## August 2022

—————————

In my August Diary I stated the following theorem.

:Theoremaandbare positive whole numbers. If (a² +b²) / (ab+ 1) is a whole number, it is a perfect square.

Having stated it, I then added:

I don't recommend you attempt a proof of the theorem … it's really hard. If you just try a few low values foraandb, though, assuming wologa≤b. the first ones you get are (1,1) giving 1², (2,8) giving 2², and (3,27) giving 3².

In all those cases,b=a³. Sure enough, if you substitutea³ forbthe expression (a² +b²) / (ab+ 1) cancels down toa².

Are there any values ofaandbfor which the expression is a whole number butbisnotequal toa³? If there are any, what's the smallest such value ofa?

In a footnote I further added:

[It doesn't seem fair to leave you wondering what that ferociously-difficult proof looks like. To see it presented several different ways — yes, including YouTube presentations — Google "Question 6."]

—————————

**• Solution**

For a proof of the theorem, I can't improve on that footnote. You can get pretty much the same solution set by Googling "IMO 1988" or "Vieta Jumping."

That just leaves open the question: "Are there any values ofaandbfor which the expression is a whole number butbisnotequal toa³?"

Indeed there are. The first few are (8, 30), (27, 240), (30,112), (64,1020), (112,418), … There are references here (although withaandbthe other way round to my usage).