There are many definitions online but they do not seem to align, or at least do not clearly correspond to, the usages I see elsewhere. For example: this website defines the difference is being $X$ refers to a set of population elements; and $x$, to a set of sample elements. $N$ refers to population size; and $n$ to sample size.

However in this post we see:

Now consider $X \sim U(0,1)$ and the sequence of random variables $X_n = \left(1 + \frac{1}{n}\right) X$. This is a sequence of RVs with $X_1 = 2X$, $X_2 = \frac{3}{2} X$, $X_3 = \frac{4}{3} X$ and so on. In what senses can we say this is getting closer to $X$ itself?

I do not understand what $X$ and $n$ mean here. I thought $X$ was a function that mapped from the sample space $\Omega$ to the reals? If that's the case what does $X_n$ mean here? Are there $n$ (the size of the sample space) possible functions that map from $\Omega$ to the reals?

  • 1
    $\begingroup$ To complement the clear answer of jld below, I would add that you are entirely correct when you say that you can think of a sequence of random variables $X_1, \dots, X_n$ as $n$ different functions (in your example) that map from $\Omega$ to the reals $\mathbb{R}$. However, when you write "Are there $n$ (the size of the sample space) possible functions..." be aware that the size of the sample space, as the number of possible outcomes in $\Omega$ is wholly distinct from the size of the sample, the number of observations of a random variable we collect. $\endgroup$
    – microhaus
    Jun 16 '21 at 20:14

In this context, capitalisation is usually used to distinguish random variables from fixed values, so you are correct to interpret $X: \Omega \rightarrow \mathbb{R}$ as a random variable. The post specifies that it is considering a sequence of random variables $X_n : \Omega \rightarrow \mathbb{R}$, so there are an infinite number of these random variables, of the form:

$$\begin{align} X_1(\omega) &= (1+\tfrac{1}{1}) X(\omega), \\[6pt] X_2(\omega) &= (1+\tfrac{1}{2}) X(\omega), \\[6pt] X_3(\omega) &= (1+\tfrac{1}{3}) X(\omega), \\[6pt] &\ \ \vdots \\[6pt] X_n(\omega) &= (1+\tfrac{1}{n}) X(\omega), \\[6pt] &\ \ \vdots \\[6pt] \end{align}$$

Finally, the value $n$ here has nothing to do with the size of $\Omega$ or any other substantive property of the sample space. It is merely an index used to specify the form of the random variables in the sequence --- the post could just as easily have used the index $i$, or any other variable notation here.


You are correct that $X$ is a Borel function $X : \Omega\to\mathbb R$. A realization of $X$ is a particular real number $X(\omega)$, and such a number may be referred to as $x$. This means, for example, the event $\{X = x\}$ is more formally written as $\{\omega\in\Omega : X(\omega) = x\}$.

We can define as many functions as we want from $\Omega$ to $\mathbb R$, so e.g. we can imagine having $X_1, X_2, \dots : \Omega\to\mathbb R$. If $X_i = a_i X$ for constants $a_i\in\mathbb R$ then each function $X_i$ is also Borel so these are all random variables too. $n$ is often used to index a sequence of random variables so I could refer to $X_n$ as an arbitrary element of the sequence of functions $X_1, X_2, \dots$ instead of using $X_i$ or some other indexing variable. This use of $n$ is totally different from that of a sample size, although often the sequence of RVs in question comes from imagining taking bigger and bigger samples so $X_n$ may correspond to a sample of $n$ things, but it doesn't have to.

We can then ask about how the sequence of RVs evolves by letting $n \to \infty$. For example, in your case since $a_n \to 1$ as $n\to\infty$ we'll have $X_n$ converging on $X$ in some sense.


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.