Hi,
Thank you very much Raphael. I can't really say at the moment how many
of the preferences would be conflicting as I'm working with a small
amount of test data, but I see your point.
I'll try your suggestion and then when I get a larger data sample, I'll
compare the two versions.
Thanks once again for all your help.
Regards
Mark
Raphael Collet wrote:
Dear Mark,
I wrote "simple", because this solution might be very efficient if
many of the preferences are conflicting. The issue is that a
propagator like
B = (X =<: Y)
will perform no propagation until either B is known, or both X and Y
are known. To get maximal performance, you need to have as much
propagation as possible. A good heuristics might be to determine the
B's as soon as possible. Something like
Bs = for X in Scenes collect:Collect do
for Y in Scenes do
if X\=Y andthen {IsPreced X Y} then %X is a precedent of Y
{Collect (Position.X=<:Position.Y)}
end
end
end
{FD.sum Bs '=:' NumPrec}
%% determine NumPrec, then the Bs, trying high values first, then the
positions
{FD.distribute generic(order:naive value:max) NumPrec|Bs}
{FD.distribute ff Position}
I haven't tried it, but that's something you could give a try at.
Cheers,
Raphael
On Thu, Feb 25, 2010 at 1:00 PM, mark richardson
<[email protected] <mailto:[email protected]>> wrote:
Hi,
Thank you so much for your help with this.
I did have another small problem that I have now solved - I had
{IsPreced X Y} returning true if Y had no precedences at all which
was causing chaos.
(Oh and I deliberately left the distribution of the position
vector out of the message, but thank you for your reminder :+) )
You say that this approach is a 'simple way to maximize the number
of preferences'.
I might regret asking this, but is there a way which is less
simple, but perhaps better?
Regards
Mark
Raphael Collet wrote:
Dear Mark,
Yes, I think you got it right. That is a simple way to maximize
the number of preferences. (Just don't forget to distribute the
vector Position in your script.)
Cheers,
Raphael
On Thu, Feb 25, 2010 at 12:26 PM, mark richardson
<[email protected] <mailto:[email protected]>> wrote:
Hi,
I've modified the script slightly to this:
{FD.sum
for X in Scenes collect:Collect do
for Y in Scenes do
if X\=Y andthen {IsPreced X Y} then %X is a precedent of Y
{Collect (Position.X=<:Position.Y)}
end
end
end '=:' NumPrec}
I realised the mistake I was making with X and Y instead of
Position.X and Position.Y and mistyping 'and' instead of
'andthen'
Regards
Mark
--
Mark Richardson MBCS
Research Assistant
University of Teesside, UK
Email: [email protected]
<mailto:[email protected]>
[email protected] <mailto:[email protected]>
Skype: mark.richardson.
_________________________________________________________________________________
mozart-users mailing list
[email protected] <mailto:[email protected]>
http://www.mozart-oz.org/mailman/listinfo/mozart-users
------------------------------------------------------------------------
_________________________________________________________________________________
mozart-users mailing list [email protected]
<mailto:[email protected]>
http://www.mozart-oz.org/mailman/listinfo/mozart-users
--
Mark Richardson MBCS
Research Assistant
University of Teesside, UK
Email: [email protected] <mailto:[email protected]>
[email protected] <mailto:[email protected]>
Skype: mark.richardson.
_________________________________________________________________________________
mozart-users mailing list
[email protected] <mailto:[email protected]>
http://www.mozart-oz.org/mailman/listinfo/mozart-users
------------------------------------------------------------------------
_________________________________________________________________________________
mozart-users mailing list
[email protected]
http://www.mozart-oz.org/mailman/listinfo/mozart-users
--
Mark Richardson MBCS
Research Assistant
University of Teesside, UK
Email: [email protected]
[email protected]
Skype: mark.richardson.
_________________________________________________________________________________
mozart-users mailing list
[email protected]
http://www.mozart-oz.org/mailman/listinfo/mozart-users
