Suppose $\dot{x} = f(x)$ is a dynamical system on the state space $X$. My notes define a conservative system as one where there exists a (nontrivial) function $H: X \rightarrow \mathbb{R}$ such that $$\frac{d}{dt}H(x) = 0 = \nabla H(x) \cdot f(x)$$

If we then define $\Sigma_E := H^{-1}(E)$, $E \in \mathbb{R}$ as level sets of $H$, it follows that $\Sigma_E$ is an invariant set due to $H$ being constant along orbits. Immediately after this proof my notes puzzlingly state "the dimension of the problem is reduced by one".

Later on when introducing Hamiltonian systems as a subset of conservative systems the classic example of the pendulum is given:

\begin{align} &\dot{\theta} = p \\ &\dot{p} = -\sin(\theta)\\ \end{align}

It is noted that there is the Hamiltonian $H(p,\theta) = \frac{1}{2}p^2 -\cos(\theta)$ and then shown how any Hamiltonian must be a conserved quantity. Immediately after it is written that "it is useful to notice conserved quantities when they exist as they reduce the dimension of the problem".

Am I missing something obvious? In the case of the pendulum the state space is 2-dimensional, having the Hamiltonian and plotting its level sets gives us the entire phase portrait which is obviously still 2-dimensional. I know a Hamiltonian system has some very nice properties over a generic system, but I do not know one relating to reducing dimensions.

I guess my main question really is "why do we care about conserved quantities?".


1 Answer 1


Once you start at a given point, you are confined to the level set of $H$ passing through that point, which (typically) is an $(n-1)$-dimensional hypersurface in the $n$-dimensional state space $X$. If you introduce an $(n-1)$-dimensional coordinate system on that hypersurface, you can write the ODEs for the motion on the surface in terms of just those $n-1$ variables.

If you know several constants of motion, you can reduce the order further, and if you know sufficiently many, you can (in principle) integrate the system of ODEs exactly.

  • $\begingroup$ Is it possible to give an example of rewriting the equation of the pendulum in 1 dimension? $\endgroup$ Sep 14, 2018 at 17:25
  • 1
    $\begingroup$ @TheoDiamantakis: Sure. You just substitute $p=\dot\theta$ into $H$ to get $\dot\theta^2 - \cos\theta = H_0$ (where $H_0$ is the constant value of $H$ on the level set that you are looking at), which gives $d\theta/dt = \pm \sqrt{H_0 + \cos \theta}$, a separable first-order ODE for $\theta(t)$ which describes the dynamics on that level set. (The sign $\pm$ is chosen according to whether you're in the upper or lower half plane, $p>0$ or $p<0$.) $\endgroup$ Sep 15, 2018 at 9:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.