Differentiation, Rigorously

Cauchy mean value theorem

Imagine two travelers running along their own routes over the same time window. The ordinary mean value theorem compares one traveler's progress against the clock. Cauchy's version instead compares the two travelers against each other, and finds an instant where their speeds stand in the same ratio as their total displacements.

Statement: if f and g are continuous on [a, b] and differentiable on (a, b), then there exists c in (a, b) with (f(b) - f(a)) g'(c) = (g(b) - g(a)) f'(c). When g(b) is not g(a) and g'(c) is not 0, this rearranges to (f(b) - f(a)) / (g(b) - g(a)) = f'(c) / g'(c) — a ratio of derivatives equaling a ratio of total changes. Taking g(x) = x recovers the ordinary mean value theorem.

The standard proof applies Rolle's theorem to the auxiliary function h(x) = (f(b) - f(a)) g(x) - (g(b) - g(a)) f(x), which satisfies h(a) = h(b); then h'(c) = 0 is exactly the assertion. Note that the symmetric product form needs no nonvanishing hypothesis, whereas the fraction form requires g(b) not equal g(a); writing it as a single c for both functions (not two separate points) is what gives the theorem its power.

A naive 'proof' that applies the ordinary MVT to f and g separately fails, because it produces two different points c1 and c2; Cauchy's theorem crucially gives a single common c.

Also called
extended mean value theorem, generalized MVT广义中值定理廣義中值定理