logo

Proof of Taylor's Theorem 📂Calculus

Proof of Taylor's Theorem

Theorem1

If a function f(x)f(x) is continuous at [a,b][a,b] and differentiable up to nn times at (a,b)(a,b), then there exists ξ(a,b)\xi \in (a,b) that satisfies

f(b)=k=0n1(ba)kk!f(k)(a)+(ba)nn!f(n)(ξ)=f(a)+(ba)f(a)++(ba)n1(n1)!f(n1)(a)+(ba)n(n)!f(n)(ξ) \begin{align*} f(b) =& \sum_{k=0}^{n-1} {{(b-a)^{k}\over{k!}}{f^{(k)}( a )}} + {(b-a)^{n}\over{n!}}{f^{(n)}(\xi)} \\ =& {f(a)} + {(b-a)f ' (a)} + \cdots + {(b-a)^{n-1}\over{(n-1)!}}{f^{(n-1)}(a)} + {(b-a)^{n}\over{(n)!}}{f^{(n)}(\xi)} \end{align*}

Explanation

This theorem, which is widely used throughout mathematics, has lent its name to the Taylor series. In terms of being differentiated nn times, it can be considered a generalization of the mean value theorem.

Customarily, when using the Taylor theorem, cc is not used, but ξ\xi is used instead.

Proof

Let’s consider

f(b):=(ba)00!f(a)+(ba)11!f(a)+(ba)22!f(a)++(ba)n1(n1)!f(n1)(a)+(ba)n(n)!c \begin{align*} f(b) :=& {(b-a)^0\over{0!}}{f(a)} + {(b-a)^1\over{1!}}{f ' (a)} + {(b-a)^2\over{2!}}{f '' (a)} \\ &+ \cdots + {(b-a)^{n-1}\over{(n-1)!}}{f^{(n-1)}(a)} + {(b-a)^{n}\over{(n)!}}c \end{align*}

Showing c=f(n)(ξ)c={f^{(n)}(\xi)} concludes the proof. Let’s define function gg as follows.

g(x):=f(b)+f(x)+(bx)11!f(x)+(bx)22!f(x)++(bx)n1(n1)!f(n1)(x)+(bx)n(n)!c=f(b)+k=0n1(bx)k(k)!f(k)(x)+(bx)n(n)!c \begin{align*} g(x):=& -f(b) + f(x) + {(b-x)^1\over{1!}}{f ' (x)} + {(b-x)^2\over{2!}}{f '' (x)} \\ & + \cdots + {(b-x)^{n-1}\over{(n-1)!}}{f^{(n-1)}(x)} + {(b-x)^{n}\over{(n)!}}c \\ =& -f(b) + \sum_{k=0}^{n-1}{(b-x)^{k}\over{(k)!}}{f^{(k)}(x)} + {(b-x)^{n}\over{(n)!}}c \end{align*}

gg is continuous at [a,b][a,b], differentiable at (a,b)(a,b), and by definition of cc, it is g(b)=g(a)=0g(b)=g(a)=0.

Rolle’s Theorem: If function f(x)f(x) is continuous at [a,b][a,b], differentiable at (a,b)(a,b), and f(a)=f(b)f(a)=f(b), then there exists at least one ξ\xi in (a,b)(a,b) that satisfies f(ξ)=0f ' (\xi)=0.

If h(x)h(x) is set as

h(x):=[k=0n1(bx)k(k)!f(k)(x)]=[(bx)0(0)!f(0)(x)+(bx)1(1)!f(1)(x)++(bx)n1(n1)!f(n1)(x)]=[f(x)+(bx)1(1)!f(1)(x)++(bx)n1(n1)!f(n1)(x)]=f(1)(x)[f(1)(x)+(bx)1(1)!f(2)(x)]+[(bx)1(1)!f(2)(x)+(bx)2(2)!f(3)(x)]+[(bx)n3(n3)!f(n2)(x)+(bx)n2(n2)!f(n1)(x)]+[(bx)n2(n2)!f(n1)(x)+(bx)n1(n1)!f(n)(x)]=(bx)n1(n1)!f(n)(x) \begin{align*} h(x):=& \left[ \sum_{k=0}^{n-1}{(b-x)^{k}\over{(k)!}}{f^{(k)}(x)} \right] ' \\ =& \left[ {(b-x)^{0}\over{(0)!}}{f^{(0)}(x)} + {(b-x)^{1}\over{(1)!}}{f^{(1)}(x)} + \cdots + {(b-x)^{n-1}\over{(n-1)!}}{f^{(n-1)}(x)} \right] ' \\ =& \left[ f (x) + {(b-x)^{1}\over{(1)!}}{f^{(1)}(x)} + \cdots + {(b-x)^{n-1}\over{(n-1)!}}{f^{(n-1)}(x)} \right] ' \\ =& f^{(1)} (x) - \left[ f^{(1)} (x) + {(b-x)^{1}\over{(1)!}}{f^{(2)}(x)} \right] \\ & + \left[ - {(b-x)^{1}\over{(1)!}} f^{(2)} (x) + {(b-x)^{2}\over{(2)!}}{f^{(3)}(x)} \right] \\ & \vdots \\ & + \left[ - {(b-x)^{n-3}\over{(n-3)!}} f^{(n-2)} (x) + {(b-x)^{n-2}\over{(n-2)!}}{f^{(n-1)}(x)} \right] \\ & + \left[ - {(b-x)^{n-2}\over{(n-2)!}} f^{(n-1)} (x) + {(b-x)^{n-1}\over{(n-1)!}}{f^{(n)}(x)} \right] \\ =& {(b-x)^{n-1}\over{(n-1)!}}{f^{(n)}(x)} \end{align*}

and since g(x)=0+h(x)(bx)n1(n1)!c\displaystyle g ' (x) = 0 + h(x) - {(b-x)^{n-1}\over{(n-1)!}}c, by Rolle’s theorem,

g(ξ)=h(ξ)(bξ)n1(n1)!c=(bξ)n1(n1)!f(n)(ξ)(bξ)n1(n1)!c=0 \begin{align*} g ' (\xi) =& h(\xi) - {(b-\xi)^{n-1}\over{(n-1)!}}c \\ =& {(b-\xi)^{n-1}\over{(n-1)!}}{f^{(n)}(\xi)} - {(b-\xi)^{n-1}\over{(n-1)!}}c \\ =& 0 \end{align*}

there exists at least one ξ\xi in (a,b)(a,b) that satisfies it. Thus,

(bξ)n1(n1)!f(n)(ξ)(bξ)n1(n1)!c=0    (bξ)n1(n1)!f(n)(ξ)=(bξ)n1(n1)!c    f(n)(ξ)=c \begin{align*} && {(b-\xi)^{n-1}\over{(n-1)!}}{f^{(n)}(\xi)} - {(b-\xi)^{n-1}\over{(n-1)!}}c =& 0 \\ \implies && {(b-\xi)^{n-1}\over{(n-1)!}}{f^{(n)}(\xi)} =& {(b-\xi)^{n-1}\over{(n-1)!}}c \\ \implies && {f^{(n)}(\xi)} =& c \end{align*}

Since c=f(n)(ξ)c={f^{(n)}(\xi)} is shown, the proof is concluded.


The proof is as shown above; however, a more commonly used form is as follows. Of course, as x[a,b]x \in [a,b] and set as x0(a,b)x_{0} \in (a,b), it essentially becomes [x0,x][a,b][x_{0} , x] \subset [a,b].

Taylor’s Theorem

If a function f(x)f(x) is continuous at [a,b][a,b] and differentiable up to nn times at (a,b)(a,b), then for x0(a,b)x_{0} \in (a,b),

f(x)=k=0n1(xx0)kk!f(k)(x0)+(xx0)nn!f(n)(ξ) f(x) = \sum_{k=0}^{n-1} {{( x - x_{0} )^{k}\over{ k! }}{f^{(k)}( x_{0} )}} + {(x - x_{0} )^{n}\over{ n! }}{f^{(n)}(\xi)}

there exists ξ(a,b)\xi \in (a,b) that satisfies it.

See Also


  1. Walter Rudin, Principles of Mathmatical Analysis (3rd Edition, 1976), p110-111 ↩︎