[gecode-users] node count

Christian Schulte cschulte at kth.se
Wed Jul 23 20:02:12 CEST 2014


You access the statistics before you actually start to search. That's wrong:
the statistics returned is always wrt the current state of the search. So,
what you want to do is to access stat() after you do the search.

Cheers
Christian

--
Christian Schulte, www.ict.kth.se/~cschulte/

> -----Original Message-----
> From: users-bounces at gecode.org [mailto:users-bounces at gecode.org] On
> Behalf Of kellen at cretin.net
> Sent: Tuesday, July 22, 2014 8:31 PM
> To: users at gecode.org
> Subject: Re: [gecode-users] node count
> 
> > I don't really see what you are doing.
> 
> 
> My code is basically:
> 
>     Search::Options so;
>     so.stop = CombinedStop::create(0, 0, 120000, true);
>     CombinedStop::installCtrlHandler(true);
>     DFS<BenchmarkBase> e(s, so);
>     Search::Statistics stat = e.statistics();
>     BenchmarkBase* ex = e.next();
>     while(ex != NULL) {
>         ex->print(out);
>         delete ex;
>         ex = e.next();
>     }
>     std::cerr
>       << "\tpropagations: " << stat.propagate << std::endl
>       << "\tnodes:        " << stat.node << std::endl
>       << "\tfailures:     " << stat.fail << std::endl
>       << "\trestarts:     " << stat.restart << std::endl
>       << "\tno-goods:     " << stat.nogood << std::endl
>       << "\tpeak depth:   " << stat.depth << std::endl
>       << std::endl;
> 
> 
> In which my benchmarking model (a class derived from BenchmarkBase) uses
my
> custom brancher and value selection. When the val() method of my value
> selection class is called, I am printing out some debugging information,
so I can
> see that it is making a value choice.
> 
> I would have expected that `stat.node` would be increased each time the
value
> selection function is called, but it always returns `0`.
> 
> Is that clearer?
> 
> > But anyway: have you tried Gist to see what search does?
> 
> I haven't tried gist as I'm running my code on a server without X
installed.
> I guess that's the next step if I can't figure out what's going on.
> 
> Cheers,
> Kellen
> 
> _______________________________________________
> Gecode users mailing list
> users at gecode.org
> https://www.gecode.org/mailman/listinfo/gecode-users




More information about the users mailing list