Originally posted by: Zak86

Hi Cplex users,

 

Anyone may help me in this problem please ?

I have to minimize sum( x[i] ) where i in [1 .. N ] and x[ ] a binary array variable ( the output solution variable ) .

The, binary LP, original problem (x[] binary)  gives me a logic solution with an optimal status and value = to 7 .

 

However, when relaxing the problem ( just to put x[ ] continuous in [0 1] using, in JAVA,  IloConversion conversion1 = cplex.conversion(x, IloNumVarType.Float); cplex.add(conversion1);  and without any other modification on constraints ) ,it gives me also optimal solution status, however with  a value = to zero .

 

This latter "Relaxed" Solution value ( zero )  is really absurd !  First, because i have a very clear constraint in which we must have at least one x[i] <> 0 . Second,  (even without considering this constraint) ,  if the Zero solution is the optimal one, the original binary problem must give this solution.

 

Note that when i read the output message, Cplex uses, automatically, the branch and bound/cuts method for the binary LP and uses the dual SIMPLEX for the relaxed LP. 

 

These are the ends of the output log of the 1) ORIGINAL LP and 2) RELAXED LP .

 

1) OUTPUT log for the original LP ( x binary in { 0 , 1 } ):

 

2) OUTPUT log for the RELAXED LP ( x continous in [ 0 , 1 ] ):

Reinitializing dual norms . . .

Dual simplex - Optimal:  Objective =  0.0000000000e+00

THANK YOU VERY MUCH FOR YOUR HELP !!

Zak,

 

#CPLEXOptimizers
#DecisionOptimization

Like I said here, it is perfectly fine for the relaxed problem to have a better objective than the original problem. The fact that the constraint x[i]>0 does not seem to have an effect may be due to bad numerics, see my other post for more details.

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Zak86

Yes Daniel, Thank you;.

I see your other answer. However, please how do you see this coefficient of the order of .....e+308?

is it in the .sav or .prm file ? What command de you make exactly to read the file and see this value ?

Indeed, in the simple console of Eclipse , i see only a 0-value.

Thank you,

Zak

#CPLEXOptimizers
#DecisionOptimization

I found the ...e+308 using the interactive optimizer:

CPLEX> read model.sav
CPLEX> disp prob stats

This displays minimal and maximal coefficients etc.

#CPLEXOptimizers
#DecisionOptimization