To solve a larger problem I must solve the sub-problem of finding a power series that satisfies: $$f^{(k)}(0) = (k!)^2.$$ My textbook states that this is "obviously" $$\sum^{\infty}_{k=0}k!z^k.$$ Why is this?
-
3Taylor's theorem? On the other hand, note that the radius of convergence of that series is zero... – Simply Beautiful Art Oct 06 '17 at 18:01
3 Answers
For any power-series $\sum_{k=0}^{\infty}a_k x^k$ one has that $f^{(k)}(0)= k! a_k$.
This is true in a 'formal' sense for any sequence of coefficients, and it is true in the analytic sense of derivative if the power-series has positive radius of convergence.
- 42,135
Disclaimer: This does not answer the posed question (which has already been answered) but it does explain how to convert your non-convergent power series into a function satisfying your differential equations.
Your series doesn't converge for any $z\ne0$. Instead, consider the Borel sum of your series:
$$\mathcal B\sum_{n=0}^\infty n!z^n=\int_0^\infty\exp(-t)\left(\sum_{n=0}^\infty z^nt^n\right)~\mathrm dt=\int_0^\infty\frac{\exp(-t)}{1-zt}~\mathrm dt=\frac1z\exp\left(-\frac1z\right)\operatorname{Ei}\left(\frac1z\right)$$
when $\Re(z)<0$ and where $\operatorname{Ei}(x)$ is the principle exponential integral.
$$\operatorname{Ei}(x)=P.V.\int_{-x}^\infty-\frac{e^{-t}}t~\mathrm dt$$
This function happens to satisfy
$$\lim_{z\to0}\frac{\mathrm d^n}{\mathrm dz^n}\frac1z\exp\left(-\frac1z\right)\operatorname{Ei}\left(\frac1z\right)=(n!)^2$$
As intended.
- 74,685