Expected value of sin of sum of n random angles

166 Views Asked by At

Consider the following problem.

Let $θ_1,θ_2,...θ_n ∈[0, \frac{π}{2}]$ be independent and uniformly distributed variables. Find $E[sin(θ_1 + ... + θ_n)].$

I was able to solve for $n=1$ (of course), but I'm not very sure how to go from there. I was thinking of using the linearity of expectation and doing induction on n (basically split up the $sin$ into two $sin*cos$ parts), but I'm not so sure if that works with deterministic functions of variables. Any help is welcome, thank you!

1

There are 1 best solutions below

0
On BEST ANSWER

This is made easy when one uses complex numbers (in particular the formula for the sine), which converts the summation to a product and makes the process of separating the independent parts easier.

Note that $\sin x = \frac{e^{ix} - e^{-ix}}{2i}$ for all $x$ real.

Therefore, we get $$ \sin(\theta_1+\ldots+\theta_n) = \frac{e^{i(\theta_1+\ldots+\theta_n)} + e^{-i(\theta_1+\ldots+\theta_n)}}{2i}= \frac{\prod_{i=1}^ne^{i\theta_i} - \prod_{i=1}^n e^{-i\theta_i}}{2i} $$

In particular, extending the definition of the expectation to complex-valued random variables using the obvious $\mathbb E[U+iV] = \mathbb E[U] + i\mathbb E[V]$ and noting that properties of the real-valued expectation carry over,

$$ \mathbb E[\sin(\theta_1+\ldots+ \theta_n)] =\frac 1{2i} \left(\mathbb E\left[\prod_{j=1}^n e^{i\theta_j}\right] - \mathbb E\left[\prod_{j=1}^n e^{-i\theta_j}\right] \right) $$

Now, if $\theta_j$ are independent, then so are the collections $\{e^{i \theta_j}\}$ and $\{e^{-i \theta_j}\}$ (this is easily seen using an argument analogous to the real number situation). In particular, we get $$ \mathbb E\left[\prod_{j=1}^n e^{i\theta_j}\right] = \prod_{i=1}^n \mathbb E[e^{i \theta_j}] = (\mathbb E[\sin \theta] + i \mathbb E[\cos \theta])^n $$ Likewise $$ \mathbb E\left[\prod_{j=1}^n e^{-i\theta_j}\right] = \prod_{i=1}^n \mathbb E[e^{-i \theta_j}] = (-\mathbb E[\sin \theta] + i \mathbb E[\cos \theta])^n $$

which upon substitution give the answer.


Alternately, if you wish to stay in the realm of real numbers for a longer time (an eventual formula will involve complex numbers in some form), define the sequences $a_n, b_n$ for $n \geq 1$ by $$ a_n = \mathbb E[\sin(\theta_1+\ldots+\theta_n)] \\ b_n = \mathbb E[\cos(\theta_1+\ldots+\theta_n)] $$

Then, the formulas $\sin(A+B) = \sin A \cos B + \cos A \sin B$ and $\cos(A+B) = \cos A \cos B + \sin A \sin B$, along with the natural split $$ \theta_1+\ldots+\theta_n = \underbrace{\theta_n}_{A} + \underbrace{\theta_1+\ldots+\theta_{n-1}}_{B} $$ and independence, lead instantly to the two recursion formulas$$ a_{n} = b_{n-1}a_1+b_1a_{n-1} \quad ; \quad b_n = b_{n-1}b_1 - a_{n-1}a_1 $$ That can be written in matrix form as $$ \begin{pmatrix} a_n \\ b_n \end{pmatrix} = \begin{pmatrix} b_1 & a_1 \\ -a_1 & b_1 \end{pmatrix} \begin{pmatrix} a_{n-1} \\b_{n-1}\end{pmatrix} $$ which has the solution $$\begin{pmatrix} a_n \\ b_n \end{pmatrix} = \begin{pmatrix} b_1 & a_1 \\ -a_1 & b_1 \end{pmatrix}^{n-1} \begin{pmatrix} a_{1} \\b_{1}\end{pmatrix} $$

$a_1,b_1$ can be obtained straightforwardly, while $a_n,b_n$ can be obtained by diagonalizing the above matrix to find an explicit formula for the $n-1$th power. This will involve complex numbers : for that formula, see here. Substitution provides an answer.