The Taylor series for $\sin(x)$, centered at any point, converges for all $x$.

The Taylor series for $e^{x}$ and $\cos(x)$ do as well.

Thus, taking an algebraic function of these (without division) forms another 'universally convergent' series:

$\frac{1}{2}(e^{x}-e^{-x}) = \sinh(x)$ converges for all $x$, for example.

And so does $(x^{2}+1)\sinh(x)$.

Note that 'universally convergent', which is a term I invented for the purpose of this question, does not mean that $\exists$ a finite region of convergence when expanding at any $x$ (analytic). It means that $\exists$ a center $x_0$ about which the series converges $\forall x$.

The question: I have not come across any 'universally convergent' functions besides these. Are there any other examples of closed form functions with a universally convergent Taylor series? I ask because such functions are extremely useful algebraically as they can immediately be substituted for their Taylor series in almost any situation.

  • 1
    $\begingroup$ $e^{e^x}$? $\frac{\sin x}x$? $\endgroup$ – bof Apr 26 '15 at 18:55

In complex analysis, one learns that if $f(z)$ is analytic at $z_0$ (differentiable in a neighborhood of $z_0$) then $f(z)$ can be represented by a Taylor series that converges to $f(z)$ in a circle with radius $r$. What's really nice is that $r$ is the exact distance between $z_0$ and the nearest singularity.

For example you have probably seen that $$\frac{1}{1+x^2} = \frac{1}{1-(-x^2)}= 1-x^2+x^4-\cdots = \sum (-1)^n x^{2n} $$ converges only for $|x|<1$ which is quite odd because this function is infinitely differentiable ($\forall n \forall x\in \mathbb{R}: \exists f^{(n)}(x)$) , but $f(z)=\frac{1}{1+z^2} $ converges in the circle $|z|<1$ in the complex plane, because the distance between $z_0=0$ and $z=i$, the nearest singularity, is $1$.

What if there isn't any singularity (e.g. $\sin z , \cos z , e^z $)? $r=\infty$ , and the Taylor series converges to $f(z)$ for all $z\in \mathbb{C}$.

So every "entire" function, i.e. a function that is (complex) differentiable at every point in $\mathbb{C}$, has a Taylor series that converges to it (for all $z\in \mathbb{C}$)

  • $\begingroup$ Excellent answer! Thank you! $\endgroup$ – doublefelix Apr 27 '15 at 20:34
  • $\begingroup$ My pleasure, feel free to ask more questions! $\endgroup$ – Ofek Gillon Apr 28 '15 at 21:19

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.