I'm at Kreyszig - "Advanced Engineering Mathematics" 10th ed. - sec. "15. Power Series, Taylor Series" - example 6. It finds the Maclaurin series of $f(z)=\arctan(z)$ by integrating
\begin{align*} f'(z)=\frac{1}{1+z^2}=\sum\limits_{n=0}^\infty (-1)^n z^{2n} && |z|<1 \tag{1} \end{align*}
term by term and, quote, 'using $f(0)=0$'.
Question: Why we can say that $\sum\limits_{n=0}^\infty \frac{(-1)^n}{2n+1} z^{2n+1}$ converges to the integral of $\frac{1}{1+z^2}$ (in $|z|<1$), given that it never mentioned that I can "switch the integral sign and the summation sign" to get the sum?
Background: up to now it proved that (I try to be concise so excuse me for highly informal math below):
- Continuity: if $f=\sum a_n (z-z_0)^n$, then $f$ is continuos at $z_0$
- Uniqueness: if $f$ has a power series, it's unique
- Termwise differentiation: the series of derived terms has the same radius of convergence of the original series.
- Termwise integration: the series of integrated terms has the same radius of convergence of the original series.
- if $f$ has a power series, then $f$ is analytic and the series of derived terms is equal to $f'$
- if $f$ is analytic, then it has a power series (the Taylor's series).
while chap. 14 was on Complex integration
Thanks, Luca
One of the main results about power series is that they converge UNIFORMLY on every compact subset of the disk of convergence (if you do not know what a compact subset is, just substitute "on any concentric disk of smaller radius").
Any uniformly convergent series can be integrated term-by term.
This is from calculus, but it is very easy to prove, once you understand the definition of uniform convergence. The only thing about integral used here is that $|\int_a^b f(z)dz|\leq \epsilon(b-a)$ whenever $|f(z)|\leq \epsilon$ for $z\in[a,b]$.