In Vector Integration and Stochastic Integration in Vector spaces, Nicolae Dinculeanu states the following theorem:
Given a measure space: $(S, \Sigma)$, a function $f : S —> R$ is $\Sigma$-measurable iff there is a sequence ($f_n$) of finite, real-valued, $\Sigma$ - step functions such that $f_n \rightarrow f$ pointwise.
I see that if $f$ is measurable then this implies the existence of measureable step functions that converge. (Using $f_n = 2^{-n}\lfloor 2^n f \rfloor$.) However I can't see why the converse holds.
Any help is much appreciated!