# Are products of independent random variables independent?

Let $Z_0, Z_1, Z_2,...$ be independent and identically distributed such that

$P(Z_n = 1) = P(Z_n = -1) = 1/2$ for $n = 0, 1, 2, ...$

Let $X_0 = Z_0$, $X_1 = X_0 Z_1$, $X_2 = X_1 Z_2$, ...

Are $X_0, X_1, X_2, ...$ independent?

What I tried:

We must prove that for Borel sets $B_1, ..., B_n$,

$$P(X_0 \in B_0, X_1 \in B_1, ..., X_n \in B_n) = \prod_{i=0}^{n} P(X_i \in B_i) \ (*)$$

since $X_0, X_1, X_2, ..., X_n$ are independent $\forall n \in \mathbb{N}$ iff $X_0, X_1, X_2, ...$ are independent.

1. $\{ X_n \}_{n=0}^{\infty}$ is Markov i.e.

$$P[X_n \in B| X_m] = P[X_n \in B| \mathscr{F}_m]$$

$\forall m \in [0,n], B \in \mathscr B$

This implies that LHS of (*) is equivalent to:

$$P(X_0 \in B_0) P(X_1 \in B_1 | X_0 \in B_0) ... P(X_n \in B_n | X_{n-1} \in B_{n-1})$$

1. $P(X_{n+1} = 1) = P(X_{n+1} = -1) = 1/2$ can be proven by induction by noting the recurrence relations:

$$P(X_{n+1} = 1) = P(X_n = 1)P(Z_{n+1} = 1) + P(X_n = -1)P(Z_{n+1} = -1)$$

$$P(X_{n+1} = -1) = P(X_n = 1)P(Z_{n+1} = -1) + P(X_n = -1)P(Z_{n+1} = 1)$$.

I made use of the fact that $X_n$ and $Z_{n+1}$ are independent, which follows from $X_n = Z_0Z_1 \dots Z_n$ and $Z_1, ..., Z_n$ and $Z_{n+1}$ are independent.

This makes the RHS of (*) to be $(1/2)^{n+1}$

1. $P(X_i \in B_i | X_{i-1} \in B_{i-1}) = 1/2$ because for $a_{n+1} \in \{-1, +1\}$

$P(X_{n+1} = a_{n+1} | X_n) = E[1_{X_{n+1} = a_{n+1}} | X_n] = E[1_{X_{n+1} = a_{n+1}} | X_n = 1]P(X_n = 1) + E[1_{X_{n+1} = a_{n+1}} | X_n = -1]P(X_n = - 1)$

This makes the LHS of (*) to be $(1/2)^{n+1}$ as well. QED

Any mistakes or missing steps?

• Functions of independent variables are independent. So if $(X, Y)$ are independent from $(Z,T)$ then $f(X,Y)$ is independent from $g(Z,T)$. – kjetil b halvorsen Sep 7 '15 at 11:30
• Your new proof of the Markov property is wrong. In the notation of the linked alternative formulation, you need to consider more $f$s than just the identity.. – ekvall Nov 6 '15 at 1:52
• @Student001 Thanks! Serves me right for trying to replace E's w/ P's. Anyway, edited OP and answer below – BCLC Nov 6 '15 at 8:45
• Sure thing. It's still wrong though; you are saying all those probabilities in the Markov statement are equal to zero. – ekvall Nov 6 '15 at 14:46

You are making this problem a lot harder than it needs to be because the random variables in question are two-valued, and the problem can be treated as one of independence of events rather than independence of random variables. In what follows, I will treat the independence of events even though the events will be stated in terms of random variables.

Let $Z_0,Z_1,Z_2,\cdots$ be independent random variables $\ldots$

I will take this as the assertion that the countably infinite collection of events $A_i = \{Z_i = +1\}$ is a collection of independent events. Now, a countable collection of events is said to be a collection of independent events if each finite subset (of cardinality $2$ or more) is a collection of independent events. Recall that $n\geq 2$ events $B_0, B_1, \cdots, B_{n-1}$ are said to be independent events if $$P(B_0\cap B_1\cap \cdots \cap B_{n-1}) = P(B_0)P(B_1) \cdots P(B_{n-1})$$ and every finite subset of two or more of these events is a collection of independent events. Alternatively, $B_0, B_1, \cdots, B_{n-1}$ are said to be independent events if the following $2^n$ equations hold: $$P(B_0^*\cap B_1^*\cap \cdots \cap B_{n-1}^*) = P(B_0^*)P(B_1^*)\cdots P(B_{n-1}^*)\tag{1}$$ Note that in $(1)$, $B_i^*$ stands for $B_i$ or $B_i^c$ (same on both sides of $(1)$) and the $2^n$ choices ($B_i$ or $B_i^c$) give us the $2^n$ equations.

For our application, $A_i = \{Z_i = +1\}$ and $A_i^c = \{Z_i=-1\}$, and so checking whether the $2^n$ equations $$P(A_0^*\cap A_1^*\cap \cdots \cap A_{n-1}^*) = P(A_0^*)P(A_1^*)\cdots P(A_{n-1}^*)\tag{2}$$ hold or not, is equivalent to checking that the joint probability mass function (pmf) of $Z_0, Z_1, \cdots, Z_{n-1}$ factors into the product of the $n$ marginal pmfs at each and every one of the points $(\pm 1, \pm 1, \cdots, \pm 1)$ which is what you would be doing if you had never heard of independent events, just about independent random variables.

Thus, the statement

Let $Z_0,Z_1,Z_2,\cdots$ be independent random variables $\ldots$

does mean, among other things, that $Z_0,Z_1,Z_2,\cdots, Z_{n-1}$ is a finite collection of independent random variables. But, does the assertion

For all $n \geq 2$, $\{Z_0,Z_1,Z_2,\cdots, Z_{n-1}\}$ is a set of $n$ independent random variables

imply that the countably infinite set $\{Z_0,Z_1,Z_2,\cdots \}$ is a collection of independent random variables?

The answer is Yes, because we know by hypothesis that some specific finite subsets of $\{Z_0,Z_1,Z_2,\cdots \}$ are independent random variables, while any other finite subset, say $\{Z_2, Z_5, Z_{313}\}$, is a subset of $\{Z_0, Z_1, \cdots, Z_{313}\}$ which are independent per the hypothesis and so the subset is also a set of independent random variables.

In your question, with each $a_i \in \{+1, -1\}$ and defining $b_i = \prod_{j=0}^i a_j$ which is also in $\{+1,-1\}$, \begin{align} P(X_0 = a_0, X_1 = a_1, \cdots, X_n = a_n) &= P(Z_0 = a_0, Z_1 = a_0a_1, Z_2 = a_0a_1a_2, \cdots, Z_n = a_0a_1...a_n)\\ &= P(Z_0=b_0, Z_1 = b_1, \cdots, Z_n = b_n)\\ &= \prod_{i=0}^n P(Z_i = b_i)\\ &= 2^{-(n+1)}\\ &= \prod_{i=0}^n P(X_i = a_i), \end{align} that is, all $2^{n+1}$ equations of the form $(2)$ hold. Thus, for each $n \geq 1$, $X_0, X_1, \cdots, X_n$ are independent random variables, and therefore the countably infinite collection $\{X_0, X_1, \cdots\}$ of random variables is a collection of independent random variables.

After reading over my revised answer, perhaps it is I who is making the problem much harder than necessary. My apologies.

• Since the question asks whether an infinite set of random variables is independent, it would seem that the sigma algebras involved must be infinite and cannot be finite. Or do you perhaps have a theorem available that allows you to draw conclusions about independence of infinite collections of random variables based on the independence of finite subsets of them? – whuber Sep 7 '15 at 16:22
• An infinite collection of random variables is said to be a set of independent random variables if every finite subset is a set of independent random variables. At least, that's the definition that I was taught at first; the stuff about tail $\sigma$-algebras came later in a course on measure-theoretic probability. I think an argument can be made that if for each $n \geq 2$, $X_0, X_1, \cdots , X_n$ are independent random variables, then the infinite collection $\{X_0, X_1, X_2, \cdots\}$ is a collection of independent random variables. – Dilip Sarwate Sep 7 '15 at 16:41
• Thank you! That looks like such helpful information that you ought to consider incorporating it in your answer. – whuber Sep 7 '15 at 16:43
• @whuber I have incorporated the argument into my answer. As you are aware, the standard shibboleth $$P\left(\bigcap_{i=1}^\infty A_i\right) = \prod_{i=1}^\infty P(A_i)\tag{1}$$ cannot be used to say that the $A_i$ are a countable collection of independent events/random variables even if we interpret both sides of $(1)$ as the limit as the number of variables increases without bound. The limit is $0$ in either case. – Dilip Sarwate Sep 7 '15 at 21:56
1. ... How do I state this precisely, if it is right? $\forall i \leq n, \sigma(X_i) \subseteq \sigma(X_n)$ ?

Your have the right idea, but I would recommend using the definition of the Markov property to state this, namely that we have $P(X_n\mid X_0,\dots,X_{n-1})=P(X_n \mid X_{n-1})$. There is nothing imprecise about this as long as you have a precise definition of conditional probabilities. The $\sigma-$algebra condition you wrote is not correct.

1. ...It seems like I assumed $X_n$ and $Z_{n+1}$ are independent. Are they?

Hint: measurable functions of independent random variables are independent (you decide if you need to prove this).

1. ...I'm stuck.

Is what I've done right so far? Which parts are wrong? Where do I go from here?

Try structure your answer some more. Specify the events $B_i$ under consideration, e.g. notice that since each variable only takes 2 values there are not that many different types of events to consider.

First solve for the right hand side using, e.g., the argument that $$P(X_i = 1)=\mathbb E P(X_i = 1 \mid X_{i-1})=\mathbb E 1/2=1/2;$$ you have the right value.

Then solve for the left hand side using the Markov property as you have attempted.

• Thanks Student001. How do I show the process is Markov? What is the exact $\sigma$-algebra relation for the $X_i$'s? How do you know $P(X_i = 1 | X_{i-1}) = 1/2$ ? – BCLC Sep 11 '15 at 15:49
• i) You show it by checking the definition - don't bother with the $\sigma-$algebras. On points where $X_n = 1$, $X_{n+1}=1$ iff $Z_{n+1}=1$, and equivalently when changing to $-1$ in either place. But $Z_{n+1}$ is independent of $X_1,\dots,X_n$. Done with the Markov part (well, you write it rigorously). ii) Check the definition of $E \left( \mathbf{1}_{\{X_i = 1 \}} \mid X_{i-1}\right)$. Intuitively, it has to be 1/2 because $Z_i$ is independent of $X_{i-1}$ and takes each value with equal probability. – ekvall Sep 11 '15 at 16:45
• Sorry! Forgot to mention I already figured out Q3. Hehehe. Thanks. I'll try out Markov later. – BCLC Sep 11 '15 at 16:54
• Ok, let me know if you can't and I can add something to the answer later. – ekvall Sep 11 '15 at 17:07