Assume we have the following quadratic program
$$\begin{array}{ll} \text{minimize} & \frac{1}{2} x^T Q x \\ \text{subject to} & Ax <b\end{array}$$
This works when the number of constraint variables is equal to the number of objective variables. But let us now assume that
$$x = (x_0,x_1,x_2)$$
but that our objective does not involve $x_0$, in other words we could say that
$$Q=\begin{pmatrix} 0 &0&0\\ 0 & Q_{11} & Q_{12}\\ 0 & Q_{21} & Q_{22} \end{pmatrix}$$
The matrix is now not positive definite, and most QP solvers don't work with this. So how should the problem be approached if the minimization function does not involve some variable $x_0$, but the constraint equations do involve this variable?
Your current implementation has infinitely many solutions that minimize the cost function (ie any value for $x_0$ that satisfies the constraint). To ensure an unique solution can be computed, either express $x_0$ as a function of $x_1$ and $x_2$ and replace $x_0$ as a variable, or make $Q$ positive definite by setting a very small value for $Q_{0,0}$, like $$\begin{bmatrix}10^{-8} & 0 & 0 \\ 0 & Q_{1,1} & Q_{1,2} \\ 0 & Q_{2,1} & Q_{2,2}\end{bmatrix}$$
That should ensure that there exist an unique solution that minimizes the cost function, while simultaneously weighting $x_0$ as minimal as possible.