[sv-ac] some thoughts on non-vacuity

From: John Havlicek <john.havlicek_at_.....>
Date: Mon Jan 14 2008 - 06:09:59 PST
Hi Doron:

I was supposed to think some more about the non-vacuity 
definitions.

Below are some notes on the thoughts I have had.  I am not
sure that these definitions are good, but there are at least
some good properties that result.

It would be good to check whether any of the definitions below
satisfies the finite prefix intuition.  I have not checked them.

I think that further review and discussion are needed.

J.H.

--------------------------------------------------------------------------

My intuition is fairly strong that if there is a finite prefix u <= w on which p
is "decided", i.e. such that either u\bot^\omega |= p or u\top^\omega |/= p
(modulo the problems with intersection that we know exist with this
formulation), then we should have w |=^{non} p iff u |=^{non} p.

Here are some possible definitions that seem intuitively reasonable to me:

   w |=^{non} always p
   iff  there exists 0 <= i < |w| such that w^{i..} |=^{non} p and such that for
        all 0 <= j < i, w^{j..} |= p

   w |=^{non} s_eventually p
   iff  there exists 0 <= i < |w| such that w^{i..} |=^{non} p and such that for
        all 0 <= j < i, w^{j..} |= not p

SANITY CHECK 1:  

   w |=^{non} not always not p
   iff  w |=^{non} always not p
   iff  there exists 0 <= i < |w| such that w^{i..} |=^{non} not p and such that for
        all 0 <= j < i, w^{j..} |= not p
   iff  there exists 0 <= i < |w| such that w^{i..} |=^{non} p and such that for
        all 0 <= j < i, w^{j..} |= not p
   iff  w |=^{non} s_eventually p

////

Here is a somewhat weak definition of |=^{non} for the untils:

   w |=^{non} p until q
   iff  w |=^{non} p until_with q
   iff  w |=^{non} p s_until q
   iff  w |=^{non} p s_until_with q
   iff  there exists 0 <= i < |w| such that (either w^{i..} |=^{non} p or 
        w^{i..} |=^{non} q) and such that for all 0 <= j < i, w^{j..} |= p and not q

Let

  vacuous_false = 0 |-> 0.
  vacuous_true = not(vacuous_false)

Assume that w is a non-empty natural word (i.e., no letter of w is \top or \bot).  
Then 

  w |/= vacuous_false and w |/=^{non} vacuous_false
  w |=  vacuous_true  and w |/=^{non} vacuous_true


SANITY CHECK 2:  Assume that w is a natural word (i.e., no \top or \bot letter):

   w |=^{non} p until vacuous_false
   iff  there exists 0 <= i < |w| such that (either w^{i..} |=^{non} p or 
        w^{i..} |=^{non} vacuous_false) and such that for all 0 <= j < i, 
        w^{j..} |= p and not vacuous_false
   iff  there exists 0 <= i < |w| such that w^{i..} |=^{non} p and such that for 
        all 0 <= j < i, w^{j..} |= p
   iff  w |=^{non} always p

////

SANITY CHECK 3:  Assume that w is a natural word (i.e., no \top or \bot letter):

   w |=^{non} vacuous_true s_until p
   iff  there exists 0 <= i < |w| such that (either w^{i..} |=^{non} vacuous_true or 
        w^{i..} |=^{non} p) and such that for all 0 <= j < i, 
        w^{j..} |= vacuous_true and not p
   iff  there exists 0 <= i < |w| such that w^{i..} |=^{non} p and such that for 
        all 0 <= j < i, w^{j..} |= not p
   iff  w |=^{non} s_eventually p

////

SANITY CHECK 4:

   w |=^{non} not(not p until_with not q)
   iff  w |=^{non} (not p until_with not q)
   iff  there exists 0 <= i < |w| such that (either w^{i..} |=^{non} not p or 
        w^{i..} |=^{non} not q) and such that for all 0 <= j < i, 
        w^{j..} |= not p and not not q
   iff  there exists 0 <= i < |w| such that (either w^{i..} |=^{non} p or 
        w^{i..} |=^{non} q) and such that for all 0 <= j < i, w^{j..} |= q and not p
   iff  w |=^{non} q s_until p

////


Note that

   w |=^{non} (p until q) and s_eventually q 
   iff  either
           w |=^{non} (p s_until q) 
        or
           w |=^{non} s_eventually q
   <=   w |=^{non} p s_until q

but this is not an equivalence if q has a non-vacuous evaluation on a suffix of
a word on which p fails vacuously (e.g., take p to be vacuous_false). 

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
Received on Mon Jan 14 06:10:47 2008

This archive was generated by hypermail 2.1.8 : Mon Jan 14 2008 - 06:10:58 PST