Originally posted by: Mark Towers

Hi,

I have a MIQCP problem and I am using the python wrapper, the file below describes how to set the optimality target however only when you load the model from a file.

https://github.com/cswaroop/cplex-samples/blob/master/globalqpex1.py

How do I set the optimality target to global optimal with a python Model

Many thanks,

Mark

You do this the exact same way. You still have an instance of the Cplex class and you use parameters.optimalitytarget.set:

c = cplex.Cplex()
# populate the model here
...
c.parameters.optimalitytarget.set(...)

Originally posted by: Mark Towers

Thanks, currently when I make my model in python I dont use the cplex class but the Model class from docplex.mp.model

Which doesnt have the parameter.optimalitytarget.values variable.

My current code is

from docplex.mp.model import Model

model = Model("test")

# populate model

model.parameters.optimalitytarget.set(3)

solution = model.solve()

However this just throws "Error: model is non-convex" which to me means that the optimality target hasnt worked

Mark

Can you show us the first twenty lines or so of the engine log? This will show us the non-default parameters that are set and we can verify whether the parameter is being set correctly. With docplex, you can do this by setting log_output=True when creating the model (see here).

Originally posted by: Mark Towers

CPXPARAM_Read_DataCheck                          1
CPXPARAM_OptimalityTarget                        3
Warning: Value CPX_OPTIMALITYTARGET_OPTIMALGLOBAL for CPXPARAM_OptimalityTarget is only available for QP and MIQP.
CPLEX Error  5002: 'q1' is not convex.
Presolve time = 0.00 sec. (0.01 ticks)
Root node processing (before b&c):
  Real time             =    0.01 sec. (0.01 ticks)
Parallel b&c, 4 threads:
  Real time             =    0.00 sec. (0.00 ticks)
  Sync time (average)   =    0.00 sec.
  Wait time (average)   =    0.00 sec.
                          ------------
Total (root+branch&cut) =    0.01 sec. (0.01 ticks)
Error: Model is non-convex
CPLEX Error  1217: No solution exists.
CPLEX Error  1217: No solution exists.

So I know that the problem does have a solution because I have programmed the same model in the cp version of cplex.

Looking at the Q1 constraint, it is (constant / variable) + (constant / variable) <= constant

However due to mp not liking division by variable, the equation is reformatted to constant * variable + constant * variable <= constant * variable * variable

Mark

Your model has non-convex quadratic constraints (as the error message indicates). This is not supported by CPLEX. Only non-convex quadratic objectives are supported.

Setting the parameter worked (you get the warning that it does not apply to this model) but it cannot cure the problem of non-convex quadratic constraints.

Originally posted by: Mark Towers

Ohh, thanks for the clarification. 

I had read this article which says that it is possible. 

https://www.ibm.com/support/knowledgecenter/en/SSSA5P_12.9.0/ilog.odms.cplex.help/CPLEX/UsrMan/topics/discr_optim/mip_quadratic/03_introMIQCP.html

However, I notice now that the quadratic term in a constraint involves only multiplication of binary variables which mine does not.

Thanks for your help,

Mark