Recursive definitions of $n<m$, $n\mid m$, and $n \bmod m$

265 Views Asked by At

Without referring to the apparatus of (primitive) recursive functions one can introduce addition into the language of successor arithmetic by two additional axioms which naturally reflect the essence of recursion:

  • $(\forall n)\ n + 0 = n$
  • $(\forall n, m)\ n+ m' = (n+m)'$

When we already have addition, it is easy to re-define $<$ explicitely:

  • $n < m :\equiv (\exists x \neq 0)\ n + x = m$

I wonder if the following is a sound recursive definition of $<$ only using the successor function:

  • $(\forall n)\ n < n'$
  • $(\forall n,m)\ n < m \rightarrow n < m'$

Other than the usual recursive definitions this one does not involve $0$, but it seems to work anyway. A proof that $0 < 2$ (with $2=1'=0''$) goes like this: Assume that $\neg(0 < 2)$. Since $\neg(n < m') \rightarrow \neg(n < m)$ it follows $\neg(0 < 0')$. Contradiction.

Note that the successor function could in turn be re-defined explicitely by the given relation $<$:

  • $m = n' :\equiv n < m \wedge \neg(\exists x)\ n < x < m$

I'd like to drive this further: Before defining multiplication (recursively) I'd like to define the intermediate relation $n \mid m$ ("$n$ divides $m$") and function $n\bmod m$ ("the remainder of $n$ divided by $m$") in a similar recursive fashion as $n < m$, especially without using multiplication (only addition).

The $n\mid m$ case seems easy:

  • $(\forall n \neq 0)\ n \mid n$
  • $(\forall n \neq 0,m \geq n)\ n \mid m \rightarrow n \mid (m + n)$

But I get stuck trying to define $n\bmod m$ in a similar fashion. Other than $n<m$ and $n\mid m$ it is a function, not a relation, but it should be possible, shouldn't it?

Any hint is welcome.

[Addendum]

Supposed that $n \bmod m$ can be defined by additional axioms in a recursive fashion, can addition be re-defined explicitely by $\bmod$ - just like the successor function can be re-defined explicitely by $<$?

1

There are 1 best solutions below

0
On

Maybe it's easier than I thought it is?

  • $(\forall m)\ 0\bmod m = 0$
  • $(\forall n,m)\ (n\bmod m)' = m \rightarrow n'\bmod m = 0$
  • $(\forall n,m)\ (n\bmod m)' \neq m \rightarrow n'\bmod m = (n\bmod m)'$