Procedure for reducing the degrees of freedom of an arbitrary system

I'm trying to understand how it's possible to always reduce the degrees of freedom of a system represented by $$n$$ coordinates $$q_{i}$$ for which holds $$f(q_{1},...,q_{n})=0.$$

For example, let's consider the system represented by $$x+\cos y=5$$. How can i show that this system has only 1 degree of freedom?

Your system is currently represented by two coordinates, $$x$$ and $$y$$. Without any constraints, any point $$(x,y)$$ on the plane could be a valid state of your system. But you have a constraint: valid states must obey the equation $$x+\cos y=5$$. This restricts the space of valid states for your system from a two-dimensional plane to a one-dimensional curve, specifically this curve: Since your state space is constrained to be one-dimensional, you now only need one parameter to distinguish between different states. This parameter is your one degree of freedom. For example, you might choose to express the state of the system as solely a function of the single parameter $$w$$, where:

$$y=w$$ $$x=5-\cos w$$

• Thanks. How can i prove formally that it's possible to always reduce the degrees of freedom of a system represented by $n$ coordinates $q_{i}$ for which holds $f(q_{1},...,q_{n})=0.$ ? Nov 25 '19 at 12:17
• @AleQuercia Simply solve the equation $f(q_1,...,q_n)=0$ for $q_1$, so that you have $q_1=g(q_2,...,q_n)$. (If for some reason such a solution is proven not to exist for $q_1$, simply solve for some other coordinate; if this is true for all coordinates, most likely you have an unphysical constraint. See Qmechanic's answer and link for formal requirements on constraints.) Since $q_1$ is completely determined by $q_2,...,q_n$, it's not a free parameter (in other words, simultaneously choosing an arbitrary $q_1$ and choosing $q_2,...,q_n$ would lead to a contradiction). Nov 25 '19 at 14:25
• Let's consider a physical system X described by a set of coordinates, is it possible to say (informally) if the system X is holonomic or no only considering his physical behaviour ? For example a rolling disk on a plane picture Nov 25 '19 at 16:49
• @AleQuercia What do you mean by the system's "physical behavior"? Are you saying that you for some reason have chosen coordinates but don't know the constraints your system is subject to? Because if you know the constraints, you have your answer: if the constraints are all expressible as $f(q_1,...,q_n,t)=0$ (note the absence of any time derivatives of the generalized coordinates), then the system is holonomic. Nov 25 '19 at 17:28
• @AleQuercia The method shown there won't apply to every possible system. Nov 25 '19 at 17:39

Technically, given a $$3N$$-dimensional position manifold, it is implicitly assumed that the zero-locus $$\bigcap_{\ell=1}^m\{f_{\ell}=0\}$$ for the $$m$$ independent holonomic constraints is an embedded submanifold, which necessarily have dimension $$3N-m$$. Hence the constrained system has $$3N-m$$ generalized coordinates/degrees of freedom.