withNoiseMILPEval_Metric_1.run

# THIS CODE IS FOR RUNNING MILP ON DATA OBTAINED AFTER APPLYING GREEDY APPROACH
# IT USES THETA FOR EVALUATION, TOLERATES VERY SMALL GAP IN THETA (E.G., 1E-5)
# TO MEASURE SUCCESS OF THIS APPROACH.


#RESET THE AMPL ENVIROMENT
reset;


#LOAD THE MODEL
#model ReducedOutliersMILPmain.mod;
model NoisyMILPModel.mod; # if data has some noise....

#CHANGE THE SOLVER (optional)
option solver gurobi;
option gurobi_options 'timelim 60'; #'bestobjstop 20';


param flag;
#SOLVE
for {ncase in 3..3} 
{   
    for {pcase in 7..7}
    { 
        for {case in 1..5}
        {     
              display ncase;
              display pcase;
              display case;
              reset data;
              option randseed 0;
              data ("./inputMILP/Noise/reducedData_n_" & ncase & "_p_" & pcase & "_" & case & ".dat");
              display p;
      
              solve;
              display theta, T, X_true, X, outliervec, Z, f;
              display _total_solve_elapsed_time;
              #display W;
              #display case, p > ("./outputMILP/reducedData_" & case & ".out");
              #display theta, T, X_true, X, outliervec, Z, f > ("./outputMILP/reducedData_" & case & ".out");
              let flag:=1;
              
              #We used .00001 for sigma=0, .001 for sigmal=1e-9, .01 for sigma=1e-3
              for { s in 1..r}
              { 
                    if(abs(T[s]-theta[s])>0.01) then { 
                        let flag:=0;
                        break;
                    }
              }
              if flag = 1 then {
                  print "Suceeded"; 
              }
              else print "Failed";
              print flag >> ("./outputMILP/EvalMetric_1/reducedData_n_" & ncase & "_p_" & pcase & ".out");
         }
         
      }              
}