Stream: Archive Mirror: Isabelle Users Mailing List

Topic: [isabelle] domintros generated by Function too weak


view this post on Zulip Email Gateway (Aug 18 2022 at 16:00):

From: Alexander Krauss <krauss@in.tum.de>

I have some more issues with the Function package, this time with the
domintros option.
[...]

I’m not sure if this is a bug in in the Function package, or an
unavoidable limitation.

It is an unavoidable limitation of the specific implementation, i.e., a
bug. I know it for a long time, and it is (again) caused by some
uncontrolled forward simplification (which is from the devil, but I
didn't know how bad it was when I wrote the code). Maybe it is time for
a new attempt of improving this... I'll have a look.

Alex

view this post on Zulip Email Gateway (Aug 18 2022 at 16:01):

From: Joachim Breitner <mail@joachim-breitner.de>
Hi,

I have some more issues with the Function package, this time with the
domintros option. Consider this (not very useful) example:


function (sequential,domintros) f where
"f v b = (if v = 1 then f 0 b else f (Suc v) b)"
apply pat_completeness
apply auto
done


The generated domintros rule is too weak:


thm f.domintros(1)[no_vars]
⟦f_dom (0, b); v ≠ Suc 0 ⟹ f_dom (Suc v, b)⟧ ⟹ f_dom (v, b)


Note that the f_rel intros are correct:
thm f_rel.intros[no_vars]


v = 1 ⟹ f_rel (0, b) (v, b)
v ≠ 1 ⟹ f_rel (Suc v, b) (v, b)


I can work around the issue by wrapping the condition of the if
statement in the identity function and removing it temporarily from the
simpset:


declare id_apply[simp del]
function (sequential,domintros) f' where
"f' v b = (if id (v = 0) then f' 0 b else f' (Suc v) b)"
apply pat_completeness
apply auto
done
declare id_apply[simp]


yields


thm f'.domintros(1)[no_vars]
⟦id (v = 0) ⟹ f'_dom (0, b); ¬ id (v = 0) ⟹ f'_dom (Suc v, b)⟧
⟹ f'_dom (v, b)


A probably related problem is that


function (sequential,domintros) f where
"f l = (case l of [a,b,0] ⇒ f [a+b])"
apply pat_completeness
apply auto
done


gives the domintro rule


thm f.domintros(1)[no_vars]
(⋀a aa. f_dom [a + aa]) ⟹ f_dom l [?]


where I am missing the connection between l and a,aa.

I’m not sure if this is a bug in in the Function package, or an
unavoidable limitation.

Thanks,
Joachim
signature.asc


Last updated: Nov 21 2024 at 12:39 UTC