Home Benchmark
Benchmark_cart
Your selection Format : wcsp
#Instances : 0
Format filter applied : wcsp
#Instance :
Benchmark


Project
login



online
We have 73 guests online
Home

Welcome on benchmark section

This section offers acces to cost function network instances collected from various sources



Quick Start

1. Select your instances in the tree directory. (please click on the Add button on the top the page )

2. Click on "Your selection" for access to the selection menu management . This menu allows to manage selection and download a shell script on your computer.

3. Execute the script on your computer. the selected instances will be fetch to the SVN server to your computer by your SVN client.

Warning

Instances can be huge . (repository space used is around 5 go) . So be careful to select only wanted instances

Required application :

the shell script generated by this web is based on SVN client call.

Under Linux like distro

  • install subversion client (apt-get install subversion subversion-tools )
  • execute the script downloaded

    Under windows

  • install svn client
  • change script extension from .sh to .bat

    (we work under linux so feedback is welcome from user working on other OS, please contact me . )

    INSTANCES FIELDS

    Graph decomposition

  • Graph representation correspond to a binary projection of the graph of constraints(i.e : global constraint will be replace by the corresponding click )
  • MCS decomposition : graph decomposition following the maximum cardinality search variable ordering
  • Min-fill decomposition : graph decomposition following Min fill variable ordering
  • Maximum space separator :(max_sep_space_size) is the cartesian product of domains of the variables included in the largest separator
  • total_separator_space : is the sum of the cartesian product of domains of the variables included in separators.

    degree distribution

  • represent the degree distribution in the constraint network after binary projection of nary constraint

    • Persistence of selection

  • By default your selection is not persistent. But for registered users, it is possible to store "your selection" for further usage or keep a pointer in the aims of allowing to the reproduction or the comparaison of articule results.


    • If you want to propose new instances , or suggest improvement please contact us :

    (costfunction_AT_toulouse.inra.fr )
    English (United Kingdom)

    Copyright © 2010 ---.
    All Rights Reserved.