3
$\begingroup$

Looking at the series expansion of $\tan x$, it appears as the monotonically increasing function which we know isn't true. So, how can we describe it's periodicity from it's Maclaurin series?

  • 3
    The McLauin series of the tangent has a restricted interval of convergence. So it doesn't really make sense to talk about periodicity here...See this link:https://proofwiki.org/wiki/Taylor_Series_Expansion_for_Tangent_Function.2017-02-03
  • 0
    For a simpler example of the same phenomenon, take the Maclaurin series for $1/(1-x)$. Looks monotonically increasing too, but for $x>1$ the function is negative. It's not safe to draw conclusions about a function by looking at its Taylor series outside its interval of convergence.2017-02-03

2 Answers 2

3

The Taylor series of a smooth function at a point need not converge everywhere; even if it does, it may not equal the function everywhere! In this case, the Taylor series at $0$ converges and agrees with tangent on the connected component of the graph containing the origin; but this isn't the whole graph! So the rest of the tangent function is invisible to the Taylor series at zero.

  • 0
    Thanks for the link. Never knew there existed such a thing.2017-02-03
0

I believe that it's because the function is discontinuous (and undifferentiable) at $x= \pm \frac{ \pi}{2}$. Since Taylor series rely only on continuity and looks at the derivatives at a single point, in theory it should rapidly diverge to infinity after $x= \pm \frac{ \pi}{2}$ (not sure if it actually does).

  • 1
    MIght want to change to $\pm \frac{\pi}{2}$2017-02-03