Rethinking Rigor in Calculus: The Role of the Mean Value Theorem (Q3130167)

The author argues that the mean value theorem (MVT) is a technical existence theorem introduced in most traditional calculus texts without rigorous proof and then used to prove intuitively obvious statements such as ``if your speedometer reads zero, you are not going anywhere.'' This practice of ``assuming the nonobvious to prove the obvious'' is said to give mathematics a bad name. Although the author expresses serious reservations about the need for formal theorems and proofs in a standard calculus course, for those who do feel that need he suggests replacing the MVT with the increasing function theorem (IFT) as the theoretical cornerstone: If \(f'\geq 0\) on an interval then \(f\) is (weakly) increasing on that interval. The article goes on to offer a suitable proof of the IFT without assuming the MVT, and to illustrate consequences, one of which is: If \(m\leq f'(x)\leq M\) on the interval \([a,b]\) then \(m(x-a) \leq f(x)- f(a)\leq M(x-a)\) for all \(x\) in \([a,b]\). Another is a nice derivation of the Taylor polynomial with Lagrange's form of the error bound. (It is Lagrange's original proof, in fact.) Connections to error behavior for various common numerical integration rules and to the fundamental theorem of calculus are discussed. (See the following review for an opposing view.).