Any help with the proof I have posted for the following question is greatly appreciated; I have listed my particular issue at the end. Thank you!

Let $p \geq 0$. For $n=1,2, \dots,$ find $P_n(x)$ about $x=0$ of $f(x)=(1+x)^p$ and find an upper bound depending on $n,p$ for the error on $x \in [0,1]$.

Since the solution depends heavily on Taylor's theorem, I will state it here:

Suppose $f \in C[a,b]$, $f^{n+1} \in C[a,b]$ and $x_0 \in [a,b]$. For every $x \in [a,b]$, there exists a number $\xi(x)$ between $x_0$ and $x$ such that $$f(x)=P_n(x)+R_n(x)$$ where $$P_n(x)=\sum_{k=0}^n \frac{f^{(k)}(x_0)}{k!}(x-x_0)^k$$ $$R_n(x)=\frac{f^{(n+1)}(\xi(x))}{(n+1)!}(x-x_0)^{n+1}$$

As $f(x)$ describes the binomial distribution, we have $$P_n(x)=\sum_{k=0}^n\frac{p(p-1)\cdots (p-k+1)}{n!} x^k$$ We will now find an upper bound on $n,p$ for the error on $x \in [0,1]$. Using $f^{(n+1)}$ obtained earlier, we have: $$R_n(x)=\frac{p(p-1)\cdots (p-n)(1+\xi(x))^{p-n-1}}{(n+1)!}x^{n+1}$$ \begin{eqnarray*} \max_{x \in [0,1]}|R_n(x)|& = & \max_{x \in [0,1]} \left|\frac{p(p-1)\cdots (p-n)(1+\xi(x))^{p-n-1}}{(n+1)!}x^{n+1}\right| \\ & = & \max_{x \in [0,1], \xi \in [0,x]} \left|\frac{p(p-1)\cdots (p-n)(1+\xi)^{p-n-1}}{(n+1)!}x^{n+1}\right| \\ & = & \left|\frac{p(p-1)\cdots (p-n)}{(n+1)!}\right| \max_{x \in [0,1], \xi \in [0,x]} \left|(1+\xi)^{p-n-1} x^{n+1}\right| \\ & \leq & \left|\frac{p(p-1)\cdots (p-n)}{(n+1)!}\right| \max_{x \in [0,1], \xi \in [0,1]} \left|(1+\xi)^{p-n-1} x^{n+1}\right| \\ & \leq & \left|\frac{p(p-1)\cdots (p-n)}{(n+1)!}\right| \max_{\xi \in [0,1]} \left|(1+\xi)^{p-n-1} \right| \\ & \leq & \left|\frac{p(p-1)\cdots (p-n)}{(n+1)!}\right| \left|2^{p-n-1}\right| \end{eqnarray*}

Now, I have a major problem with the transition from the second last step to the last step. Specifically, is the maximum of $\left|2^{p-n-1}\right|$ always achieved when $\xi=1$? It seems to me that for example, if $p=1, n=1$, then $p-n-1<0$, and the previous condition fails. Any advice would be helpful here. Should I take it on a case-by-case basis? Is there anything else wrong with the proof? Thanks for reading my large post, I hope that I have provided enough context.

  • $\begingroup$ The statement of Taylor's theorem you wrote says that $f^{(n+1)}$ is continuous on $[a,b]$. So why not just invoke the extreme value theorem and say there exists a point for which $|f^{(n+1)}(x)|$ attains a maximum value? Of course you'd need to verify that $f^{(n+1)}$ is indeed continuous. $\endgroup$ – Yeldarbskich Sep 29 '15 at 18:59
  • $\begingroup$ How would I be able to use this to find the explicit maximum value (or bound), though? $\endgroup$ – Douglas Fir Sep 29 '15 at 19:08
  • 1
    $\begingroup$ I wasn't sure you needed an explicit bound. The bound would certainly be dependent on if $p<n+1$ or if $p \geq n+1$, so maybe you can work out what it should be in those cases. $\endgroup$ – Yeldarbskich Sep 29 '15 at 19:14

I think you will have to do a case based analysis. If p is an integer, then for $$n \geq p\ , R_n =0 .$$ and $$n \leq p\ , \xi =1 .$$ gives maximum.

If not, then $ p-n\geq 1$ gives maximum for $\xi =1 $.

Or else,if $ p-n\leq 1$ gives maximum for $\xi =0$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.