Graham's Number: When Mathematics Stopped Fitting in My Head
I first ran into Graham's number in college.
I was studying mathematics and found it by accident. At first I took it for one more giant number. A million, a billion, a trillion, followed by a ridiculous wall of zeros.
So I started digging. One link led to another, then came the formulas, Ramsey theory, hypercubes, Knuth's arrows.
When I understood the scale, I just sat there.
It wouldn't fit.
Big isn't the point
There is no largest number. Add one to any number and you have a larger one. You learn that early.
Graham's number is interesting for a different reason. It is finite, precisely defined, and came from an actual mathematical problem. Nobody invented it by typing zeros until the keyboard gave up.
That was what caught me.
The problem comes from Ramsey theory. Take the vertices of an n-dimensional cube, connect every pair, and color every line red or blue. Then ask a simple question: starting at what dimension does every possible coloring force four coplanar vertices whose six connections are all the same color?
The exact answer wasn't known. Ronald Graham and Bruce Rothschild proved that a threshold exists, and Graham's number became famous as an upper bound for the problem.
The question sounds almost harmless.
The number does not.
Arrows instead of zeros
Ordinary decimal notation becomes useless fast.
A googol is 1 followed by 100 zeros:
10100
A googolplex is 10 raised to the power of a googol:
10(10100)
You couldn't write its full decimal form inside the observable universe. There isn't enough room.
When I first read that, I thought this had to be about as far as numbers could go.
It wasn't even close.
For numbers at this scale, mathematicians use Knuth's up-arrow notation.
One arrow means ordinary exponentiation:
3 ↑ 3 = 27
Two arrows build a power tower:
3 ↑↑ 3 = 3(33)
That gives:
7,625,597,484,987
More than seven trillion.
Three arrows repeat the double-arrow operation. Four arrows repeat the triple-arrow operation. Every added arrow changes the level of the operation itself.
Now we can write the first term in the sequence:
g1 = 3 ↑↑↑↑ 3
Even g1 is already beyond every physical comparison I could find. Stars, atoms, Planck volumes, the whole observable universe. None of it helps.
But g1 is not Graham's number.
Sixty-four steps
The next term, g2, is written as a 3, followed by g1 arrows, followed by another 3.
Not four arrows. Not a million arrows.
g1 arrows.
Then comes g3, where the number of arrows equals g2.
After that, g4 uses g3 arrows.
This continues until g64.
The last term, g64, is Graham's number.
At that point I stopped trying to picture it. Until then I had been building power towers in my head, filling space with digits, comparing numbers with stars and particles. Then it became clear that every comparison had already failed.
My imagination was done.
The number wasn't.
Why it stayed with me
The size alone wasn't what fascinated me. The strange part was that a person could define this number exactly even though no person could ever write it out or picture it.
It has a precise construction and a place in a proof. We can reason about its properties even though we will never see more than a negligible fraction of its decimal representation.
That was the moment mathematics changed for me.
It stopped looking like a collection of formulas from a textbook. I saw that it could keep working after human intuition had laid down and refused to get back up.
And here's the funny part. Mathematics doesn't treat Graham's number as mystical. It is an ordinary finite integer.
Inconveniently large, sure.
Still an integer.
Since then, mathematicians have pushed the upper bound for the original problem far below Graham's number. The known lower bound is 13, and a 2019 paper reduced the upper bound to less than 2 ↑↑↑ 5.
That doesn't change what happened when I first found it.
I remember thinking I had finally understood the scale.
Then I understood that I couldn't.
I just sat there.
Sources
Graham's Number, Wolfram MathWorld
Improved Lower Bound on a Euclidean Ramsey Problem
Further Improving of Upper Bound on a Geometric Ramsey Problem
· · ·
More notes and essays at denisostapenko.com.
Comments
Post a Comment