorry, I was too lazy to make up new problems, so I just took problems from my paper.Log in to reply

Comment:

Observe that the base case of \(n=1, 2 \) are trivially true for similarly ordered sequences, and we do not require the condition of non-negativity.

I believe the condition on non-negative sequences, can be relaxed to having \( a_3 + b_3 \) be positive (where \( a_1, b_1 \) are the smallest terms in their sequence). I tried tracking down where non-negative sequence was used in the proof, and the only place that I see it is in the induction hypothesis where we have to multiply by \( (a_{k+1} + b_{k+1} ) \). We apply it when \(k = 2 \), hence we just require \( a_3 + b_3 \) to be positive, which makes \( a_ i + b_i \) positive for \( i \geq 3 \).

If this were true, it could lead to very interesting results! We could have \(a_1, a_2, a_3, b_1, b_2 \) as negative values. Thoughts? – Calvin Lin Staff · 2 years ago

Log in to reply

I'll ask Cody Johnson do a quick check of this on Mathematica to see if what you said is actually true. – Daniel Liu · 2 years ago

Log in to reply

Fascinating read. Outstanding job, Daniel! – Ryan Tamburrino · 2 years, 1 month ago

Log in to reply

wow man that was really great. :o kudos!! God..you guys are geniuses.. what am I even doing here among y'all :/ anyways....again..brilliant job :D – Aritra Jana · 2 years, 1 month ago

Log in to reply

– Tytan Le Nguyen · 2 years, 1 month ago

Don't worry, you will get there with persistence and ardor. As for the document, it is very intriguing indeed. Daniel, you have certainly cracked upon a new discovery that will be talked about for a while. I will spread the word immediately to my school and to my friends.Log in to reply

The Reverse Rearrangement Inequality Lower Bound is proven in the Rearrangement Inequality section in Math Olympiad Treasures by Titu Andreescu. – Rahul Saha · 2 years, 1 month ago

Log in to reply

@Daniel Liu Please can you post it on Brilliant. No matter what i tried I have mot been able to find your inequality. P.s. Is it in the paper .If so where? – Sualeh Asif · 2 years, 1 month ago

Log in to reply

– Daniel Liu · 2 years, 1 month ago

It should be right under section 2.Log in to reply

Your status says they didn't accept your submission. Can you tell what was wrong? – Kartik Sharma · 2 years, 1 month ago

Log in to reply

EDIT: Dr. Andreescu said that they were not interested in articles about inequalities to publish right now. Bad timing, I guess. – Daniel Liu · 2 years, 1 month ago

Log in to reply

@Daniel Liu how did you wrote in pdf ( can you explain me , i too want to create a pdf) , i know how to write in latex , can it be converted into pdf? – Sandeep Rathod · 2 years, 1 month ago

Log in to reply

– Daniel Liu · 2 years, 1 month ago

I used a program called TexWorks. It creates PDF's with the nice latex font that is characteristic in mathematical research papers.Log in to reply

Amazing. There you are creating (i.e-discovering) new concepts, here I am struggling to understand even the basic ones :/ (My status) – Krishna Ar · 2 years, 1 month ago

Log in to reply

Hi Daniel, Can you remember what you were thinking when you discovered the inequality? In other words, What was going through your mind when you created something new? – Anup Navin · 2 years, 1 month ago

Log in to reply

I couldn't prove it, so I just gave up on the problem.

A month later, I came back, wondering if I could make a more generalized version of that inequality. It looked so much like the Rearrangement Inequality, it just had to be true. So I created the inequality as it is now, and then managed to prove it. – Daniel Liu · 2 years, 1 month ago

Log in to reply

\[ \prod ( 1 + a_i ) \geq [ 1 + \sqrt[k]{ \prod{a_i} } ] ^k \quad - (1)\]

Substitute in \( a_i = \frac{ \sigma(i) } { i } \times \frac{ n}{m} \), clear out denominators and the result follows.

Note: The simplest approach that I know to prove inequality (1) is to take logarithms and apply Jensens to \( f(x) = \ln (1 + x ) \). – Calvin Lin Staff · 2 years, 1 month ago

Log in to reply

– Anup Navin · 2 years, 1 month ago

Have you considered presenting it to the AMS journal?Log in to reply

– Daniel Liu · 2 years, 1 month ago

I tried submitting it to AwesomeMath journal, but they declined because they weren't looking for inequality articles at the time. And now I already made it public, so I think it's too late already. These types of topics I doubt the AMS journal cares about, since it is purely competition-math style.Log in to reply

Where is the inequality? – Sualeh Asif · 2 years, 1 month ago

Log in to reply

– Daniel Liu · 2 years, 1 month ago

Click on the link above, then download the PDF. Now look right under section 2.Log in to reply

@Daniel Liu . Could you please help me – Sualeh Asif · 2 years, 1 month ago

The link redirects me to AoPS. However I still could not find the pdf link on the page.Log in to reply

As for the inequality, I also posed it on Brilliant Wiki, you can find it here. – Daniel Liu · 2 years, 1 month ago

Log in to reply