9
$\begingroup$

The Wikipedia page for $\sigma$-algebra says this set is called a "sigma-algebra" by some, and called a "sigma-field" by others. I'm writing a paper on measure theory, where the topic of sigma-algebra comes up, and wanted to use the correct term. So, I need to figure out which term is more appropriate: field or algebra?

I recall from abstract algebra that the definition of a field is a commutative ring (which itself is a triple $(S,+,\times)$ where $+:S\times S \to S$ and $\times: S \times S \to S$ are binary operators satisfying a number of properties), so that every nonzero element has a multiplicative inverse. The classic example of a field is the rational numbers, $\mathbb{Q}$.

The definition of an algebra, however, I do not recall learning. Wikipedia says an algebra is a vector space (which is itself a triple $(S,+,\cdot)$ where $+:S\times S \to S$ and $\cdot: \mathbb{R} \times S \to S$ are operators satisfying a number of properties) equipped with a bilinear product (what is this?).

Now, I can't immediately see how EITHER of these definitions relate to a $\sigma$-algebra. Let's look at the definition. A collect $\Sigma$ of subsets of $S$ is a $\sigma$-algebra in $S$ if $S \in \Sigma$, $\Sigma$ is closed under complementation, and $\Sigma$ is closed under countable unions.

So, we have the definitions of field, algebra and $\sigma$-algebra in front of us. I can't see how $\sigma$-algebra relates to either algebra or field. And back to the original question -- which name is more appropriate: "sigma-algebra" or "sigma-field"?

Thanks!


Edit As an additional question: When defining the $\sigma$-blank, would it be rigorous to define an algebra first, and then teach $\sigma$-algebra as an example (special case) of an algebra?

Edit 2 Thank you JHance for the awesome answer! With the new construction, let's show all the necessary properties are satisfied:

Additive properties:

  • $(A \cup B) - (A \cap B) = (B \cup A) - (B \cap A)$ (commutativity: $A+B=B+A$)
  • $(A \cup 0) - (A \cap 0) = A - 0 =A$ (additive identity: $A+0=A$)
  • $(A \cup A) - (A \cap A) = A - A =\emptyset$ (additive inverse: $A+A^{-1}=A$)

Multiplicative properties:

  • $A \cap B = B \cap A$ (commutativity: $A \times B = B \times A$)
  • $(A \cap B) \cap C) = A \cap (B \cap C)$ (associativity: $(A \times B) \times C = A \times (B \times C)$)
  • $A \cap X = A$ (multiplicative identity: $A \times 1 = A$)
  • $A \cap \emptyset = \emptyset$ (that is, $A \times 0 = 0$)

Distributive:

  • $((A \cup B) - (A \cap B)) \cap C =...$
$\endgroup$
  • 2
    $\begingroup$ Look up the Wikipedia entries on "universal algebra," "algebra of sets," "field of sets," "boolean algebra," and "boolean ring." Everything is answered in them. $\endgroup$ – whacka Aug 18 '14 at 2:45
  • 1
    $\begingroup$ The terms $\sigma$-algebra and $\sigma$-field are interchangeable in measure theory. They have nothing to do with the terms algebra and field one finds within topics of algebra. $\endgroup$ – user139388 Aug 18 '14 at 6:15
  • $\begingroup$ Well, that's disappointing. If that's the case, then neither name is appropriate. No wonder I couldn't see the connection. $\endgroup$ – Mathemanic Aug 18 '14 at 15:39
  • 3
    $\begingroup$ @Mathemanic In the most loose sense, an algebra is a set with operations, so the usage in measure theory is not completely inappropriate. $\endgroup$ – rschwieb Aug 27 '14 at 22:59
  • $\begingroup$ What would be the operations? An algebra is a vector space (or more loosely a set of elements) with two operations (addition and scalar multiplication) and equipped with a bilinear product, according to Wikipedia. However the elements of a $\sigma$-blank are subsets, not vectors. $\endgroup$ – Mathemanic Aug 28 '14 at 0:38
10
$\begingroup$

Let $X$ be a total space. If we let symmetric difference take the place of addition: $A \Delta B = (A \cup B) - (A \cap B)$, and let intersection be multiplication, then a sigma algebra of subsets of $X$ becomes a boolean ring with the empty set being $0$ and the total space being $1$. In general a family of sets closed under intersection and symmetric difference, and having the emptyset is a called a ring of sets and is actually a ring (or for some, a rng since it need not have unity), in addition it is a Boolean ring of characteristic 2 as for any $A$ we have $A+ A = \emptyset$ and $A^2 = A$.

An algebra of sets is a ring containing the total space, or in other words, it is a ring with unity, note that this is equivalent to closing the family under complements.

The notion of $\sigma$-rings/algebras corresponds to a more measure theoretic desire for countable operation, although they still behave well since all operations are Abelian.

Note that a $\sigma$-algebra with addition and multiplication given by symmetric difference and intersection is actually a (unital) algebra over the field $\mathbb{Z}/2\mathbb{Z}$, which can be imbedded in the algebra as the set $\{\emptyset,X\}$

Edit: Looking back at your question, you also expressed some confusion as to what an algebra is. For most things you'll see an algebra is a set $A$ that is at once equipped with the structures of a ring and vector space (or module) over a field ( or ring) $K$. However we want ring addition and vector addition on $A$ to coincide, and we want ring multiplication to be bilinear, which really just boils down to saying that for $v,w \in A$ and $\lambda \in K$, we want $$(\lambda v)\cdot w = v \cdot(\lambda w) = \lambda(v \cdot w)$$

If you're comfortable with rings, an elegant way of phrasing this in that context is simply that a unital algebra is a ring $A$ together with a ring of scalars $R$ and homomorphism of rings $f: R \to A$. Scalar multiplication is just $\lambda v = f(\lambda) \cdot v$

$\endgroup$
  • $\begingroup$ Dear JHance, thank you so much for the awesome answer! I'm updating the original post to try to show all the necessary properties are satisfied! $\endgroup$ – Mathemanic Aug 28 '14 at 2:10
  • $\begingroup$ I don't think that's really the most general sense of "an algebra"... $\endgroup$ – Malice Vidrine Aug 28 '14 at 2:39
  • $\begingroup$ @MaliceVidrine: Yes, my notion of algebra here is over a field, associative, and the ring homomorphism definition assumes the algebra is unital. I'll edit my language. $\endgroup$ – JHance Aug 28 '14 at 2:44

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.