3
$\begingroup$

Is there a proof that all terms of type $\forall{a}. a$ are operationally $\bot$, in a non-weakly-normalising version of System F?

If you ask a free theorem calculator such as this one for the free theorem of $\forall{a}. a$ (written just $a$, as the tool adds implicit $\forall$s for each free type variable), you get the rather unhelpful

forall t1,t2 in TYPES, R in REL(t1,t2). (f_{t1}, f_{t2}) in R

Which seems to me only to imply that instantiations of $f$ at any two types are related by any relation.

Does this imply that $f$ must be a looping term?

$\endgroup$
1
$\begingroup$

Sure, just take R to be $\{(\bot,\bot)\}$, the relation which only relates bottoms.

This is a strict continuous relation, so it satisfies the requirements of the free theorem you linked to.

$\endgroup$

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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