Derivative of Sum not Equal to Sum of Derivatives

jinydu · 2008-11-22, 07:57

One of my students complained this week after I took off points for assuming that a power series can be differentiated term-by-term with no further justification. He asked me for an example why this isn't ok.

I know of one example. The Taylor series for log(1+x) about x = 0 converges when |x| <= 1 except when x = -1. Evidently, this series is differentiable at x = 1 with derivative 1/(1+1) = 1/2. However, differentiating the series term-by-term and evaluating at x = 1 gives a series that does not converge.

I'm looking for further examples. More specifically:

1) Is there an example that only involves positive terms?

2) Is there an example where differentiating the series term-by-term and evaluating at a specific point gives a convergent series that converges to the wrong value (i.e. the derivative of the infinite fails to exist at the point or is has a different value)?

Of course, I know these things can only happen on the boundary of the disk of convergence

Thanks

Kevin · 2008-11-23, 13:39

The basic idea is when you differentiate, you're multiplying terms in the power series by n. So you want things that converge, but slowly enough so that they won't if terms are multiplied by n.

For the log(1+x) example, the original power series is sum of (-1)^n*x^n/n, and when you differentiate it you get sum of (-1)^n*x^n. So it goes from being alternating series with terms going to 0 linearly to something that just alternates. If you want an example of something using just positive terms, I'd figure out what originally has a power series where the value at x=1 correspond to zeta(2), and becomes the harmonic series after you differentiate (or just say consider the power series sum x^n/n^2 at x=1).

I'm pretty sure that if things converge, it's going to converge to the right thing, but I don't feel like thinking about which theorem of analysis would justify it (EDIT: try abel's theorem http://en.wikipedia.org/wiki/Abel%27s_theorem)

tmorrow · 2008-11-24, 10:22

Kevin showed the way, here's an example I made up.

f(x) = sum{1,inf} x^n/[n(n+1)] ...(1)...

Using log(1-x) = - sum{1,inf} on |x|<r=1 and the fact that 1/n(n+1) = 1/n - 1/(n+1), the above power series can be expressed in terms of familiar functions:

f(x) = (1/x)(1-x)log(1-x) + 1 on |x|<1, x!=0 (x=0 works in the limit x->0) ...(2)...

Formal differentiation of the power series (1) gives:

f'(x) = sum{1,inf} x^(n-1)/(n+1) ...(3)...
which clearly diverges at x=1 by comparison with the harmonic series.

Routine differentiation of our elementary form (2) gives (after a tidy up)

f'(x) = -(1/x)[1+(1/x).log(1-x)] ...(4)...

From (4) it is clear that we are seeing the same difficulty with non differentiability at x=1. The equality between (3) and (4) can be easily established as well.

Using the above approach students can relate the problems of formal power series differentiation with that of routine differentiation of elementary functions and illustrates a power series can represent poorly behaved functions (a_n may look well behaved on the surface but looks are deceiving). I'm sure simpler function examples can be found if you look hard enough.

2008-11-22, 07:57	#1
jinydu Dec 2003 Hopefully Near M48 2·3·293 Posts	Derivative of Sum not Equal to Sum of Derivatives One of my students complained this week after I took off points for assuming that a power series can be differentiated term-by-term with no further justification. He asked me for an example why this isn't ok. I know of one example. The Taylor series for log(1+x) about x = 0 converges when \|x\| <= 1 except when x = -1. Evidently, this series is differentiable at x = 1 with derivative 1/(1+1) = 1/2. However, differentiating the series term-by-term and evaluating at x = 1 gives a series that does not converge. I'm looking for further examples. More specifically: 1) Is there an example that only involves positive terms? 2) Is there an example where differentiating the series term-by-term and evaluating at a specific point gives a convergent series that converges to the wrong value (i.e. the derivative of the infinite fails to exist at the point or is has a different value)? Of course, I know these things can only happen on the boundary of the disk of convergence Thanks

2008-11-23, 13:39	#2
Kevin Aug 2002 Ann Arbor, MI 110110001₂ Posts	The basic idea is when you differentiate, you're multiplying terms in the power series by n. So you want things that converge, but slowly enough so that they won't if terms are multiplied by n. For the log(1+x) example, the original power series is sum of (-1)^nx^n/n, and when you differentiate it you get sum of (-1)^nx^n. So it goes from being alternating series with terms going to 0 linearly to something that just alternates. If you want an example of something using just positive terms, I'd figure out what originally has a power series where the value at x=1 correspond to zeta(2), and becomes the harmonic series after you differentiate (or just say consider the power series sum x^n/n^2 at x=1). I'm pretty sure that if things converge, it's going to converge to the right thing, but I don't feel like thinking about which theorem of analysis would justify it (EDIT: try abel's theorem http://en.wikipedia.org/wiki/Abel%27s_theorem) Last fiddled with by Kevin on 2008-11-23 at 13:42

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
square root equal to a negative	LaurV	Homework Help	34	2013-03-03 09:35
Derivative of Gamma	nibble4bits	Math	0	2008-01-08 05:53
Laptop - why does colder equal slower??	kdelisle2005	Hardware	11	2007-03-02 03:54
P-1 not "equal" to x curves ECM with same bounds?	Mystwalker	Factoring	1	2004-10-02 17:44
Special exponents with equal digits in series	hbock	Lone Mersenne Hunters	0	2004-01-03 20:51

2008-11-24, 10:22	#3
tmorrow Jan 2004 103 Posts	Kevin showed the way, here's an example I made up. f(x) = sum{1,inf} x^n/[n(n+1)] ...(1)... Using log(1-x) = - sum{1,inf} on \|x\|<r=1 and the fact that 1/n(n+1) = 1/n - 1/(n+1), the above power series can be expressed in terms of familiar functions: f(x) = (1/x)(1-x)log(1-x) + 1 on \|x\|<1, x!=0 (x=0 works in the limit x->0) ...(2)... Formal differentiation of the power series (1) gives: f'(x) = sum{1,inf} x^(n-1)/(n+1) ...(3)... which clearly diverges at x=1 by comparison with the harmonic series. Routine differentiation of our elementary form (2) gives (after a tidy up) f'(x) = -(1/x)[1+(1/x).log(1-x)] ...(4)... From (4) it is clear that we are seeing the same difficulty with non differentiability at x=1. The equality between (3) and (4) can be easily established as well. Using the above approach students can relate the problems of formal power series differentiation with that of routine differentiation of elementary functions and illustrates a power series can represent poorly behaved functions (a_n may look well behaved on the surface but looks are deceiving). I'm sure simpler function examples can be found if you look hard enough.