Consider the structure of the positive real numbers $(0, \infty) $ with its unit $1$, its addition $+$, its multiplication $\times $, and its strict ordering $> $. Is this structure $$( (0, \infty), 1, +, \times, >) $$ o-minimal?

  • $\begingroup$ Note that $>$ is definable from $+$ here: $a<b$ iff $\exists c(a+c=b)$. $\endgroup$ – Noah Schweber Feb 27 '19 at 16:36

Another way to see o-minimality is to note that $log$ induces an isomorphism $$log : ((0,\infty), 1, +, \times, >) \cong (\mathbb R, 0, \oplus, +, >),$$ where $\oplus$ is the binary operation defined by $\oplus(x,y) = \log(e^x + e^y)$. But $\oplus$ is definable in $\mathbb R_{exp}$. Hence the latter structure is a reduct of $\mathbb R_{exp}$ and therefore is o-minimal.


Yes. To see this, first note that we can interpret the positive reals in $(\mathbb{R},0,1,+,-,\times)$ as squares of non-zero real numbers. We can by induction on complexity of formulas define a translation map from formulas in the language of the positive reals to formulas in the language of the reals as follows:

  • For an atomic formula $\phi(x_1,\dots x_k)$ by its translation is $T(\phi)(y_1, \dots, y_k) = \phi(y_1^2, \dots , y_k^2)$

  • $T(\phi \wedge \psi) = T(\phi) \wedge T(\psi)$

  • $T(\neg \phi) = \neg T(\phi)$

  • $T(\exists x \phi(x)) = \exists y (y \neq 0 \wedge T(\phi)(y^2))$

Then the postive reals model $\phi(c)$ for $c > 0$ if and only the reals models $T(\phi)(\sqrt{c})$, for formulas with a single free variable of arbitrary complexity.

But then using o-minimality of the reals, we know $\{y : T(\phi)(y)\}$ is a finite union of points and intervals, so the same is true of $\{c : \phi(c)\} = \{c : T(\phi)(\sqrt{c})\}$

  • $\begingroup$ So by the same argument, also the nonnegative reals $[0, \infty) $ is o-minimal, right? Thank you, James! $\endgroup$ – Colin Tan Feb 27 '19 at 14:59
  • 4
    $\begingroup$ @ColinTan More simply, your structure is an expansions-by-definitions of a definable substructure of $(\mathbb{R}; +,\times)$, and so o-minimality immediately follows. $\endgroup$ – Noah Schweber Feb 27 '19 at 16:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.