logo

Smooth Systems in Each Segment of Dynamics 📂Dynamics

Smooth Systems in Each Segment of Dynamics

Definition

Piecewise Smooth System

Let’s consider a dynamical system whose state space is Rn\mathbb{R}^{n} with variables xRnx \in \mathbb{R}^{n} and parameters μRp\mu \in \mathbb{R}^{p} represented as follows:
x˙=f(x;μ)xRn,μRp \dot{x} = f \left( x ; \mu \right) \qquad x \in \mathbb{R}^{n} , \mu \in \mathbb{R}^{p}
If there exist finitely many open sets SkRnS_{k} \subset \mathbb{R}^{n} and ff that satisfy the following for smooth functions Fk:SkRnF_{k} : S_{k} \to \mathbb{R}^{n}, then this system is called a Piecewise Smooth System:
f(x,μ)=Fk(x,μ)(x,μ)SkRn f \left( x , \mu \right) = F_{k} \left( x , \mu \right) \qquad \forall (x,\mu) \in S_{k} \subset \mathbb{R}^{n}
The region Σij\Sigma_{ij} between SiS_{i} and SjS_{j} is assumed to be a (n1)(n-1)-dimensional differential manifold, and FF may implicitly represent a multivalued map, not strictly a function in the traditional sense.

Degree of smoothness

Considering sufficiently smooth FiF_{i} and FjF_{j},
Fi(k)Fj(k)=dkdxkFidkdxkFj F_{i}^{(k)} - F_{j}^{(k)} = {{ d^{k} } \over { d x^{k} }} F_{i} - {{ d^{k} } \over { d x^{k} }} F_{j}
there exists an integer d0d \ge 0 such that for all kk that satisfy 0k<d0 \le k < d, if (Fi(k)Fj(k))\left( F_{i}^{(k)} - F_{j}^{(k)} \right) is continuous in Σij\Sigma_{ij} and (Fi(d)Fj(d))\left( F_{i}^{(d)} - F_{j}^{(d)} \right) is not, then dd is referred to as the Degree of smoothness in Σij\Sigma_{ij}.

Explanation 1

Many theories in dynamics have developed alongside the history of ordinary differential equations, primarily focusing on smooth functions. However, practical applications (especially in control) tend to break these conventions. PWS offers a transitional view before exploring mature non-smooth systems. Though its definition might be somewhat vague mathematically, its expression is clear and communicates well, making it acceptable for use. A famous example is the DC-DC buck converter represented as follows:
V˙=1RCV+ICI˙=VL+{0,if VVr(t)E/L,if V<Vr(t)Vr(t)=γ+η(tmod  T) \begin{align*} \dot{V} =& - {{ 1 } \over { RC }} V + {{ I } \over { C }} \\ \dot{I} =& - {{ V } \over { L }} + \begin{cases} 0 & , \text{if } V \ge V_{r} (t) \\ E / L & , \text{if } V < V_{r} (t) \end{cases} \\ V_{r} (t) =& \gamma + \eta \left( t \mod T \right) \end{align*}
This system is evidently non-smooth, but whether or not it conforms to the definition of PWS is not of significant importance.

Degree of smoothness

Though typically translated as “Degree,” in this context, it closely matches the concept of Degree of smoothness. Simply put, it explains how many times x˙=f(x)\dot{x} = f(x) needs to be differentiated before discontinuities arise. The reference literature states that in
x˙=signx \dot{x} = - \sign x
Σ12={x=0}\Sigma_{12} = \left\{ x = 0 \right\} is d=0d = 0 and in
x˙=x \dot{x} = - \left| x \right|
Σ12={x=0}\Sigma_{12} = \left\{ x = 0 \right\} is d=1d = 1. Similarly, these assertions are not derived from meticulous consideration of the definition of the sign function sign\sign or the differentiability of absolute value. It’s more intuitive to understand and move on.


  1. Di Bernardo, M., Budd, C. J., Champneys, A. R., Kowalczyk, P., Nordmark, A. B., Tost, G. O., & Piiroinen, P. T. (2008). Bifurcations in nonsmooth dynamical systems. SIAM review, 50(4), 629-701. https://doi.org/10.1137/050625060: p632~635. ↩︎