Haskell - 使用归纳来证明其含义

时间:2015-12-07 16:50:07

标签: haskell theorem-proving induction implication

我要通过归纳来证明

no f xs ==> null (filter f xs)

其中:

filter p []    = []
filter p (x:xs) 
  | p x        = x : filter p xs
  | otherwise  = filter p xs

null [] = True; null _ = False 

no p [] = True
no p (x:xs)
  | p x = False
  | otherwise = no p xs

Logic implication:
True ==> False = False
_    ==> _     = True

所以,我认为以下是我的假设和我的主张:

Assumption: no f xs ==> null (filter f xs)
Claim: no f (x:xs) ==> null (filter f (x:xs))

我开始尝试证明基本情况(空列表):

no f [] ==> null (filter f [])
== {- Filter.1, filter p [] = [] -}
no f [] ==> null ([])
== {- No.1,  no p [] = True-}
True ==> null ([])
== {- Null.1, null [] = True -}
True ==> True 

但是我不确定它是否正确,因为我已经证明它们都是真的而不是如果左边的部分是真的而第二部分是假的,那么暗示就是假的(那个是==>)的定义。 它是否正确? 我怎样才能继续证明? 我不清楚如何使用归纳来证明其含义......

提前谢谢!

1 个答案:

答案 0 :(得分:1)

这是完整的证据。后来,当我有更多的时间,我将在Agda或Idris上证明这一点并在此处发布代码。

通过xs的归纳证明。

案例xs = []

no f [] ==> null (filter f [])
== {- Filter.1, filter p [] = [] -}
no f [] ==> null ([])
== {- No.1,  no p [] = True-}
True ==> null ([])
== {- Null.1, null [] = True -}
True ==> True 

案例xs = y : ys。假设no f ys == null (filter f ys)。请考虑以下情况:

案例f y == True

no f (y : ys) ==> null (filter f (y : ys))
== {- no - f y == True -}
False ==> null (filter f (y : ys))
== 
True

案例f y == False

no f (y : ys) ==> null (filter f (y : ys))
=={- By definition of filter and no -}
no f ys ==> null (filter f ys)
== {By I.H.}
True

完成证明。