Originally posted by: llattila

I'm trying to create a QP model but for some reason my code cannot extract the minimization function. Getting the following error code:

CPLEX(default) cannot extract expression: minimize distance.

Here I have listed the pieces of code, which have an impact on the distance expression.

tuple target{
int id;
float x;
int day;
}

{object} objects = ...;
{target} targets = ...;

dvar float locationXhttp://0..maxDaysobjects;
dvar boolean objectAndTargetobjectstargets;

dexpr float distance = sum (o in objects, t in targets) objectAndTarget[o][t] * locationXhttp://t.day[o]^2 + (other conditions come here)

As soon as put the locationXhttp://t.day[o] to the second power, it gives the error message. Other parts of the distance functions seems to be working (such as (t.x)^2 ).

Basically the targets have a day between 1 to maxDays and the locationX is a different value for different days for different objects. I have included day 0 for the location as well as that is the initial location of my objects. objectAndTarget is a boolean value, true if a specific object chooses a specific target.

I'm using IBM ILOG CPLEX Optimization Studio 12.3 on Windows 7.
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: llattila

And of course it didn't properly show the code. Let's try again :)

tuple target{
int id;
float x;
int day;
}
 
{object} objects = ...;
{target} targets = ...;
 
dvar float locationX[0..maxDays][objects];
dvar boolean objectAndTarget[objects][targets];
 
dexpr float distance = sum (o in objects, t in targets) objectAndTarget[o][t] * locationX[t.day][o]^2 + (other conditions come here)

As soon as put the

locationX[t.day][o]

to the second power, it gives the error message. Other parts of the distance functions seems to be working (such as (t.x)^2 ).
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: llattila

And of course it didn't properly show the code. Let's try again :)

tuple target{
int id;
float x;
int day;
}
 
{object} objects = ...;
{target} targets = ...;
 
dvar float locationX[0..maxDays][objects];
dvar boolean objectAndTarget[objects][targets];
 
dexpr float distance = sum (o in objects, t in targets) objectAndTarget[o][t] * locationX[t.day][o]^2 + (other conditions come here)

As soon as put the

locationX[t.day][o]

to the second power, it gives the error message. Other parts of the distance functions seems to be working (such as (t.x)^2 ).
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

The dexpr 'distance' as it currently stands is a cubic expression- since 'locationX' (a float dvar) is being squared and then being multiplied with 'objectAndTarget' (a boolean dvar). CPLEX only supports quadratic expressions; higher order non-linear algebraic forms are not currently supported and thus you get an 'extraction error'. You will need to reformulate the model to a quadratic form in order to solve it.
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: llattila

Thanks for answering. Trying to reformulate the objective function and try again :)
#DecisionOptimization
#OPLusingCPLEXOptimizer