I found this notation in one paper that focuses on copulas:

Consider a $d$ -dimensional continuous random vector $X=\left\{X_{1}, X_{2}, \cdots X_{d}\right\}$ with marginals $F_{i}\left(x_{i}\right)=$ $\mathbb{P}\left(X_{i} \leq x_{i}\right) .$ Given a $x \in \mathbb{R}^{d},$ the distribution function $F(x)=\mathbb{P}\left(X_{1} \leq x_{1}, \cdots X_{d} \leq x_{d}\right)$ specifies all marginal distributions $F_{i}\left(x_{i}\right)$ as well as any dependencies between $X$.

I understood $X_i$ are random variables. I would kindly ask what is the $x_i$ in this case. It should also be random variable, but then there is just $x$ which is a random vector then.

When I think of marginal probabilities I imagine something like this:

  $X_1$ $X_2$ M
  10 20 30
  30 10 40
  20 10 30
Totals 60 40 100

So marginal probabilities are these form the M column. So isn't it better to define the marginal probabilities with the uppercase letter, just for consistency. Your sanity may help me understand the notations.


1 Answer 1


The vector $X=(X_1,\ldots,X_n)$ is a random vector made of random variables as components. For instance, $X=(X_1,X_2)$ with $X_1\sim\mathcal N(0,1)$ and $X_2|X_1\sim\mathcal N(-2X_1,2^2)$ is such a thing. The marginal distribution of $X_1$ is $\mathcal N(0,1)$ and the marginal distribution of $X_2$ is $\mathcal N(0,8)$

The joint cdf is defined as $$F(x)=F(x_1,x_2,\ldots,x_n)=\mathbb P(X_1\le x_1,\cdots,X_n\le x_n)$$ where $x=(x_1,\ldots,x_n)\in\mathbb R^n$. In the above example, $$F(x)=F(x_1,x_2)=\mathbb P(X_1\le x_1,X_2\le x_2)=\int_{-\infty}^{x_1} \varphi(x_1)\Phi([x_2+2x_1]/2)\,\text dx_1$$ It defines marginal distributions${}^*$ through their cdfs $$F_i(x_i)=F(+\infty,\ldots,\underbrace{+\infty,x_i,+\infty,}_\text{$i$-th position}\ldots,+\infty)$$

${}^*$I do not understand the table produced in the question, as the probabilities (?) do not add up to 100.

  • $\begingroup$ Are the $x_i$ scalars then? $\endgroup$
    – Good Luck
    Mar 12, 2021 at 14:33
  • $\begingroup$ I will gladly accept the answer if you touch the later casing part. $\endgroup$
    – Good Luck
    Mar 12, 2021 at 14:44
  • $\begingroup$ Am I wrong when I imagine $x_i$ is a sum of $i$-th state of random variables $X_i$? You see my upper table and just replace $M$ with $x$ then. $\endgroup$
    – Good Luck
    Mar 12, 2021 at 14:47
  • 1
    $\begingroup$ The table is not appropriate because the rows would correspond to the possible values of the rv $X_1$ while the columns would correspond to the possible values of the rv $X_2$. $\endgroup$
    – Xi'an
    Mar 12, 2021 at 15:09
  • 2
    $\begingroup$ Gender would be the random variable $X_1$ with possible values "female" and "male" while the number of ice-creams per day would be the random variable $X_2$. These two random variables have a joint distribution given by a contingency table. $\endgroup$
    – Xi'an
    Mar 12, 2021 at 15:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.