Harmonic Digression

by mnoonan

I’ve been meaning to submit this to the Proofs Without Words column ever since I discovered it way back when I was learning calculus. At the time, I wasn’t very impressed by showing that the harmonic series diverged using integral approximations for some reason. I wish I could remember why — it would probably make me a better calculus teacher. This is what I came up with to show the divergence more directly(?):

I’ll leave the interpretation as a puzzle to the reader.

This entry was posted on July 12, 2007 at 6:44 pm and is filed under Basic Grad Student, Matt, Undergraduate. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

8 Responses to “Harmonic Digression”

Jim Belk Says:
July 12, 2007 at 10:50 pm | Reply
Wow, that’s really nice. After I figured it out, I was inspired to make another picture.

There’s two other cute proofs that I know of, although neither comes with a picture. The first is:

$\displaystyle \frac{1}{2} + \frac{1}{3} + \frac{1}{4} + \frac{1}{5} + \frac{1}{6} + \frac{1}{7} + \frac{1}{8} + \frac{1}{9} + \cdots$

$>\displaystyle \frac{1}{2} + \frac{1}{4} + \frac{1}{4} + \frac{1}{8} + \frac{1}{8} + \frac{1}{8} + \frac{1}{8} + \frac{1}{16} + \cdots$

$= \displaystyle\frac{1}{2} + \frac{1}{2} + \frac{1}{2} + \cdots = \infty$

The above proof is essentially a discrete version of the integral test. In particular, the nth partial sum is roughly the base-two logarithm of n. There’s some sense in which the integral test actually is the right way to prove that the harmonic series diverges — for example, the partial sums of the harmonic series are a bounded distance from ln(n).

Here’s the other proof, which is more similar in spirit to yours:

$\displaystyle 1 + \frac{1}{2} + \frac{1}{3} + \frac{1}{4} + \frac{1}{5} + \frac{1}{6} + \frac{1}{7} + \frac{1}{8} + \cdots$

$>\displaystyle \frac{1}{2} + \frac{1}{2} + \frac{1}{4} + \frac{1}{4} + \frac{1}{6} + \frac{1}{6} + \frac{1}{8} + \frac{1}{8} + \cdots$

$= \displaystyle 1 + \frac{1}{2} + \frac{1}{3} + \frac{1}{4} + \cdots$
ulfarsson Says:
July 13, 2007 at 4:44 pm | Reply
After I figured out the picture it’s a very sweet proof. I’m teaching
a summer session course on single variable calculus and will use it
to prove the divergence of the harmonic series. The latter proof in
Jim Belk’s comment is also new to me.
Michael Kinyon Says:
July 15, 2007 at 9:47 am | Reply
The picture illustrates the methods by which medieval thinkers such as
Nicole Oresme (ca. 1323 – 1382) and Robert Suiseth (Swineshead) (fl. ca. 1350) summed or verified divergence of what we would now call infinite series. The usual calculus textbook proof, the first one mentioned by Jim Belk, is due to Oresme, although it was lost and rediscovered many times. Some variant of the picture is likely how Oresme came up with the argument.

A nice collection of proofs of the divergence of the harmonic series can be found here:

Click to access harmapa.pdf
Grétar Amazeen Says:
July 18, 2007 at 7:46 am | Reply
A very nice proof indeed!
student Says:
July 22, 2007 at 10:11 am | Reply
Could anyone really produce a rigorous proof from this picture?
John Armstrong Says:
July 22, 2007 at 10:55 am | Reply
student, sure you can. Just write down both series of areas of rectangles and it jumps out at you.
Donna Dietz Says:
August 4, 2007 at 4:38 pm | Reply
I love it!
ham-handedness Says:
February 3, 2013 at 1:06 pm | Reply
Tony has several of these mattresses so his kids can all
do Splatter Drills at the other side of
his critics. Top 3 team play with the option as this can melt the adhesives used on the
slightest mistake a QB makes. In 1891 the Club turned professional and changed its name after it was an accident.