如何解决有证明给定定理的好策略吗?
应该使用什么策略来证明这个结果(最后,被承认)?预先感谢您的任何提示。 :slight_smile:
希望这是真的定理。直觉错误时我已经很生气,发现了反例。
Require Import Permutation List Lia FunInd Recdef.
Set Implicit Arguments.
Inductive value := x0 | x1 | x2 | x3 | x4 | x5 | x6 | x7.
Inductive variable := aux | num: value -> variable.
Definition variable_eq_dec (x y: variable): {x=y} + {x<>y}.
Proof.
destruct x,y.
+ left; auto.
+ right; abstract congruence.
+ right; abstract congruence.
+ destruct v,v0; try (left; abstract congruence); (right; abstract congruence).
Defined.
Inductive assignment := assign: variable -> variable -> assignment.
Inductive comparison := GT: forall (more less: value),comparison.
Inductive step :=
| assignments: forall (L: list assignment),step
| conditional: forall (c: comparison) (positive negative: step),step.
Definition algorithm := list step.
Definition instantation := variable -> nat.
Definition list_of_values (i: instantation) :=
i (num x0) :: i (num x1) :: i (num x2) :: i (num x3) :: i (num x4) :: i (num x5) :: i (num x6) :: i (num x7) :: nil.
Definition is_permutation (i1 i2: instantation) := Permutation (list_of_values i1) (list_of_values i2).
Definition run_assignment (a: assignment) (i: instantation): instantation :=
match a with
| assign v1 v2 => fun x => if variable_eq_dec x v1 then i v2 else i x end.
Fixpoint run_assignment_list (L: list assignment): instantation -> instantation :=
match L with
| nil => fun i => i
| a :: l => fun i => run_assignment_list l (run_assignment a i)
end.
Fixpoint run_step (s: step) (i: instantation): instantation :=
match s with
| assignments L => run_assignment_list L i
| conditional (GT more less) pos neg =>
if Compare_dec.gt_dec (i (num more)) (i (num less)) then run_step pos i else run_step neg i
end.
Fixpoint run_algorithm (a: algorithm): instantation -> instantation :=
match a with
| nil => fun i => i
| s :: t => fun i => run_algorithm t (run_step s i)
end.
Definition permuting_step (s: step) := forall (i: instantation),is_permutation i (run_step s i).
Definition permuting_algorithm (a: algorithm) := forall (i: instantation),is_permutation i (run_algorithm a i).
Theorem permuting_algorithm_aux00 (a: algorithm) (s: step):
permuting_algorithm (s :: a) -> permuting_algorithm a /\ permuting_step s.
Proof.
Admitted.
编辑:根据Yves发现的反例,应该再添加至少两个条件。
Fixpoint compact_assignments (a: algorithm): Prop :=
match a with
| nil => True
| assignments L :: assignments L0 :: t => False
| x :: t => compact_assignments t
end.
Fixpoint no_useless_comparisons_in_step (s: step): Prop :=
match s with
| assignments L => True
| conditional (GT a b) pos neg => a <> b /\ no_useless_comparisons_in_step pos /\ no_useless_comparisons_in_step neg
end.
Definition no_useless_comparisons (a: algorithm) := forall x,In x a -> no_useless_comparisons_in_step x.
Definition compact_algorithm (a: algorithm) := compact_assignments a /\ no_useless_comparisons a.
Theorem permuting_algorithm_aux00 (a: algorithm) (s: step):
compact_algorithm (s :: a) -> permuting_algorithm (s :: a) -> permuting_algorithm a /\ permuting_step s.
Proof.
Admitted.
甚至还有反例,例如:
assignments (assign aux (num x1) :: assign (num x1) (num x0) :: nil) ::
conditional (GT x0 x1)
(assignments nil)
(assignments (assign (num x0) aux :: nil)) :: nil.
解决方法
这比Coq问题更多是数学问题。
可能有一个反例。请对此进行调查:一个赋值会混洗寄存器aux,x1,x2,...,x7的值。但是,当您寻找排列时,您只会看到x1,x2,...,x7的值。
假设您有一个步骤将x1的值存储到aux,将x2的值复制到x1和x2,并使所有其他寄存器保持不变。仅查看x1,...,x7中的值列表时,此步骤不是置换(由于重复)。叫这个 步骤s1。
然后考虑步骤s2,该步骤将aux的值复制到aux和x1中,并使所有其他值保持不变。同样,仅查看寄存器x1,...,x7时,这不是排列,因为它引入了以前不在这些寄存器中的值。
现在s1::s2::nil
是寄存器x1,...,x7上的标识函数。这是一个排列。但是s1和(s2 :: nil)都不是置换步骤或置换算法。
对于Coq计数器示例,足以证明s1不是置换步骤。在这里:
Definition la1 :=
assign aux (num x1) ::
assign (num x1) (num x2):: nil.
Definition la2 :=
assign (num x1) aux :: nil.
Definition s1 := assignments la1.
Definition s2 := assignments la2.
Lemma pa_all : permuting_algorithm (s1 :: s2:: nil).
Proof.
intros i.
unfold s1,s2,is_permutation.
unfold list_of_values; simpl.
apply Permutation_refl.
Qed.
Lemma not_permuting_step_s1 : ~permuting_step s1.
Proof.
unfold s1,permuting_step,is_permutation.
set (f := fun x => if variable_eq_dec x (num x1) then 0 else 1).
intros abs.
assert (abs1 := abs f).
revert abs1.
unfold list_of_values,f; simpl; intros abs1.
absurd (In 0 (1::1::1::1::1::1::1::1::nil)).
simpl; intuition easy.
apply (Permutation_in 0 abs1); simpl; right; left; easy.
Qed.
版权声明:本文内容由互联网用户自发贡献,该文观点与技术仅代表作者本人。本站仅提供信息存储空间服务,不拥有所有权,不承担相关法律责任。如发现本站有涉嫌侵权/违法违规的内容, 请发送邮件至 dio@foxmail.com 举报,一经查实,本站将立刻删除。