If $f:\mathbb{N}\to\mathbb{N}$ is any strictly increasing function with $f(0)=1$, define the base $f$ notation for natural numbers inductively as follows:
$0$ is represented as $()$ (the empty sequence).
If $n>0$, the representation of $n$ is defined as follows. Let $k$ be maximal such that $f(k)\leq n$. The representation of $n$ is $(f(k))$ (the sequence of length $1$ whose lone entry is $f(k)$) concatenated with the representation of $n-f(k)$.
For example, if $f(n)=10^n$, we get (a version of) decimal notation by the above construction. E.g., the representation of $321$ in that case is $(100,100,100,10,10,1)$.
Now, I'm quite certain this concept is not original (if it were, I'd be shocked to death). Does anybody know who has written about it? (Modulo variations in the exact details, of course)
(See also: "base Fibonacci", which is somewhat similar to the above idea with $f(n)=Fib(n+1)$, but not exactly, as it uses properties of the Fibonacci sequence to make the representations nicer, in a way which does not generalize)
