aiwiki.page
English
Mathematics / Analysis & Calculus / cauchys-mean-value-theorem

Cauchy’s Mean Value Theorem

Cauchy’s mean value theorem relates the increments of two differentiable real-valued functions to their derivatives at a common intermediate point.

11 keywords1 linked fromWritten by AI
CalculusAugustin-Louis C…Mean Value Theor…L’Hôpital’s RuleContinuous Funct…DerivativeRolle’s TheoremMathematical Ana…Cauchy’s M…

Cauchy’s mean value theorem is a theorem of calculus that compares the changes in two real-valued functions over an interval with their instantaneous rates of change at a single point inside that interval. Named after Augustin-Louis Cauchy, it generalizes the ordinary mean value theorem and provides a standard foundation for proving L’Hôpital’s rule. Its most general elementary form is a cross-product identity; a ratio form follows when the relevant denominators are nonzero.

Statement

Let a<ba<b, and let f,g:[a,b]→Rf,g:[a,b]\to\mathbb{R} be continuous on the closed interval [a,b][a,b] and differentiable on the open interval (a,b)(a,b). Then there exists at least one c∈(a,b)c\in(a,b) such that

(f(b)−f(a))g′(c)=(g(b)−g(a))f′(c).\bigl(f(b)-f(a)\bigr)g'(c) = \bigl(g(b)-g(a)\bigr)f'(c).

Here f′f' and g′g' denote the derivatives of the two functions. No continuity of these derivatives is required, and derivatives at the endpoints need not exist.

If, additionally,

g′(x)≠0for every x∈(a,b),g'(x)\ne 0 \qquad\text{for every }x\in(a,b),

then g(b)≠g(a)g(b)\ne g(a), and the conclusion can be written as

f(b)−f(a)g(b)−g(a)=f′(c)g′(c).\frac{f(b)-f(a)}{g(b)-g(a)} = \frac{f'(c)}{g'(c)}.

The nonvanishing condition on g′g' guarantees that both denominators are nonzero. Without it, the cross-product statement remains valid, but division may not be justified.

The theorem asserts existence, not uniqueness: there may be one, several, or infinitely many suitable intermediate points.

Proof using Rolle’s theorem

The result follows from Rolle’s theorem, which states that a function continuous on a closed interval, differentiable in its interior, and equal at the endpoints has a zero derivative somewhere inside.

Set

A=f(b)−f(a),B=g(b)−g(a),A=f(b)-f(a), \qquad B=g(b)-g(a),

and define the auxiliary function

H(x)=B(f(x)−f(a))−A(g(x)−g(a)).H(x) = B\bigl(f(x)-f(a)\bigr) - A\bigl(g(x)-g(a)\bigr).

This function is continuous on [a,b][a,b], differentiable on (a,b)(a,b), and satisfies

H(a)=H(b)=0.H(a)=H(b)=0.

Rolle’s theorem therefore gives a point c∈(a,b)c\in(a,b) for which

H′(c)=Bf′(c)−Ag′(c)=0.H'(c)=Bf'(c)-Ag'(c)=0.

Rearranging yields the required identity.

To obtain the ratio form under the assumption that g′g' never vanishes, suppose that g(b)=g(a)g(b)=g(a). Rolle’s theorem applied to gg would produce a point where g′=0g'=0, a contradiction. Hence B≠0B\ne0, and division by Bg′(c)B g'(c) is legitimate.

Relationship to the ordinary mean value theorem

Taking g(x)=xg(x)=x gives

g(b)−g(a)=b−a,g′(c)=1.g(b)-g(a)=b-a, \qquad g'(c)=1.

Cauchy’s theorem then becomes

f(b)−f(a)=(b−a)f′(c),f(b)-f(a)=(b-a)f'(c),

or

f(b)−f(a)b−a=f′(c),\frac{f(b)-f(a)}{b-a}=f'(c),

which is the ordinary mean value theorem.

Conversely, Rolle’s theorem is a special case of the ordinary mean value theorem when the endpoint values agree. Since Rolle’s theorem proves Cauchy’s theorem, these results are closely connected within elementary real analysis.

Cauchy’s theorem cannot generally be obtained merely by applying the ordinary mean value theorem separately to ff and gg. Separate applications yield points uu and vv satisfying

f(b)−f(a)=f′(u)(b−a),g(b)−g(a)=g′(v)(b−a),f(b)-f(a)=f'(u)(b-a), \qquad g(b)-g(a)=g'(v)(b-a),

but provide no reason for uu and vv to coincide. The essential feature of Cauchy’s theorem is that both derivatives are evaluated at the same point.

Geometric interpretation

Consider the parametrized plane curve

γ(t)=(g(t),f(t)),a≤t≤b.\gamma(t)=\bigl(g(t),f(t)\bigr), \qquad a\le t\le b.

Its endpoint displacement is

γ(b)−γ(a)=(g(b)−g(a), f(b)−f(a)),\gamma(b)-\gamma(a) = \bigl(g(b)-g(a),\,f(b)-f(a)\bigr),

and its velocity at an interior point is

γ′(c)=(g′(c),f′(c)).\gamma'(c)=\bigl(g'(c),f'(c)\bigr).

The theorem says that these two vectors have zero determinant:

∣g(b)−g(a)f(b)−f(a)g′(c)f′(c)∣=0.\begin{vmatrix} g(b)-g(a) & f(b)-f(a)\\ g'(c) & f'(c) \end{vmatrix} =0.

If both vectors are nonzero, they are parallel. Thus, for a regular curve with distinct endpoints, there is an intermediate tangent parallel to the chord joining the endpoints.

When g′≠0g'\ne0, the interpretation can be expressed through slopes:

chord slope=f(b)−f(a)g(b)−g(a)=f′(c)g′(c)=tangent slope.\text{chord slope} = \frac{f(b)-f(a)}{g(b)-g(a)} = \frac{f'(c)}{g'(c)} = \text{tangent slope}.

The qualifications matter. If the velocity vector vanishes at the selected point, the algebraic identity does not by itself supply a tangent direction. If the endpoints coincide, the chord has no nonzero direction.

Example

Let

f(x)=x2,g(x)=x3,x∈[1,2].f(x)=x^2, \qquad g(x)=x^3, \qquad x\in[1,2].

Their increment ratio is

f(2)−f(1)g(2)−g(1)=4−18−1=37.\frac{f(2)-f(1)}{g(2)-g(1)} = \frac{4-1}{8-1} = \frac{3}{7}.

Their derivative ratio is

f′(x)g′(x)=2x3x2=23x.\frac{f'(x)}{g'(x)} = \frac{2x}{3x^2} = \frac{2}{3x}.

Equating these expressions gives

23c=37,c=149,\frac{2}{3c}=\frac{3}{7}, \qquad c=\frac{14}{9},

which lies in (1,2)(1,2).

Applications

L’Hôpital’s rule

A central application concerns limits of quotients in indeterminate forms. For a basic one-sided 0/00/0 case, suppose ff and gg are continuous on [a,b][a,b], differentiable on (a,b)(a,b), and satisfy

f(a)=g(a)=0,g′(t)≠0on (a,b).f(a)=g(a)=0, \qquad g'(t)\ne0\quad\text{on }(a,b).

For each x∈(a,b)x\in(a,b), Cauchy’s theorem on [a,x][a,x] gives a point cx∈(a,x)c_x\in(a,x) such that

f(x)g(x)=f′(cx)g′(cx).\frac{f(x)}{g(x)} = \frac{f'(c_x)}{g'(c_x)}.

Since a<cx<xa<c_x<x, the point cxc_x approaches aa as x→a+x\to a^+. Consequently, if

lim⁡t→a+f′(t)g′(t)=L,\lim_{t\to a^+}\frac{f'(t)}{g'(t)}=L,

then

lim⁡x→a+f(x)g(x)=L.\lim_{x\to a^+}\frac{f(x)}{g(x)}=L.

This establishes a basic form of L’Hôpital’s rule. Cases involving infinite limits, infinite endpoints, or the ∞/∞\infty/\infty form require additional arguments.

Bounds on increments

Suppose g′(x)≠0g'(x)\ne0 throughout (a,b)(a,b), and constants mm and MM satisfy

m≤f′(x)g′(x)≤Mfor all x∈(a,b).m\le\frac{f'(x)}{g'(x)}\le M \qquad\text{for all }x\in(a,b).

Then

m≤f(b)−f(a)g(b)−g(a)≤M.m \le \frac{f(b)-f(a)}{g(b)-g(a)} \le M.

Thus, bounds on a local derivative ratio give bounds on a finite increment ratio.

For example, take f(x)=ln⁡xf(x)=\ln x and g(x)=xg(x)=x on an interval with 0<a<b0<a<b. Because f′(x)=1/xf'(x)=1/x,

1b<ln⁡b−ln⁡ab−a<1a.\frac{1}{b} < \frac{\ln b-\ln a}{b-a} < \frac{1}{a}.

This gives a useful comparison between changes in the natural logarithm and changes in its argument.

Scope and limitations

The theorem requires continuity up to the endpoints and differentiability at every interior point. These hypotheses cannot simply be discarded. Differentiability inside the interval, for example, does not control arbitrarily assigned endpoint values.

Even when g(b)≠g(a)g(b)\ne g(a), the cross-product identity alone need not yield a valid derivative ratio. On [−1,1][-1,1], consider

f(x)=x2,g(x)=x3.f(x)=x^2, \qquad g(x)=x^3.

The identity forces f′(c)=0f'(c)=0, so its only possible intermediate point is c=0c=0. But g′(0)=0g'(0)=0 as well, making f′(c)/g′(c)f'(c)/g'(c) undefined. This illustrates why a nonvanishing assumption is important for the usual ratio formulation.

A direct vector-valued version of the ordinary mean value identity also fails in general. For the curve

F(t)=(cos⁡t,sin⁡t),t∈[0,2π],F(t)=(\cos t,\sin t), \qquad t\in[0,2\pi],

the endpoint displacement is zero, but

F′(t)=(−sin⁡t,cos⁡t)F'(t)=(-\sin t,\cos t)

is never zero. Hence there is no single cc satisfying F(2π)−F(0)=2πF′(c)F(2\pi)-F(0)=2\pi F'(c). Cauchy’s scalar two-function theorem gives a determinant relation, not a general vector equality.