$\sigma$-algebra of a $\sigma$-algebra: $\sigma(\sigma(\mathbb{R}))$

66 Views Asked by At

Let $\sigma(\cdot)$ denote the $\sigma$-algebra of its input. So $\sigma(\mathbb{R})$ is the $\sigma$-algebra of real numbers.

Question: I am wondering whether $\sigma(\sigma(\mathbb{R}))$ is equal to $\sigma(\mathbb{R})$, or it is different. (If the analysis is simpler, I am only wondering Borel $\sigma$-algebras.)

More explanation: The reason for my confusion is the following: $\sigma(\mathbb{R})$ is already closed under complements, countable unions, and countable intersections. So $\sigma(\sigma(\mathbb{R}))$ should be equal to $\sigma(\mathbb{R})$. However, the reason I do not fully agree with this statement is that $\sigma(\mathbb{R})$ is a set of intervals, hence $\sigma(\sigma(\mathbb{R}))$ should be a set of sets. I think $\sigma(\sigma(\mathbb{R}))$ is the powerset of $\sigma(\mathbb{R})$, but I have never heard of a powerset of an uncountable set.

Edit: Thanks to the amazing comments, this question is now resolved.

1

There are 1 best solutions below

2
On

The power set of $X$ is a $\sigma$-algebra of $X$. After all, it's closed under all the required operations. And, you can have power sets of uncountable sets just fine.

This is true for both $\mathbb{R}$ and $\sigma(\mathbb{R})$.

But, it's probably not the only one on $\sigma(\mathbb{R})$.

Your intuition that it can't be the same thing as $\sigma(\mathbb{R})$ because it should be composed of sets of sets is correct.

I don't know what a smaller algebra over $\sigma(\mathbb{R})$ would be.