Suppose we have $X=\{1,2,3\}$ and $M$ a $\sigma$-algebra over X.

I've stumped across the question of finding the dimension of the vector space of all the $M$-measurable real functions over $X$ (e.g.$\quad f:X \to \mathbb{R}$). For the best of me i can't seem to wrap my head around this, even if i think it's rather intuitive.

My difficulty is coming up with a possible basis of functions for the vector space of functions from $X$ to $\mathbb{R}$. Once i get that checking for misurability is trivial in this case, but i don't know how to approach basis of functions.

Of course functions from $X$ to $\mathbb{R}$ have at most 3 distinct values. I thought of using as basis 3 separate functions, one for each value of $X$: $e_i(i)=1\quad i=1,2,3$.

But i don't think this cuts it since i can't come up with a way of writing every f via linear combination of these basis functions. Any hint on how to tackle this problem? Or maybe a general idea of how a base function should look like?



The dimension depends on $M$. If $M=\{X,\emptyset\}$ then measurable functions are constant, so the dimension is $1$. If, as you seem to be assuming, $M$ is the power set of $X$ then every function is measurable. In that case if you define $$e_x(y)=\begin{cases}1,&(y=x),\\0,&(y\ne x).\end{cases}$$then yes, the functions $e_x$ for $x\in X$ form a basis.

Hint for that: $(2,3,7)=2(1,0,0)+3(0,1,0)+7(0,0,1)$.

  • $\begingroup$ If we assume $M=\{X, \emptyset\, \{1\}, \{2,3\}\}$ then this has dimension 2 with $(1,0,0), (0,1,1)$ as basis using your vectorial notation. Am i correct? $\endgroup$ – WhiteEyeTree Jan 12 '19 at 18:06
  • $\begingroup$ @WhiteEyeTree Yes. $\endgroup$ – David C. Ullrich Jan 12 '19 at 18:07

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.