On the antiderivative of inverse functions

Abstract

One of the basics of calculus is the following proposition: If $F$ and $G$ are antiderivatives of two real functions $f$ and $g$ resp., then an antiderivative of $\lambda f+g$ is $\lambda F + G$. It may be surprising that such systematic an integration formula exists for the inverse of a function $f$, a fact that seems to have been discovered for the first time by Laisant in 1905, and seems not to be well known. More precisely, if $f$ is an invertible real function, and if $F$ is an antiderivative of $f$, then the antiderivative of $f^{-1}$ is $xf^{-1}(x)-F\circ f^{-1}(x)+C$. Laisant, and other authors after him, assumes that $f^{-1}$ is differentiable, in which case the proof of this formula is immediate. Recently, it has been shown by Key that this additional assumption is unnecessary. In this paper, we give two different proofs of this result. The first proof, of geometrical spirit, relies to Fubini's theorem, while the second proof, purely analytic, is based on the Stieltjes integral.