I want to sum up the partials of a harmonic series, how do I do it?

If I was using the 'Lagrange trigonometric identity to solve this problem', how would I plot it on Wolfram mathematica (using which input)?

  • $\begingroup$ A harmonic series ($\sum \frac1n$) or a sum of sines, $\sum \sin (ka)$? $\endgroup$ – Daniel Fischer Oct 29 '13 at 19:32
  • $\begingroup$ A sum of sines as described in the question: sina+sin2a+sin3a+...+(n-1)a $\endgroup$ – Alex Oct 29 '13 at 19:33
  • 2
    $\begingroup$ I asked because the question body speaks of a harmonic series, that is something different. $\endgroup$ – Daniel Fischer Oct 29 '13 at 19:34
  • $\begingroup$ Basically, I want to show that when a string on a piano is plucked, not only the fundamental frequency of the string plucked eg. 'a' for example resonates, but also multiples of that frequency, such as the octave above....and then I wanted to add these sin waves and create a function from there... $\endgroup$ – Alex Oct 29 '13 at 19:35
  • 2
    $\begingroup$ Harmonic in music $\neq$ harmonic in mathematics. Not entirely unrelated, but different. Do you already know Euler's formulae relating the exponential function and the trigonometric functions? $\endgroup$ – Daniel Fischer Oct 29 '13 at 19:38

Using De Moivre's formula, you find that


for every real number $a$.

Then, using the linearity of the imaginary part, your sum is clearly equal to the imaginary part of another much simpler sum :


If $a$ is an integer multiple of $2\pi$, then your sum is clearly equal to $0$.

If $a$ is not an integer multiple of $2\pi$, then $r=e^{ia}\neq 1$, and the terms of the sum $e^{ia}+\ldots+\left(e^{ia}\right)^{n-1}$ are just the terms of the geometric sequence


Now, you should remember that if $(u_n)_{n\geqslant 0}$ is a geometric sequence with ratio $r\neq 1$, then the sum of the $N$ first terms of this sequence is :


In our case, we have $u_0=r$ and $N=n-1$, so :


Now, we'll use this useful formula :

$$\forall \alpha\in\mathbb R,\ e^{i\alpha}-1=2ie^{i\frac\alpha 2}\sin\left(\frac\alpha 2\right)$$

(To prove it, just use the fact that $\sin(\theta)=\dfrac{e^{i\theta}-e^{-i\theta}}{2i}$ for all real number $\theta$ and develop the RHS).

Using this formula on both the numerator and denominator of our latest identity, we get :

$$e^{ia}+\ldots+\left(e^{ia}\right)^{n-1}=e^{ia}\frac{2ie^{i\frac{N-1}2a}\sin\left(\frac{N-1}2a\right)}{2ie^{i\frac a2}\sin\left(\frac a2\right)}=e^{i\frac N2a}\frac{\sin\left(\frac{N-1}2a\right)}{\sin\left(\frac a2\right)}$$

The imaginary part of the RHS is just the sum you wanted :

$$\sin(a)+\ldots+\sin((n-1)a)=\sin\left(\frac N2a\right)\frac{\sin\left(\frac{N-1}2a\right)}{\sin\left(\frac a2\right)}\quad (a\notin 2\pi\mathbb Z)$$


They are called Lagrange's trigonometric identities

$$\sum_{n=1}^N \sin (na)= \frac{1}{2}\cot\frac{a}{2}-\frac{\cos(N+\frac{1}{2})a}{2\sin\frac{a}{2}}$$

$$\sum_{n=1}^N \cos(na)= -\frac{1}{2}+\frac{\sin(N+\frac{1}{2})a}{2\sin\frac{a}{2}}$$

Source: http://en.wikipedia.org/wiki/List_of_trigonometric_identities

  • $\begingroup$ that looks very good, thank you! $\endgroup$ – Alex Oct 29 '13 at 19:50
  • 1
    $\begingroup$ Do you have a proof for that? $\endgroup$ – Alex Oct 29 '13 at 19:51

Hint: Multiple by $\sin\frac{a}{2}$ and use the equality $\sin x\sin y=\frac{1}{2}(\cos(x-y)-\cos(x+y))$:

$$ \sin\frac{a}{2}(\sin a+\sin 2a+\ldots)= \frac{1}{2}(\cos\frac{a}{2}-\cos\frac{3a}{2}+\cos\frac{3a}{2}-\cos\frac{5a}{2}+\ldots) $$ etc.

  • $\begingroup$ What do you mean by that? Could you show an example? $\endgroup$ – Alex Oct 29 '13 at 19:42
  • $\begingroup$ I corrected and added the answer. $\endgroup$ – Boris Novikov Oct 29 '13 at 19:53
  • $\begingroup$ is it true that all the -cos(3a/2)+cos(3a/2) cancel? $\endgroup$ – Alex Oct 30 '13 at 12:18
  • $\begingroup$ Yes, of course, except the first and the last summands. $\endgroup$ – Boris Novikov Oct 30 '13 at 13:33
  • $\begingroup$ That's a very nice trick! $\endgroup$ – Prism Nov 11 '13 at 19:18

$\sin a + \sin2a+\cdots+\sin((n-1)a)$ is the imaginary part of $e^{ia}+e^{2ia}+e^{3ia}+\cdots+e^{(n-1)ia}$. That is a finite geometric series whose sum is expressible in closed form. It will be $\dfrac{\text{something}}{1-e^{ia}}$. Multiplying the top and bottom both by $e^{-ia/2}$ will make it

$$ \frac{\text{something}}{e^{-ia/2}-e^{ia/2}} = 2i\cdot \frac{\text{something}}{\sin(a/2)}. $$

Once it's in that form you only have to look at the numerator to find the imaginary part.


Use the formula $$2\sin(x)\sin(y)= \cos(x-y)-\cos(x+y)$$


$$2\sin(\frac{a}{2})\sin(a)=\cos(\frac{a}{2})-\cos(\frac{3a}{2}) \\ 2\sin(\frac{a}{2})\sin(2a)=\cos(\frac{3a}{2})-\cos(\frac{5a}{2}) \\ 2\sin(\frac{a}{2})\sin(3a)=\cos(\frac{5a}{2})-\cos(\frac{7a}{2}) \\ ....\\ 2\sin(\frac{a}{2})\sin(na)=\cos(\frac{(2n-1)a}{2})-\cos(\frac{(2n+1)a}{2}) \\$$

Add them together.

  • $\begingroup$ add what together? $\endgroup$ – Alex Oct 29 '13 at 22:39
  • $\begingroup$ @Alex All these equalities. Note that on the RHS you get a telescopic sum, while on the LHS $2 \sin(\frac{a}{2})$ is a common factor. $\endgroup$ – N. S. Oct 29 '13 at 22:46
  • $\begingroup$ doesn't your last term already add all these? $\endgroup$ – Alex Oct 29 '13 at 22:48
  • $\begingroup$ so I would just have to divide the RHS by 2sin(a/2) and then I would have an expression for sin(na)? $\endgroup$ – Alex Oct 29 '13 at 22:48
  • $\begingroup$ @Alex no it doesn't. The last term is just the relation with $x=\frac{a}{2}$ and $y=na$. $\endgroup$ – N. S. Oct 29 '13 at 22:49

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.