Running ParSA

From UConn PAN
Jump to navigation Jump to search

This page will serve as a progress report for the weekly work done on running the ParSA algorithm. Plots of cost function versus step number and the parameters used will be included for each completed run.


Week of 21 January 2008

The goal of the first run was to ensure that the MIR scheduler and solver would be able to successfully complete an entire run.

The parameters used for this run are as follows:

 {
         SA_Solver       SA_MIRSolver
         {
                 SA_Solver
                 {
                         datafilename            data.ser
                         outputfilename          parsaoutput
                         solutionfilename        *
                         overwritesolution       better
                         writefunction           path
                         verbose                 on
                         verboselevel            10
                         StartSolution           Init
                 }
                 SA_Scheduler SA_MIRScheduler
                 {
                         SA_Schedulerdouglas gregory
                         {
                                 picturedirectory        .douglas gregory
                                 OptType                 MIN
                                 thresholdvalue          0
                                 initialtemperature      -1.0
                                 initaccratio            0.9
                                 verboselevel            10
                                 timelimit               100000
                         }
                         Alpha                  0.5
                         Betta                  2
                         endtemperature         0
                         Temperature_Reset      0
                 }
                 Betta_Runtime           1.1
                 Ebest                   0
                 Epsilon                 0.01
                 Maximum_RunLength       RunFactor*GetLocalN()
                 Minimum_RunLength       GetLocalN()
                 Samples                 10
                 RunFactor               5
                 algorithm               MIR
         }
 }
A plot of BestCostFunktion versus Step Number