15. Dynamical systems#

We will now formally define dynamical system, i.e, what does it mean to be a dynamical system, what elements are in a dynamical system, and as well some axioms that a dynamical system has to satisfy.

Definition#

A dynamical system \(D(U, \Sigma, Y, s, r)\) with:

  • \(U\): input space. The input function \(u(\cdot): T \rightarrow U = \mathbb{R}^{n_i}\), with \(n_i\) the dimension of the input.

  • \(\Sigma\): state space. Function \(x(\cdot): T \rightarrow \Sigma: \mathbb{R}^n\).

  • \(Y\): output space. Output function \(y(\cdot): T \rightarrow Y = \mathbb{R}^{n_o}\) with \(n_o\) the dimension of the output.

  • \(s\): state transition function, defines as \(s(\cdot, \cdot, \cdot, \cdot): T \times T \times \Sigma \times U \rightarrow \Sigma\). An example is \(x(t) = s(t, t_0, x_0, u(\cdot))\Big|_{t_0}^t\)

  • \(r\): read-out map (output map), defines as \(r(t, x(t), u(t))\rightarrow y(t)\). With \(T\) be the set of time, it can either be \(\mathbb{R}_+\) or \(\{nT\}, n \in \mathbb{Z}\).

Axioms#

If a system satisfies the following axioms, then it is a dynamical system.

State transition axiom#

If the inputs are the same over a time interval, and we start at the same state, then the state transition function will give us the same result at the end of that time interval.

\[\begin{split} \begin{align*} u(t) &= \tilde{u}(t) \quad \text{over} \quad [t_0, t_1] \\ \text{then} \quad & s(t, t_0, x_0, u) = s(t, t_0, x_0, \tilde{u}) \end{align*} \end{split}\]

What is substantial about this is that we do not care what happen before and after the chosen time interval. We can think of it as at \(t_0\), whatever happened before has already been encapsulated into \(x_0\), regardless of the previous control sequences, and if \(u = \bar{u}\) over \([t_0, t_1]\), then the state transition function will go through the same sequence of states.

Semigroup axiom#

Given \(t^* \in [t_0, t_1]\)

\[ s(t_1, t_0, x_0, u) = s(t_1, t^*, s(t^*, t_0, x_0, u), u) \]