From c8e67614bbf6945d457a808ecf9e8c4a7c8ce607 Mon Sep 17 00:00:00 2001 From: Erik Strand <erik.strand@cba.mit.edu> Date: Mon, 25 Feb 2019 16:16:01 -0500 Subject: [PATCH] Add first part of 4.6 --- _psets/3.md | 29 +++++++++++++++++++++++++++++ 1 file changed, 29 insertions(+) diff --git a/_psets/3.md b/_psets/3.md index 9e0472f..076e4f6 100644 --- a/_psets/3.md +++ b/_psets/3.md @@ -423,3 +423,32 @@ $$\num{1.5e8} \si{Hz}$$. Let $$(x_1, x_2, \ldots, x_n)$$ be drawn from a Gaussian distribution with variance $$\sigma^2$$ and unknown mean value $$x_0$$. Show that $$f(x_1, \ldots, x_n) = n^{-1} \sum_{i = 1}^n x_i$$ is an estimator for $$x_0$$ that is unbiased and achieves the Cramér–Rao lower bound. + +The probability of seeing a particular sequence $$x_1, \ldots, x_n$$ of samples is + +$$ +p(x_1, \ldots, x_n) += \prod_{i = 1}^n \frac{1}{\sqrt{2 \pi \sigma^2}} e^{-\frac{(x_i - x_0)^2}{2 \sigma^2}} \mathrm{d} x +$$ + +So the expected value of this estimator is + +$$ +\begin{align*} +\left \langle \frac{1}{n} \sum_{i = 1}^n x_i \right \rangle +&= \int_{-\infty}^\infty \cdots \int_{-\infty}^\infty + \left( \prod_{i = 1}^n \frac{1}{\sqrt{2 \pi \sigma^2}} e^{-\frac{(x_i - x_0)^2}{2 \sigma^2}} + \right) \left( \frac{1}{n} \sum_{i = 1}^n x_i \right) \mathrm{d} x_1 \ldots \mathrm{d} x_n \\ +&= \frac{1}{n} \sum_{i = 1}^n \prod_{j = 1}^n + \int_{-\infty}^\infty + \frac{1}{\sqrt{2 \pi \sigma^2}} e^{-\frac{(x_j - x_0)^2}{2 \sigma^2}} + \left( 1 - \delta_{ij} (1 + x_i) \right) + \mathrm{d} x_j \\ +&= \frac{1}{n} \sum_{i = 1}^n x_0 \\ +&= x_0 +\end{align*} +$$ + +The [Kronecker delta](https://en.wikipedia.org/wiki/Kronecker_delta) is used to indicate that only +one term in the product contains $$x_i$$. This term integrates to $$x_0$$ (since it is the expected +value of a single Gaussian), and the others integrate to one. -- GitLab