Problem with GA PopInitRange and lineqr equqlity constraints

Hello,

I have a problem with GA initial population range when using linear equality constraints. For example, if I use three variables x1, x2, and x3, and want to constraint the first two ones (x1 and x2), the constraints seem to somehow influence the distribution of the third variable in the initial population, which was not supposed to be constrained. That is, the third variable is not distributed evenly over the whole initial range, but rather focused around the lower bound. Upper and lower bounds and the initial population range are the same.

Even when I set the Aeq and beq matrices to zeros (which doesn't make any sense, but is always fulfilled, for the purpose of testing), the distribution of the third variable is not evenly distributed over initial range.

Any help would be most appreciated,

Slobodan