I'm reading Halmos's Measure Theory and his definition of semiring seems to disagree with the ones that I find on the internet.
Halmos's definition (p. 22):
A semiring is a non empty class $\mathbf{P}$ of sets such that
- if $E\in\mathbf{P}$ and $F\in\mathbf{P}$. then $E\cap F\in\mathbf{P}$. and
- if $E\in\mathbf{P}$ and $F\in\mathbf{P}$ and $E\subset F$, then there is a finite class $\{C_0, C_1, \cdots, C_n\}$ of sets in $\mathbf{P}$ such that $E=C_0\subset C_1\subset\cdots\subset C_n=F$ and $D_i=C_i-C_{i-1}\in\mathbf{P}$ for $i=1,\cdots,n$.
Wikipedia's definition (for example):
A semiring (of sets) is a non-empty collection $S$ of sets such that
$\emptyset \in S$
If $E\in S$ and $F\in S$ then $E\cap F\in S$.
If $E\in S$ and $F\in S$ then there exists a finite number of mutually disjoint sets $C_{i}\in S$ for $i=1,\ldots ,n$ such that $E\setminus F=\bigcup _{i=1}^{n}C_{i}$.
These definitions are not equivalent! For example, the collection $\{\emptyset,\{a\},\{b\}, \{c\}, \{a,b,c\}\}$ is a semiring under the second definition, but not the first.
Questions:
- History question: why does Halmos use a different definition than we do today? Was the definition weakened at some point in order to be more general?
- Math question: what are the advantages/disadvantages of these two definitions from the standpoint of measure theory?