# Confused about the definition of holonomic constraints [duplicate]

I'm reading Goldstein's Classical Mechanics and he defines a constraint on particles having radii $\mathbf{r}_i$ to be holonomic if it can be written as $f(\mathbf{r}_1, \mathbf{r}_2, \dots , t) = 0$. He then says that an example of a non-holonomic constraint would be particles constrained to the exterior of a sphere: $r^2 - a^2 \ge 0$.

However, this could be rewritten as $\mathbf{1}_{\left\{\mathbf{r}\mid \:|r|^2 < a^2\right\}}(\mathbf{r}) = 0$, where $\mathbf{1}$ denotes an indicator function. As such it seems to me that the given example is a holonomic constraint. What am I missing?

• I am not used to the indicator function, but can it be exactly written as $f(\mathbf{r}_1, \mathbf{r}_2, \dots , t) = 0$? If not, then the constraint is not holonomic. – Diracology Jul 15 '17 at 0:18
• The indicator function is $f$ in this case. It takes a value of 1 when the coordinates are in the set of points which violate the constraint. (And 0 otherwise) – Davis Yoshida Jul 15 '17 at 0:41
• I think to be holonomic $f$ should also be differentiable, and the indicator function is not. – hyportnex Jul 15 '17 at 1:06
• – Qmechanic Jul 15 '17 at 2:08