I was in the middle of answering this question that asked how to get the Taylor series of $cos(x)$ centered at $a=\frac{\pi}{3}$. I was reaching the point where I was about to write $(-1)^n$ making the it an alternating series, when I began to realize why calculus II courses only talk about the Maclaurin series of $cos(x)$. In the Maclaurin series when $a=0$, the function alternates signs every iteration using $(-1)^n$, and has the pattern of $\{-,+,-,+,-,+\}$. However, when the function is centered at $a=\frac{\pi}{3}$, the series has the pattern $\{+,-,-,+,+,-,-,+,+,...\}$
What form of $(-1)^{kn}$ can create a sequence with this kind of pattern?

  • $\begingroup$ I guess this behaviour is due to $\sin$ and $\cos$ terms rather than to $(-1)^n$, isn'it? $\endgroup$ – 7raiden7 May 11 '14 at 19:13

For sequences of period 4, one has to rely on the fourth roots of unity, that is, to linear combinations of the sequences of general term $1$, $\mathrm i^n$, $(-1)^n$ and $(-\mathrm i)^n$. In your case, starting at $n=1$, the sequence is $$ \frac{1-\mathrm i}2\cdot\mathrm i^n+\frac{1+\mathrm i}2\cdot(-\mathrm i)^n. $$ Another way to express the same sequence is to use cosine and sine functions with period 4, in your case, still starting at $n=1$, one would consider $$ \sqrt2\cdot\cos\left(n\frac\pi2-\frac\pi4\right). $$

  • $\begingroup$ If an expanded Taylor series is what we would use to estimate $cos(x)$, is it "legal" to use $cos(x)$ in its closed form? Wouldn't that be finding the value of $cos(x)$ by using a series that requires knowing the value of $cos(x)$? $\endgroup$ – Ephraim May 11 '14 at 19:27
  • $\begingroup$ There is no Taylor series involved here, whether of the cosine function or otherwise. The second formulation in my post only uses the values of $\cos(\pi/4)$, $\cos(3\pi/4)$, $\cos(5\pi/4)$ and $\cos(7\pi/4)$. $\endgroup$ – Did May 11 '14 at 19:33
  • $\begingroup$ The original reason for my question was to figure out how to express the Taylor series of the function $cos(x)$ centered at $a=\frac{\pi}{3}$. The first part of your answer looks like it would work no matter what. What I was asking in my comment was whether or not $\sqrt2\cdot\cos\left(n\frac\pi2-\frac\pi4\right)$ can be used to write a taylor series for the function $cos(x)$ $\endgroup$ – Ephraim May 11 '14 at 19:39

$$(-1)^{\lceil n/2\rceil}$$ where $\lceil x\rceil$ is the ceiling function, i.e. the smallest integer $\ge x$.


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.