Counterexample to proposition on $\sigma$-algebras

190 Views Asked by At

If $\sigma$-algebras $\mathscr{F}$ and $\mathscr{G}$ are independent and if sigma-algebras $\mathscr{F}$ and $\mathscr{H}$ are independent then $\mathscr{F}$ and $\mathscr I := \sigma(\mathscr{G},\mathscr{H})$ are independent.

What is a counterexample to this?

If $\sigma$-algebras $\mathscr{F}$ and $\mathscr{G}$ are independent, then

$\forall F \in \mathscr{F}, G \in \mathscr{G}$,

$$P(F \cap G) = P(F)P(G)$$

If $\sigma$-algebras $\mathscr{F}$ and $\mathscr{H}$ are independent, then

$\forall F \in \mathscr{F}, H \in \mathscr{H}$,

$$P(F \cap H) = P(F)P(H)$$

So I need an $I \in \mathscr I$ s.t.

$$P(F \cap I) = P(F)P(I)$$

1

There are 1 best solutions below

0
On BEST ANSWER

Consider a probability space $(\Omega, \mathscr{F}, \mathbb{P})$, events A, B and C, pairwise independent but not mutually independent.

Then

$$P(A \cap B) = P(A)P(B)$$

$$P(A \cap C) = P(A)P(C)$$

$$P(C \cap B) = P(C)P(B)$$

$$P(A \cap C \cap B) \ne P(A)P(C)P(B)$$

Let:

$$\mathscr F = \sigma(A)$$

$$\mathscr G = \sigma(B)$$

and $\mathscr H = \sigma(C)$

Then $\mathscr I = \sigma(B, C)$.

$\mathscr F = \sigma(A)$ and $\mathscr G = \sigma(B)$ are independent.

$\mathscr F = \sigma(A)$ and $\mathscr H = \sigma(C)$ are independent.

We must show that $\mathscr F = \sigma(A)$ and $\mathscr I = \sigma(B, C)$ are not independent i.e. $\exists I \in \mathscr I$ s.t.

$$P(A)P(I) \ne P(A \cap I)$$

Now choose $I = B \cap C$ which works because $\mathscr G = \sigma(B)$ and $\mathscr H = \sigma(C)$ are independent.