## A curiosity

From Futility Closet, a fun blog of random tidbits I enjoy reading, comes the following curious sequence of equations, attributed to J.A.H. Hunter:

I managed to extend this pattern for a few more digits before I got bored. Does it continue forever or does it eventually stop? Is there any deeper mathematical explanation lurking behind this supposed “curiosity”? What’s so special about $f(x) = 2x^2 - 1$? Do patterns like this exist for other functions?

About these ads
This entry was posted in arithmetic, modular arithmetic, number theory, pattern. Bookmark the permalink.

### 20 Responses to A curiosity

1. Taking f(x)=x^2-6 I find 264453123^2-6 = 69935454 264453123 and so on for truncations of this number, but there doesn’t appear to be such a pattern for f(x)=x^2-1

2. A Brook says:

It starts here: http://oeis.org/A151752 Hint: look at the leading digits.

• Brent says:

Fascinating! However, the pattern seems not to continue: the 13th term in that sequence is 3773193359375, suggesting that $2 * 377319335937^2 - 1$ ought to end in 377319335937, but it is actually equal to 284739762543877319335937 instead! In fact, the right digit to use here is 8: $2 * 877319335937^2 - 1 = 1539378434417 877319335937$.

What’s the connection to all-odd-digit number divisible by powers of 5, and why does the pattern stop?

• A Brook says:

One (obvious) thing I see here is that the pattern basically says that a number of the form $2n^2-n-1$ is divisible by a large power of 10. Maybe I’ll have time to look at more details tonight.

3. Brenton Bostick says:

It seems to stop at 88384389877319335937

4. Brenton Bostick says:

Actually, my code had a bug that did not account for the first zero in the sequence! It certainly seems to go on forever:

…5548722601433298591064947197786574820441719412220106407559357953405720
0770206342858319449522214639585592951097866225470440776483509324976437
1295716500836135134846344163506159929966088384389877319335937

• Brent says:

Cool, what language did you use to implement that? Can you share your code?

• Brenton Bostick says:

I used Mathematica. Here is the code I used:

digs = {7};
Module[{d},
Do[
d = Select[Range[0, 9],
Module[{t},
t = digs~Prepend~#;
FromDigits[t] ==
Mod[2 FromDigits[t]^2 – 1, 10^(Length[digs] + 1)]] &][[1]];
digs = digs~Prepend~d
,
{200}
];
];
FromDigits[digs]

It assumes that there is always some digit that will work, and selects the first one that works in each iteration. It seems to work so far.

• Matt Gardner Spencer says:

Your assumption is valid. There is always exactly one number that works to continue the sequence. What’s more, that number depends completely on the digit of 2n^2-1 appearing immediately before n appears. I haven’t found a better way to find that digit yet other than simply computing 2n^2-1 out to enough digits though.

5. Brenton Bostick says:

It seems like there should something to say concerning 10-adic numbers, but I’m not sure what.

• Brent says:

I thought of that too, but didn’t follow it up because I thought the p-adic numbers are only defined when p is prime. But after your comment I dug a little deeper and realized that although the n-adic numbers for n composite do not form a *field* they do still form a *ring*, which is just fine for the purposes of this problem.

I conjecture the following: $2x^2 - 1 = x$, considered as an equation in the ring of 10-adic numbers, has a unique solution, of which 7, 37, 937, … are all suffixes.

• Brent says:

Hmm, this is not quite right (re: unique solution), since the 10-adics admit the solutions 1 and (-1/2) = …99999.5, just like the reals. But it seems the 10-adics also admit one other, weirder solution…?

• Brent says:

An interesting link sent to me by Matt Gardner Spencer, which seems relevant: http://www.numericana.com/answer/p-adic.htm#decimal

6. Brent says:

I wonder if it is possible to come up with an algorithm for generating the infinite sequence without using brute-force guessing. One simple observation: any such algorithm will necessarily use an unbounded amount of memory; a bounded-memory algorithm would produce a periodic sequence, but periodic 10-adic numbers are equivalent to negative rationals, and we know there are no rational solutions other than 1 and -1/2.

• Matt Gardner Spencer says:

Let k_i denote the number just to the left of the gap (running down the pyramid) in the i-th row. Then d_{i+1} (the next digit to be prepended) is 7k_i mod 10. So getting the next digit from the previous line is straightforward. But finding k_{i+1} seems to require both k_i and the digit to the left of k_i, at least the way I’m approaching it. That means it’s really no better than simply computing k_{i+1} with brute force ((2x^2-x-1)/10^{i+1} mod 10)

• Brent says:

Hmm, well, but at least that gives a way of directly computing each successive digit (rather than having to guess and check).

• Brent says:

Using this method I have computed the first 20,000 digits (takes about 15 seconds on my computer). You can find them here.

Comments are closed.