A kind of status bar to see where the Search is in csp

Tim Heath wgheath at saber.net
Fri Jun 9 18:57:00 CEST 2006


Is it an insane idea to have a type of heuristic for finding a solution
that says something like:

Find at least 1 solution

Spend an hour or some amount of time looking for a better solution.  If
in an hour no superior solution has been found take the first solution.

The reason I like this idea is as the speed of computers go up the
heuristic will improve automatically.

Luis Quesada wrote:
> Dear Torsten,
>> I really like this approximation you propose. Of course, the 
>> 'progress  bar' may quick go to, say, 90% and then stay there forever 
>> and  eventually fall back again. Still, that's at least some feedback..
>
> Indeed the progress reported does not evolve monotonically. It is also 
> true that you may get stuck at 90%. However, this might suggest that 
> it is time to abort your search and have a look at the remaining 10%. 
> Maybe the reason why it is so hard to determine them is a bug in the 
> input data...
>
> In fact, I have found this approximation useful when dealing with 
> tough CSP where finding one solution may take hours.
>
> Cheers,
> Luis
>>
>> On 08.06.2006, at 22:47, Luis Quesada wrote:
>>
>>> Well, you could consider the following approximation: to associate 
>>> the  status of your search with the number of variables determined 
>>> so far.
>>> For this you got two options.
>>> 1. (naive) have a thread per variable that sends a message to a 
>>> port  when the variable has been determined:
>>> thread {Wait X} {Send P incCountOfVarsDetermined)} end
>>> The thread reading the messages will print the status when 
>>> receiving  the incCountOfVarsDetermined messages.
>>> 2. To keep a cell and update the contents of that cell at each  
>>> distribution step with the number of variables that have been 
>>> filtered  out. For this you will need to write your own distributor. 
>>> Have a look  at  
>>> http://www.mozart-oz.org/documentation/fdt/node40.html#chapter.user- 
>>> defined
>>>
>>> Luis
>>>
>>> Torsten Anders wrote:
>>>
>>>> The short answer is: this is not possible unfortunately.
>>>> To realise this you would need to know the shape of your search 
>>>> tree   and where the solutions are in this search tree. But if you 
>>>> already   know that, then there is no search required anyway -- you 
>>>> would just   directly go to that solution.
>>>> I'm no guru, though ;-)
>>>> Best,
>>>> Torsten
>>>> On 08.06.2006, at 21:45, Tim Heath wrote:
>>>>
>>>>> Hi Oz Gurus,
>>>>>
>>>>> I want to make a kind of status bar showing how the search is 
>>>>> going   for csp application.  Any ideas how to do that?  I know if 
>>>>> I look at   the explorer it shows the tree as it is being 
>>>>> populated.  What I am   not clear on is if there is a way to know 
>>>>> if it is like 50% done or   75% done at any period of time 
>>>>> programatically.
>>>>>
>>>>> Oz rocks!
>>>>>
>>>>> _____________________________________________________________________ 
>>>>> __ __________
>>>>> mozart-users mailing list                                 
>>>>> mozart-users at ps.uni-sb.de
>>>>> http://www.mozart-oz.org/mailman/listinfo/mozart-users
>>>>>
>>>>>
>>>> -- 
>>>> Torsten Anders
>>>> Sonic Arts Research Centre • Queen's University Belfast
>>>> Frankstr. 49 • D-50996 Köln
>>>> Tel: +49-221-3980750
>>>> www.torsten-anders.de
>>>> strasheela.sourceforge.net
>>>
>>>
>>> _______________________________________________________________________ 
>>> __________
>>> mozart-users mailing list                                
>>> mozart-users at ps.uni-sb.de
>>> http://www.mozart-oz.org/mailman/listinfo/mozart-users
>>>
>>>
>> -- 
>> Torsten Anders
>> Sonic Arts Research Centre • Queen's University Belfast
>> Frankstr. 49 • D-50996 Köln
>> Tel: +49-221-3980750
>> www.torsten-anders.de
>> strasheela.sourceforge.net
>>
>>
>
> _________________________________________________________________________________ 
>
> mozart-users mailing list                               
> mozart-users at ps.uni-sb.de
> http://www.mozart-oz.org/mailman/listinfo/mozart-users
>






More information about the mozart-users mailing list