2

Consider the set $\mathbb{N}$ of natural numbers, including $0$. The structure $(\mathbb{N};+,\cdot)$ can define the standard order relation $\leq$. In fact, you don't even need multiplication to define the order relation. That raises the question, is it possible to define the standard order relation in the structure $(\mathbb{N};\cdot)$?

user107952
  • 20,508

1 Answers1

7

Any permutation of the primes extends to an automorphism of $(\mathbb{N};\cdot)$. Since we can permute primes without preserving order (to put it mildly), this means that the ordering is not definable in $(\mathbb{N};\cdot)$ via any logic, let alone first-order logic.

(OK, that's not quite fair - there isn't a universally agreed upon definition of "logic." However, any of the usual notions - e.g. "regular logic" a la Ebbinghaus/Flum/Thomas - have as a required property isomorphism invariance. The above applies to any such logic.)

To put this "logic-independence" claim in perspective, consider $\mathfrak{C}:=(\mathbb{N};\mathsf{suc})$ where $\mathsf{suc}$ is the successor map. The structure $\mathfrak{C}$ is rigid, so automorphism-based arguments cannot be used here. Moreover, $<$, $+$, and $\cdot$ are each definable in $\mathfrak{C}$ via second-order logic. However, none of them is $\mathsf{FOL}$-definable in $\mathfrak{C}$. Showing this latter point requires a more complicated argument. Basically, the point is that while automorphism-based arguments don't always succeed in proving non-definability results, when they do they prove extremely strong non-definability results.

Noah Schweber
  • 245,398
  • 1
    Separately, I believe that this exact point has come up repeatedly in your prior questions, e.g. here or here. In general, if you're curious whether something is definable, always start by thinking about the (orbits of the action of the) automorphism group. – Noah Schweber Nov 04 '21 at 16:10