Constraint


A constraint is a limitation put on a system that prevents it from moving in certain locations in space. Applying a constraint to a system reduces the number of unknowns of that systems, leading to a simplified description of the same phenomenon. A bound system can only move in a subset of the space it would have otherwise had access to. In mathematical terms, each constraint shrinks the configuration space of the system by reducing the degrees of freedom: one constraint removes one degree of freedom.

Most mechanical bindings can be described as forces with Newton's second law:

ma=F+Φm\mathbf{a}=\mathbf{F}+\Phi

mm is mass, a\mathbf{a} is acceleration, F\mathbf{F} is an active force, which is known, and Φ\Phi is a constraint reaction, which is not known.

Properties

A constraint is said to be ideal if the configuration space is "smooth", that is, if the constraint reaction in any configuration PP is always orthogonal to QQ. Mathematically, this is like saying that the Scalar product of Φ\Phi and any tangent element δr\delta \mathbf{r} is always zero:

Φδr=0PRn and δrTPQΦrqi where i=1,n\Phi\cdot \delta\mathbf{r}=0\quad\forall P\in \mathbb{R}^{n}\text{ and }\forall \delta \mathbf{r}\in T_{P}Q\quad\Leftrightarrow\quad \Phi\cdot \frac{ \partial \mathbf{r} }{ \partial q_{i} } \text{ where }i=1\ldots,n

where TPQT_{P}Q is the tangent space of QQ in the point PP. If the constraint applies to NN point masses, only the total reaction needs to be orthogonal:

i=1NΦδri=0δriTPQ\sum_{i=1}^{N} \Phi\cdot \delta\mathbf{r}_{i}=0\quad \forall \delta \mathbf{r}_{i}\in T_{P}Q

A constraint is said to be scleronomous (or mobile) if it changes in time, otherwise it is rheonomous (or fixed). In other words, the equation of a scleronomous constraint is explicitly dependent on time, whereas a rheonomous one isn't. An example of scleronomous constraint would be an object attached to an elevator: the object is always constrained in the same way but the elevator (the constraint) moves, pulling the object with it.

A constraint is said to be holonomic if it can be expressed as a function of generalized coordinates and potentially time that is set to zero: f(q1,,qN,t)=0f(q_{1},\ldots,q_{N},t)=0. Common examples are forcing a particle to move on a Curve or Surface, in which case the equation of the curve or surface is the constraint.