I need to prove that: \begin{align} c_k =&\; \cos(k\!\cdot\!x)\\ c_k :=&\; c_{k-1} +d_{k-1}\\ d_k :=&\; 2d_0\!\cdot\!c_k +d_{k−1}\\ d_0 :=&\; −2\!\cdot\!\sin^2{(x/2)}\\ \end{align}

I've got an explicit formula for $d_k$ which should be:

\begin{align} d_k&=d_o+\sum_{i=1}^k{2\!\cdot\!d_o\!\cdot\!c_i} &&\implies& c_k &=c_{k-1}+ \sum_{i=1}^k{2\!\cdot\!d_o\!\cdot\!c_i} \end{align}

Now I want to do a proof by induction. Assuming that $c_p=\cos(p\!\cdot\!x)$ for every $p<k$. This would get me the following: $$c_k =\cos\left(\left(k-1\right)\!\cdot\!x\right)+\sum_{i=1}^k{2\!\cdot\!d_o\!\cdot\!\cos(p\!\cdot\!x)}$$ Using this formula I found:

\begin{align} \sum_{k=1}^n \cos(kx) & = \frac{\sin\left(\frac{nx}2\right)}{\sin \left(\frac{x}2\right)}\, \cos\left(\frac{(n+1)\,x}2\right) \end{align}

I tried to play around with trigonometric addition formulas but I am getting nowhere.

  • 1
    $\begingroup$ Your $c_k$ coincide with $T_k(\cos x)$ where $T_k$ is a Chebyshev polynomial of the first kind. $\endgroup$
    – Piquito
    Feb 4 '16 at 0:08
  • 2
    $\begingroup$ So the first line has to be proven, using lines 2-3-4? But then what is $c_0$? $\endgroup$ Feb 4 '16 at 13:08
  • $\begingroup$ @Justpassingby If $c_k=\cos(k*x)$, then $c_0=1$. $\endgroup$ Feb 4 '16 at 21:49
  • 1
    $\begingroup$ @simpleArt if we are required to prove the relation that $c_k=\cos(k*x)$ then how can we use this relation and say $c_0=1$ and then use this ...to again prove the same thing that is $c_k=\cos(k*x)$ ...it is logically incorrect ...to use the information that we are required to prove .... $\endgroup$
    – Freelancer
    Feb 5 '16 at 3:41
  • 1
    $\begingroup$ Very confusing. $c_k = \cos(kx)$ is the definition of $c_k$? Or you want to prove that is a consequence of the other conditions (with $c_0 = 1$)? $\endgroup$ Feb 5 '16 at 11:00

The last equation reads $$d_0=-2\sin^2\left(\frac x2\right)=\cos(x)-1.$$

Then eliminating $d$ from the second and the third,

$$d_k=c_{k+1}-c_k,\\ c_{k+1}-2c_k+c_{k-1}=2(\cos(x)-1)c_k.$$

If we assume that $c_k=\cos(kx)$, then after some simplification,


Hence $c_k=\cos(kx)$ is indeed a solution, such that $c_0=1$ (not specified in the OP).

We also have


Now knowing these relations, formally proving by induction isn't a problem anymore.

By the induction hypothesis, $$c_{k-1}=\cos((k-1)x),\\d_{k-1}=\cos(kx)-\cos((k-1)x).$$


$$\color{green}{c_k}=c_{k-1}+d_{k-1}=\cos((k-1)x)+\cos(kx)-\cos((k-1)x)\color{green}{=\cos(kx)}$$ and $$\color{green}{d_k}=2(\cos(x)-1)c_k+d_{k-1}\\ =2(\cos(x)-1)\cos(kx)+\cos(kx)-\cos((k-1)x)\\ =2\cos(x)\cos(kx)-2\cos(kx)+\cos(kx)-\cos(kx)\cos(x)-\sin(kx)\sin(x)\\ =\cos(kx)\cos(x)-\sin(kx)\sin(x)-\cos(kx)\\ \color{green}{=\cos((k+1)x)-\cos(kx)}.$$

And the base case is

$$\color{green}{c_0}=\cos(0\cdot x)=\color{green}{1},\\ \color{green}{d_0}=\cos(1\cdot x)-\cos(0\cdot x)=\color{green}{\cos(x)-1}.$$

  • 1
    $\begingroup$ Probably easier using $\cos((k+1)x)+\cos((k-1)x)=2\cos(kx)\cos(x)$. $\endgroup$
    – user65203
    Feb 6 '16 at 15:35

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.