Oh, that's a very good point! I never really thought of that technique as this kinda trick, but yeah. For some reason I always thought of it as squaring the integral (or the contour integration method).
This was pretty cool. Once on an exam in a differential equations class, the instructor made a mistake in the problem that required integrating sec(x).
In my first Calculus course I had failed to memorize some trig integral, and rederived it on the spot. Unfortunately what I came up with differed from the usual by some complex trigonometric substitution. The grader saw a ton of work, not the usual answer, and didn't bother differentiating it to demonstrate that it was correct, and marked me wrong.
I was ticked off about that for some time. (Wouldn't have changed my grade, but it was the principle of the thing.)
Haha, yeah. That one's pretty cool too. There are a couple of slightly more easier-to-see ways, but they still involve seemingly arbitrary multiplications of cosine, etc (or product-to-sum rules).
Man... integrals. Love/hate relationship.
I was equally inspired during undergrad and grad school by Feynman's "unusual tools", and ended up checking out a copy of Advanced Calculus by Woods which was apparently the book he used to learn calculus in high school. If I recall correctly, Woods goes over integration by both differentiation and integration under the integral sign, including some interesting ways to set boundary conditions to get the answer you want. It's a nice book if you ever get the chance.
http://mathworld.wolfram.com/GaussianIntegral.html
You make the single integral into a double integral, and then the double integral turns out to be easy to do using polar coordinates.