net.sf.cpsolver.ifs.example.csp
Class Test

java.lang.Object
  net.sf.cpsolver.ifs.example.csp.Test

public class Test
extends Object

Test of Structured CSP problems. It takes one argument -- property file with all the parameters. It allows to execute given number of tests. It also allows to define several configurations which will be executed. For instance CSP(20,15,5%..95%,5..95%), 10 runs of each configuration. All such configuration are processed in one run of Test class.

In Structured CSP, variables are divided into several kernels (some variables may remain ouside kernels). Different constraints (in density and tightnes) are generated according to whether variables are from the same kernel or not.

Test's parameters:

Parameter	Type	Comment
General.MPP	String	Minimal perturbation problem (if true), this mj. means that initial assignment will be generated
CSP.Seed	Long	Random number generator seed, System.currentTimeMillis() is taken if not present
CSP.ForceSolutionExistance	Boolean	If true, generated problem will always have at least one feasible solution
CPS.NrTests	Integer	Number of tests (for each input configuration)
CSP.NrVariables	Integer	Number of variables
CSP.NrVariablesMin CSP.NrVariablesMax CSP.NrVariablesStep	Integer	Range of the number variables (a set of different configurations will be generated) Use either CSP.NrVariables or these CSP.NrVariablesMin, CSP.NrVariablesMax, CSP.NrVariablesStep
CSP.DomainSize	Integer	Number of values of every variable
CSP.DomainSizeRatio	Double	Number of values as a ration of the number of variables. This way we can model for instance CSP(N,2N,p1,p2) problems with one configuration. Use either CSP.DomainSize or CSP.DomainSizeRatio
CSP.Tightness	Double	Tightness of constraints outside kernels
CSP.TightnessMin CSP.TightnessMax CSP.TightnessStep	Double	Tightness of constraints outside kernels given as a range -> respective configurations will be generated and tested
CSP.Density	Double	Density of constraints outside kernels
CSP.DensityMin CSP.DensityMax CSP.DensityStep	Double	Density of constraints outside kernels given as a range -> respective configurations will be generated and tested
CSP.NrKernels	Integer	Number of kernels (Structured CSP, use 0 for "normal" CSP)
CSP.KernelSize	Integer	Number of variables in each kernel
CSP.KernelTightness	Double	Tightness of constraints inside a kernel
CSP.KernelDensity	Double	Density of constraints inside a kernel
CSP.SameProblemEachStep	Boolean	If true, each configuration will start with the same seed
CSP.SameProblemEachTest	Boolean	If true, each test of the same configuration will start with the same seed
General.Output	String	Output folder where a log file and tables with results. In order not to overwrite the results if executed more than once, a subfolder with the name taken from current date and time will be created in this folder and all results will go to this subfolder.

Also, the configuration file can consist only from one parameter (named INCLUDE_REGEXP) which is processed as a regular expression of semicolon separated list of property files, for instance

INCLUDE_REGEXP=general.ini;{CSP(50,12,250,p2)|CSP(25,15,198,p2)}.ini;{std|opt}.ini;{10x1min}.ini;{cbs|rw1|tabu20}.ini

where {a|b|c|...} means a selection of a, b, c, .. All possible combinations are taken and for each of them an input configuration is combined from the relevant files. So, for instance, the above example will result into the following configurations:

• general.ini;CSP(50,12,250,p2).ini;std.ini;10x1min.ini;cbs.ini
• general.ini;CSP(50,12,250,p2).ini;std.ini;10x1min.ini;rw1.ini
• general.ini;CSP(50,12,250,p2).ini;std.ini;10x1min.ini;tabu20.ini
• general.ini;CSP(50,12,250,p2).ini;opt.ini;10x1min.ini;cbs.ini
• general.ini;CSP(50,12,250,p2).ini;opt.ini;10x1min.ini;rw1.ini
• general.ini;CSP(50,12,250,p2).ini;opt.ini;10x1min.ini;tabu20.ini
• general.ini;CSP(25,15,198,p2).ini;std.ini;10x1min.ini;cbs.ini
• general.ini;CSP(25,15,198,p2).ini;std.ini;10x1min.ini;rw1.ini
• general.ini;CSP(25,15,198,p2).ini;std.ini;10x1min.ini;tabu20.ini
• general.ini;CSP(25,15,198,p2).ini;opt.ini;10x1min.ini;cbs.ini
• general.ini;CSP(25,15,198,p2).ini;opt.ini;10x1min.ini;rw1.ini
• general.ini;CSP(25,15,198,p2).ini;opt.ini;10x1min.ini;tabu20.ini

To be able to distinguish such configuration a subfolder in General.Output folder is created, its name is combined from the names which are in parenthesis. So, for instance the first bunch of tests will output into the folder:

${General.Output}\CSP(50,12,250,p2)_std_10x1min_csb\25-Feb-05_191136

If one parameter is defined in more than one configuration files (e.g. in general.ini as well as cbs.ini) the one from the file more on the right is taken.

An example of the configurations:
File general.ini

#Default settings common for all configurations
General.MPP=false
General.InitialAssignment=false
General.Output=output\\RandomCSP\\IFS

#Value selection heuristics
Value.Class=net.sf.cpsolver.ifs.heuristics.GeneralValueSelection
Value.WeightWeightedConflicts=0.0
Value.RandomWalkProb=0.0
Value.WeightConflicts=1.0
Value.WeightNrAssignments=0.0
Value.WeightValue=0.0
Value.Tabu=0

#Variable selection heuristics
Variable.Class=net.sf.cpsolver.ifs.heuristics.GeneralVariableSelection
Variable.RandomSelection=true

#Termination condition
Termination.Class=net.sf.cpsolver.ifs.termination.GeneralTerminationCondition
Termination.MaxIters=-1
Termination.TimeOut=-1
Termination.StopWhenComplete=true

#Solution comparator
Comparator.Class=net.sf.cpsolver.ifs.solution.GeneralSolutionComparator


File CSP(50,12,250,p2).ini

#Sparse problem CSP(50,12,250/1225,p2)
CSP.NrVariables=50
CSP.DomainSize=12
CSP.Density=0.2
CSP.TightnessMin=0.10
CSP.TightnessMax=0.95
CSP.TightnessStep=0.02

CSP.Seed=780921

CSP.ForceSolutionExistance=false
CSP.SameProblemEachStep=false
CSP.SameProblemEachTest=false

CSP.NrKernels=0


File std.ini

#Standard problem
CSP.ForceSolutionExistance=false


File opt.ini

#Optimization problem (minCSP)
#Value selection: use weigh of a conflict, but when there are more than one value
# with the same number of conflicts, use the one with lower value
Value.WeightValue=0.0001
Value.WeightConflicts=1.0
#Do not stop when a complete solution is found
Termination.StopWhenComplete=false


File 10x1min.ini

#For each configuration, execute 10 tests, each with 1 minute timeout
CPS.NrTests=10
Termination.TimeOut=60


File cbs.ini

#Use conflict-based statistics
Extensions.Classes=net.sf.cpsolver.ifs.extension.ConflictStatistics
Value.WeightWeightedConflicts=1.0


File tabu20.ini

#Use tabu-list of the length 20
Value.Tabu=20


File rw1.ini

#Use 1% random walk selection
Value.RandomWalkProb=0.01


Version:

IFS 1.1 (Iterative Forward Search)
Copyright (C) 2006 Tomáš Müller
muller@unitime.org
Lazenska 391, 76314 Zlin, Czech Republic

This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2.1 of the License, or (at your option) any later version.

This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details.

You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA

See Also:

StructuredCSPModel, ConflictStatistics, GeneralValueSelection, GeneralVariableSelection, GeneralTerminationCondition, GeneralSolutionComparator

Constructor Summary

Test()

Method Summary

static void

main(String[] args)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

Test

public Test()

Method Detail

main

public static void main(String[] args)