Originally posted by: tavs

GLPK allows to set a row's bounds by use of function

void glp_set_row_bnds(glp_prob *lp, int i, int type, double lb, double ub);

if type is GLP_DB, i.e. a "double-bound", the routine sets the i-th row's lower and upper bounds to lb and ub respectively

In CPLEX's C API I found the functions CPXchgrhs and CPXchgrngval, but I couldn't figure out how to use them to achieve the same functionality. What should I do ?

TIA
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Quoted from the documentation of CPXcopylpwnames (I agree, this is not the first place I would look for that):

If rngval>0, then row activity is in [rhs,rhs+rngval]
and if rngval<0, then row activity is in [rhs+rngval,rhs]

Does that answer your question?
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

At least in CPLEX 8 or 9 there was a bug in presolve with handling negative range values. I don't know if this bug is still there, bug you should probably prefer non-negative range values.

So, given a linear constraint

lb <= a*x <= ub

you should set the regular right hand side to lb, and the range value to ub-lb.

Tobias
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: tavs

Thanks, it worked!
#CPLEXOptimizers
#DecisionOptimization