Dear Mark,
For applying constraints that depend on the truth value of other
constraints you may also want to check out Oz' support for Constraint
Combinators (http://www.mozart-oz.org/documentation/system/
node44.html#chapter.combinator). Raphael already mentioned or. If you
want to make constrainable whether some arbitrary Oz code is executed
or not then Combinator.'reify' might be what you are looking for.
However, it is less efficient than your plain if statement (it
creates a space copy in the background). Here is a simple example
(there is no example in the doc): B = 1 if X = Y, and B = 0 if X \= Y.
B = {Combinator.'reify' proc {$} X = Y end}
Best
Torsten
On Jul 19, 2009, at 10:41 PM, mark richardson wrote:
Hi,
Thanks for your reply.
I really, really wasn't expecting this sort of reply. I'm now in
the uncomfortable position of trying to decide exactly what I DO
mean :+)
To put the problem into a clearer perspective, it concerns 3
variables A,B and C.
A is a FD integer which represents a virtual actor, B is a FS which
represents one or more other virtual actors.
C is a FD integer which represents an interaction which is possible
between virtual actors depending upon characteristics of those agents.
At the moment I'm using the format:
thread
if C==x then
<post constraints on A and B to match interaction x>
end
end
My thought was that I could possibly make speed improvements on
this by replacing the 'if' statement with something which would
cause the thread to fail sooner - namely the (C==x)=true idea.
Now I'm struggling to decide if saying something completely
different semantically by doing this will fundamentally alter the
function of my program or even whether I'm structuring my problem
logic correctly in the first place.
I think some more thought on this might be in order :+)
Regards
Mark
Raphael Collet wrote:
Dear Mark,
Your two threads simply implement different semantics! See below:
On Sun, Jul 19, 2009 at 10:05 PM, mark richardson
<[email protected]> wrote:
Hi,
I have a constraint based problem that implements constraints as
explicitly defined threads. An if statement guards whether the
constraint is imposed or not. For example one of the constraints
could be something like:
thread
if B==2 then
<post some constraint>
end
end
This threads tries to enforce the logic statement (B==2 and
<constraint>) or (B\=2). If B is different from 2, the logic
statement is equivalent to 'true'.
My program isn't as simple as this and there IS a good reason why
I'm doing things this way :+), but it demonstrates the basic idea.
My question is this (although I can't think of a simplified
example to test it with); would the above code block prove more
efficient like this:
thread
(B==2)=true
<post some constraint>
end
This second threads enforces the logic statement (B==2 and
<constraint>). In this case, if B is different from 2, the logic
statement is equivalent to 'false'. Therefore both threads are
not equivalent from a logic point of view.
Here I'm assuming this thread would signal failure whenever B was
not 2 and subsequently fail the space it was in.
In other words, are there any efficiency gains to be made by using
failure as a guard for the constraint rather than a logic
conditional statement? (Obviously I'm talking about a very large
number of calls to this thread.)
The first question is: which of the two logic statements
correspond to the problem?
For the first semantics, it is possible to use an 'or' statement,
and let the posted constraint be evaluated in a subspace.
or B=2 <post some constraint>
[] B\=:2
end
If the posted constraint fails, the 'or' statement will
automatically enforce the remaining clause, i.e., B\=2. Note that
this technique should be used with care, because it creates more
computation spaces, but it can pay off if the posted constraint is
likely to fail...
Cheers,
raph
--
Torsten Anders
Interdisciplinary Centre for Computer Music Research
University of Plymouth
Office: +44-1752-586219
Private: +44-1752-558917
http://strasheela.sourceforge.net
http://www.torsten-anders.de
_________________________________________________________________________________
mozart-users mailing list
[email protected]
http://www.mozart-oz.org/mailman/listinfo/mozart-users
