Stream: Mirror: Isabelle Users Mailing List

Topic: [isabelle] Sledgehammer proof reconstruction failed


view this post on Zulip Email Gateway (Oct 13 2020 at 18:34):

From: Jakub Kądziołka <kuba@kadziolka.net>
Hello,

I was trying to infer the following:

theory Scratch
imports Complex_Main
begin

notepad begin
fix a n :: nat
assume "a dvd n" "a < n powr (1/3)" "1 < n"
then have "a < n" by (smt divide_le_eq_1_pos of_nat_1_eq_iff of_nat_less_iff powr_less_cancel powr_one_gt_zero_iff)
end

This smt call has been suggested by Sledgehammer, but with the words
"proof reconstruction failed". Here is the exact error it generates:

exception THM 1 raised (line 2237 of "thm.ML"):
RSN: no unifiers
1 < n
¬ real n + - 1 * real 1 ≤ 0 ⟹ 1 < n

Is this a bug? It certainly seems like one...

Regards,
Jakub Kądziołka

view this post on Zulip Email Gateway (Oct 31 2020 at 08:50):

From: Mathias Fleury <mathias.fleury12@gmail.com>
Hi Jakub,

I had a look at that and it is a bug in the proof produced by Z3.

One proof step (modus ponens) is:

1 < n      real n + - 1 * real 1 ≤ 0 <--> 1 < n
================================================ mp
        real n + - 1 * real 1 ≤ 0

The problem is that the rule mp is specified to have the form:

P      P<-->Q
============== mp
    Q

However, in this proof step produced by Z3, the terms P and Q of the
equivalence are swapped. So, this is an issue in the proof.

The bad news is: it makes no sense to report the issue, because our Z3
version is too old -- and the proof format has changed compared to the
version that is currently used, so it is not only about updating the
binary shipped with Isabelle.

The second bad news: using veriT (the new alternative backend for smt)
works but it takes 7s, so no chance here either...

Thanks for reporting,

Mathias


Last updated: Sep 25 2021 at 09:17 UTC