0-1 variable definition

[Kevin Zheng]

When I choose to define a 0-1 variable with binvar, why does the yalmip solution show that the variable is an sdpvar variable?Is there a way to define the variable strictly as 0-1?Thank you very much!

[Johan Löfberg]

binvar is simply syntactic sugar for an sdpvar with an associated binary constraint. Everything is sdpvar objects

[Kevin Zheng]

Thank you Dr.Löfberg.Is there a way to strictly define the 0-1 variable in yalmip?

[Johan Löfberg]

If you define a variable with binvar, or equivalently with sdpvar  + binary constraint, you have defined a binary variable.

[Kevin Zheng]

Thank you very much!I defined a number of 0-1 variables in yalmip with binvar, but the solution results showed that some variables were not strictly 0-1 variables, and the value was 0.999995 instead of 1. May I ask how to solve this problem?

[Johan Löfberg]

solvers have options for declaring integer tolerance, and will terminate when they are satisfied

[Kevin Zheng]

May I ask if we can manually modify the integer tolerance?How should I modify it?

[Johan Löfberg]

you use sdpsettings like always, and which option depends on the solver of course

>> ops = sdpsettings;
>> ops.cplex.mip.tolerances

ans = 

  struct with fields:

           absmipgap: 1.0000e-06

         integrality: 1.0000e-05

         lowercutoff: -1.0000e+75

              mipgap: 1.0000e-04

       objdifference: 0

    relobjdifference: 0

         uppercutoff: 1.0000e+75

[Kevin Zheng]

Thank you very much!