Which quantity is larger, ab or ba, for positive numbers a and b?
As it turns out, sometimes ab is the larger number, other times it is
ba, and it may even be the case that the two quantities are equal,
even though a=b.
Let us start by rewriting the inequality:
abeblnablnaalna<ba<ealnb<alnb<blnb.⇔⇔⇔
(Here we use the fact that x↦ex is a (strictly) increasing
function and that a and b are positive.)
The variables have now been separated and let us introduce
s(x)=xlnx for x>0.
Note how s(a)<s(b)⇔ab<ba for all a,b>0
(and similarly for the relations = and >).
We see that s is continuous on the positive real numbers,
that s(1)=0,
that s(x)<0 for 0<x<1 and
that s(x)>0 for x>1.
This gives us our first, partial, result:
0<a<1≤b⇔s(a)<s(b)⇔ab<ba.
The function s is, in fact, differentiable on the positive reals
and the derivative can give us all the information we need:
s′(x)=x2x1⋅x−lnx⋅1=x21−lnx.
We see that s′(x)>0 for 0<x<e, s′(e)=0, and
s′(x)<0 for x>e.
See a plot of x↦s(x) in Figure 1.
From the information of the derivative of s, we have the following:
0<a<b≤ee≤a<b⇔s(a)<s(b)⇔s(a)>s(b)⇔ab<ba,⇔ab>ba.
To summarize, we now have the following relations, provided that a<b:
0<b≤1
1≤b<e
e<b
0<a≤1
ab<ba
ab<ba
ab<ba
1≤a<e
-
ab<ba
?
e<a
-
-
ab>ba
The only unresolved case is when 1≤a<e<b.
In fact, given the fact that s(x)→0 as x→∞,
we have that s(a)=s(b) has infinitely many solutions with
1<a<e and b>e.
Or, equivalently:
ab=ba has infinitely many solutions for 1<a<e and b>e.
Let us conclude by considering integer a and b.
The case a=1 is trivial (1b<b1 for all b>1).
The case a=2 is especially interesting because 2 is the only integer
between 1 and e.
This means, as argued above, that there is a nontrivial solution to
ab=ba and, in fact, 24=42.
So
23242b<32,=42,>b2for all integer b≥3.
When a≥3 it is more simple as we are in the e<a<b category.
We have
ab>bafor integer a≥3 and integer b>a.
Feel free to leave any question, correction or comment in this
Mastodon thread.