Skip to content

Refutation unsoundness on NRA formula #4765

@wintered

Description

@wintered

Hi,

Consider the following trace. The second part (after the reset) shows that the given model makes the formula sat.

[506] % z3release small.smt2
unsat
sat
[507] % cat small.smt2
(declare-fun a () Real)
(declare-fun b () Real)
(declare-fun c () Real)
(declare-fun d () Real)
(declare-fun g () Real)
(assert
 (not
 (exists ((e Real))
  (forall ((f Real))
  (=> (xor (=> (= f 0) (= a 0))
     (>= (+ (* (/ (- 1) b) e e) g) d)
     (<= (/ (- 1) b e a) c))
   (> b 0)
   (>= (div (to_int a) (to_int a)) (- b)))))))
(check-sat)
(reset)
(define-fun a () Real 0.0)
(define-fun b () Real 1.0)
(define-fun c () Real (- 2.0))
(define-fun d () Real 1.0)
(define-fun g () Real 0.0)
(assert
 (not
 (exists ((e Real))
  (forall ((f Real))
  (=> (xor (=> (= f 0) (= a 0))
     (>= (+ (* (/ (- 1) b) e e) g) d)
     (<= (/ (- 1) b e a) c))
   (> b 0)
   (>= (div (to_int a) (to_int a)) (- b)))))))
(check-sat)
[508] %

OS: Ubuntu 18.08
Commit: da0e140

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions