January 11th, 2017, 04:58 PM  #1 
Member Joined: Dec 2016 From:  Posts: 44 Thanks: 7  Cinf smooth function
Hi all, I was reading through Roger Penroe's "the road to reality" and I found the following function \begin{eqnarray} h(x)&=&e^{1/x} & x>0\\ h(x)&=&0 & \text{elsewhere} \end{eqnarray} where it says this functions is $C^{\infty}$ smooth, meaning all derivatives to all orders are smooth (the singular point here is $x=0$). It says to give it a try but doesnt show any proof, according to the book the proof might be a bit difficult. I have tried unsuccesfully to tackle it, any ideas to solve this? Thanks! 
January 11th, 2017, 07:58 PM  #2 
Senior Member Joined: Mar 2015 From: New Jersey Posts: 1,053 Thanks: 85 
h(x) is infinitely differentiable because no matter how many times you differentiate you get a sum of differentiable products which is differentiable.

January 11th, 2017, 09:33 PM  #3  
Senior Member Joined: Sep 2016 From: USA Posts: 114 Thanks: 44 Math Focus: Dynamical systems, analytic function theory, numerics  Quote:
1. Convince yourself that it is continuous at $x=0$. 2. The derivative exists at $x=0$ if the following limit exists: $$ \lim_{h\rightarrow 0} \frac{e^{\frac{1}{h}}}{h} = \lim_{u\rightarrow \infty} ue^{u}.$$ Convince yourself that $h'(0) = 0$. 3. Now, you compute $h'(x)$ piecewise and you argue that $h \in C^1$. With this established, it is immediate that $h \in C^{\infty}$.  
January 12th, 2017, 07:28 AM  #4 
Member Joined: Dec 2016 From:  Posts: 44 Thanks: 7 
Ok, I get the first point, that is easy to demonstrate, but I dont get the second. Is that the condition of continuity of the derivative or what? (Actually the derivative should be divided by $x^{2}$ and not by $x$, why are you using $h$ as variable here?) I dont also see the 3rd point, since it is a proof that the first derivative exists, but not to the derivative to all orders, moreover, is not only that they exist, but that they are all smooth functions (no peaks or discontinuities). Last edited by greg1313; January 12th, 2017 at 11:09 AM. 
January 14th, 2017, 11:31 AM  #5 
Senior Member Joined: Mar 2015 From: New Jersey Posts: 1,053 Thanks: 85 
SDK is right, but you have to show h$\displaystyle ^{(n)}$(0) exists for all n to show C$\displaystyle ^{\infty}$ for x $\displaystyle \geq$ 0. You can do so because h$\displaystyle ^{(n)}$(0) will consist of a sum of terms of the form: $\displaystyle \lim_{\delta\rightarrow 0}\frac{e^{\frac{1}{\delta}}}{\delta^{n}}=\lim_{X\rightarrow \infty}\frac{X^{n}}{e^{X}}=0$ by L'hospitals rule. And the limit of a sum is the sum of the limits. 

Tags 
cinf, function, smooth 
Thread Tools  
Display Modes  

Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
nonpiecewise smooth function  mariam  Applied Math  1  December 27th, 2013 06:56 AM 
smooth function solution check  rayman  Real Analysis  0  February 18th, 2013 04:41 AM 
Smooth function on a smooth manifold  martexel  Real Analysis  1  November 26th, 2010 12:49 PM 
smooth function and its derivatives  regfor3  Real Analysis  3  November 10th, 2008 07:53 AM 
yet another smooth function problem  NoSoup4u  Real Analysis  2  November 6th, 2008 07:11 PM 