Dear Uncle Colin,

Which is larger, x=10001001 or y=10011000?

- Calculating Oversize Massive Powers And Realising… Eurgh

Hi, COMPARE, and thanks for your message!

These feel like they ought to be a similar size! In fact, the first one, x is the larger. Let’s show that in several ways.

Numbers nearby

Let’s think about the ‘nearby’ number A=10001000. Clearly, both of our numbers are bigger than this.

Now, x=1000A, and y=1.0011000A. A sensible question is, is 1.0011000 bigger than 1000? No - in fact, (1+11000)1000 is very close to e.

So x is the larger, by a factor of about 1000/e.

What’s xy?

If we divide the pair, we get 1000100110011000, which is (10001001)1000×1000.

Take logs: this is 1000ln(111000)+ln(1000).

Since ln(1t)t, this is roughly 1+ln(1000), and ln(1000)>1.

(This is morally the same as the other way.)

Multivariable calculus

I’m not recommending this way, but it popped into my head.

Consider the function f(X,Y)=XY=eYln(X)

Differentiating partially:

  • δδXf=YXf(X,Y)
  • δδYf=ln(X)f(X,Y)

When X=Y=1000, these work out to be 10001000 and 10001000ln(1000).

The derivative with respect to Y is much larger, so increasing the power makes the value of the function increase far more quickly than increasing the base – again, x is the larger.

One more way

(This one is @realityminus3’s idea, adapted slightly.)

Consider f(x)=xln(x). Its derivative is f(x)=ln(x)1(ln(x))2, which is positive for all x>e.

That implies f(1001)>f(1000), so 1001ln(1001)>1000ln(1000).

In turn, 1001ln(1000)>1000ln(1001), and as a result 10001001>10011000.

(Trust RM3 to point me towards the neatest way. Good rat.)


I hope one of those methods makes sense to you!

- Uncle Colin