[gecode-users] number of propagators

[Christian Schulte]

That's a feature and not a bug. The result of propagation is deterministic,
the order of how propagators are executed is not. Different order can the
result in a different number of propagations.

 

The order can differ because some propagators normalize their variables (for
example, linear, distinct, etc). Typically, normalization involves ordering
variables according to their addresses in memory (which is platform
dependent). Performing operations on differently ordered variables then will
also schedule propagators in different orders.

 

Hope that helps

Christian

 

--

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

 

From: users-bounces at gecode.org [mailto:users-bounces at gecode.org] On Behalf
Of Alejandro Arbelaez
Sent: Sunday, May 25, 2008 10:08 PM
To: users at gecode.org
Subject: [gecode-users] number of propagators

 

Hi,


I've been trying some gecode examples on different machines, a macbook-pro
2.4Ghz under  ubuntu 7.01  (i386),  gcc version 4.0.1 and a Intel Xeon  2.33
Ghz under Mandriva 2008 (x86_64), gcc version 4.2.3. Of course I'm running
the same gecode version in both machines (gecode-2-1-1) 

Despite of I'm getting the same search tree, the number of propagations is
different for some instances of some problems, for instance:


-------- Mandriva
magic-square -c-d 1 -a-d 1 order 7

initial
    propagators:   19
    branchings:    1

Summary
    runtime:       3.040 (3040.000000 ms)
    solutions:     1
    propagations:  7435467
    failures:      481301
    clones:        481365
    commits:       962666
    peak memory:   1214 KB

-----Ubuntu
magic-square -c-d 1 -a-d 1 7

Initial
        propagators:   19
        branchings:    1

Summary
        runtime:       3.870 (3870.000000 ms)
        solutions:     1
        propagations:  7430108
        failures:      481301
        clones:        481365
        commits:       962666
        peak memory:   609 KB


-----Mandriva
partition -a-d 1 -c-d 1 13
Initial
    propagators:   58
    branchings:    1

Summary
    runtime:       0.710 (710.000000 ms)
    solutions:     0
    propagations:  2397328
    failures:      49734
    clones:        49733
    commits:       99466
    peak memory:   569 KB

----ubuntu
partition -c-d 1 -a-d 1 13
Initial
    propagators:   58
    branchings:    1

Summary
    runtime:       1.813 (1813.487000 ms)
    solutions:     0
    propagations:  2397273
    failures:      49734
    clones:        49733
    commits:       99466
    peak memory:   285 KB

---Mandriva
partition -a-d 1 -c-d 1 15

Initial
    propagators:   66
    branchings:    1

Summary
    runtime:       8.060 (8060.000000 ms)
    solutions:     0
    propagations:  26991030
    failures:      546230
    clones:        546229
    commits:       1092458
    peak memory:   682 KB

-----ubuntu
partition -c-d 1 -a-d 1 15

Initial
    propagators:   66
    branchings:    1

Summary
    runtime:       20.200 (20200.376000 ms)
    solutions:     0
    propagations:  26995378
    failures:      546230
    clones:        546229
    commits:       1092458
    peak memory:   342 KB



Can someone explain me this behavior.



Thanks in advance,
Alejandro

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.ps.uni-sb.de/pipermail/users/attachments/20080526/4f78e89b/attachment.htm>