An Integral from Hardy
Here's an integral from G. H. Hardy:
It doesn't fit any of the standard undergraduate forms. There's no obvious \(u\)-substitution, integration by parts doesn't help, and a trigonometric substitution turns it into a mess.
The Setup
The trouble is the sum of two square roots in the denominator. With just one square root the integral is a power-rule exercise:
But the sum resists the usual substitutions. Its derivative involves both roots in a form that doesn't cancel cleanly against anything you'd put in the numerator.
Multiply by the Conjugate
Multiply numerator and denominator by the conjugate \(\sqrt{x+a} - \sqrt{x+b}\):
By the difference-of-squares identity \((u+v)(u-v) = u^2 - v^2\), with \(u = \sqrt{x+a}\) and \(v = \sqrt{x+b}\), the denominator becomes:
The \(x\)'s cancel and the denominator is a constant.
Putting It Together
The integral is now:
Each piece is a power-rule integral. Using \(\int (x+c)^{1/2} \, dx = \tfrac{2}{3}(x+c)^{3/2}\):
So:
Sanity Check by Differentiation
Let
Differentiating:
To match the original integrand, multiply top and bottom by \(\sqrt{x+a} + \sqrt{x+b}\):
The numerator is \(a - b\), which cancels, leaving:
The \(a = b\) Case
The answer has \(a - b\) in the denominator, so it's undefined at \(a = b\). But the original integral degenerates there: the denominator becomes \(2\sqrt{x+a}\) and
The singularity is removable — taking \(b \to a\) in the closed form (via L'Hôpital or a Taylor expansion of \((x+b)^{3/2}\) around \(b = a\)) recovers \(\sqrt{x+a} + C\).
References
- G. H. Hardy, A Course of Pure Mathematics, Cambridge University Press. Available on Project Gutenberg.
- Wikipedia, Rationalisation (mathematics).