betaFeedback

WorksMathematics

On the multiplicity of crossing limit cycles of double diagonalizable piecewise linear vector fields

Abstract

We consider planar piecewise linear vector fields with two zones separated by a straight line and having two diagonalizable components, which we call the double diagonalizable configuration. In this configuration, the Poincaré half-maps correspond to separatrices of saddle points in the foliations of symmetric differential forms ωr(x)−ωr(y)ω_r(x)-ω_r(y), where ωrω_r is a rational 11-form with exactly three simple poles. Limit cycles correspond to non-diagonal intersections of two such separatrices, and the multiplicity of a limit cycle is precisely the order of contact at the intersection point. We prove that two distinct separatrices which are not nested segments of the same straight line have a contact of order at most 44 at a diagonal point and at most 22 at a non-diagonal point. As a consequence, we obtain that the multiplicity of limit cycles in the double diagonalizable configuration is at most two, this bound being sharp.

Introduction and statement of the main results

The qualitative study of differential systems has been one of the main interests of the mathematical community for more than a century, which is a consequence of Hilbert’s 16th Problem whose object is to determine, if it exists, the upper bound on the number of limit cycles of planar polynomial vector fields. Stemming from its original formulation, several different versions have been considered. One of these versions consists of studying, instead of polynomial vector fields, nonsmooth vector fields.

The investigation of limit cycles in nonsmooth planar vector fields itself has a history that dates back to the 1930s with the work of Andronov et al. [1]. Perhaps one of the most studied subclasses of nonsmooth vector fields is that of the planar piecewise linear vector fields, with two zones separated by a straight line. In other words, up to an affine change of variables, vector fields of the form

X=1y>0X++1y<0X−,(1)X = \mathbf{1}_{y>0}X_{+} + \mathbf{1}_{y<0}X_{-}, \tag*{(1)}

where X±X_{\pm} are linear vector fields, and 1S\mathbf{1}_{S} is the characteristic function of a set SS. The dynamics on the switching set Σ={y=0}\Sigma= \{y = 0\} are defined by the Filippov convention [11]. In this setting, a limit cycle (also referred to as crossing limit cycle) is a crossing non-trivial closed orbit which is isolated from other closed orbits.

Several authors have studied system (1) in the last decades and significant progress has been made concerning the upper and lower bounds on the number of limit cycles. Freire et al. [12] proved that (1) has at most one limit cycle in the case that the Filippov vector field is continuous, which was first conjectured by Lum and Chua [22]. In the discontinuous case, several upper bounds have been obtained for particular cases of system (1), for instance: in [18], it was proved that the maximum number of limit cycles is two if one of the components has a focus at Σ\Sigma, and one if either component has a center or a weak saddle (a saddle with hyperbolicity ratio 1). The upper bound is also one for refracting systems [17] and for sewing systems, that is, systems without sliding region [23, 7].

Recently, Carmona, Fernández-Sánchez and Novaes [6] established that the uniform upper bound on the number of limit cycles of (1) is L∗≤8L^{*} \le8. Their breakthrough was based on the integral characterization of Poincaré half-maps [5] and an extension of Khovanskii’s theory, and was further summarized into a methodology to determine upper bounds on limit cycles of piecewise linear vector fields in [8]. However, the best known lower bound on the maximum number of limit cycles of (1) is three (see, for instance, [15, 19, 4, 14]). The known examples of piecewise linear vector fields with three limit cycles involve at least one focus, whereas for configurations without foci the best lower bound is two [9, 20].

Let us now formulate our main result. We say that a piecewise linear vector field XX of the form (1) is double diagonalizable (or of D-D type, for short) if both its components X+X_{+} and X−X_{-} have a diagonalizable linear part with non-zero eigenvalues. Equivalently, each component X±X_{\pm} of XX has a saddle or a proper node singularity, which can be either visible or invisible. The particular case of D-D piecewise linear vector fields where the components have facing saddles has been considered in [2, 21, 25]. For two nodes, Chen and Liu [9] proved, by a computer-assisted analysis, that there are at most two limit cycles counting multiplicities. Other types of D-D vector fields were considered in [20].

The main result of the paper is the following Theorem:

Theorem A. All limit cycles of piecewise linear vector fields of D-D type have multiplicity at most two. Moreover, this bound is sharp.

To our knowledge, Theorem A is new for the saddle–saddle and node–saddle configurations whose fold points do not coincide. Indeed, in the node–node case it follows from [9], and when the two fold points coincide there is no sliding region and every limit cycle is hyperbolic by [7]. Our proof is analytic and treats all D-D configurations at once. Since the displacement function depends analytically on the parameters, Theorem A also implies that at most two limit cycles bifurcate from a crossing limit cycle of a D-D vector field under small perturbations within the class (1).

Due to the diagonalizability hypothesis, for each component X±X_{\pm} a Poincaré half-map x↦σ±(x)x \mapsto\sigma_{\pm}(x) is defined on some interval contained in Σ\Sigma, and X±X_{\pm} has a first integral of Darboux type. Through this first integral, one can express the graphs of these maps as separatrices associated with suitably defined differential forms. More precisely, parametrizing the discontinuity line Σ\Sigma by the xx-coordinate and denoting by Σin±\Sigma_{\mathrm{in}}^{\pm} and Σout±\Sigma_{\mathrm{out}}^{\pm} the domain and image of the Poincaré half-map of X±X_{\pm} respectively, the curves {(x,y)∣x∈Σin±,y=σ±(x)}\{(x,y) \mid x \in\Sigma_{\mathrm{in}}^{\pm}, y = \sigma_{\pm}(x)\} can be interpreted as separatrices of a saddle point (see Figure 3) within the foliation of the symmetric differential form Ωr=ωr(x)−ωr(y)\Omega_r = \omega_r(x) - \omega_r(y). Here,

The foliation of $\Omega_r$ for different configurations of $r_0$, $r_1$, $r_2$ in $\mathbb{M}$

Figure 3. The foliation of Ωr\Omega_r for different configurations of r0r_0, r1r_1, r2r_2 in M\mathbb{M}. The manifold M\mathbb{M} is diffeomorphic to a 2-torus, which is topologically obtained by gluing the opposite sides of these squares. The separatrices of the saddle ss are highlighted as thick red lines. Colors available online.

ωr(z)=z−r0(z−r1)(z−r2) dz\omega_r(z) = \frac{z-r_0}{(z-r_1)(z-r_2)}\,\mathrm{d}z

and r=(r0,r1,r2)r = (r_0,r_1,r_2) is a triple of distinct real numbers. Using this correspondence, each limit cycle of multiplicity nn for a discontinuous vector field XX of the form (1) corresponds precisely to a point of contact of order nn between the separatrices of the corresponding symmetric forms.

The present work is structured as follows: In Section 2, we establish the above-mentioned connection between the Poincaré half-maps of piecewise linear vector fields of D-D type and separatrices associated to symmetric differential forms Ωr\Omega_r as defined above. From this point, we state the intermediate result (Theorem B) that if the order of contact of two such separatrices is either 5 at a diagonal point or 3 at a non-diagonal point, then these separatrices either coincide or are nested line segments. In Section 3, we prove Theorem B by studying the contact at diagonal points in subsection 3.1, whereas subsection 3.2 is dedicated to the investigation of 3-contacts at non-diagonal points which is translated via a projective change of coordinates to the determination of the fibers of an analytic map Θ:D⊂R2→R2\Theta: \mathbf{D} \subset\mathbb{R}^{2} \to\mathbb{R}^{2}, where D\mathbf{D} is an appropriate open subset of R2\mathbb{R}^{2}. Some illustrative examples are presented in Section 4. Section 5 proves all technical results needed to describe the fibers of the map Θ\Theta. As we shall see, this description is fully determined by the sign of the Jacobian determinant det⁡DΘ\det D\Theta. To determine this sign, we express all relevant functions as parameter-dependent weighted integrals over the real line. This allows us to reduce the evaluation of sign⁡(det⁡DΘ)\operatorname{sign}(\det D\Theta) to comparing the growth rates of two moment ratios associated with these integral representations.

Half-return maps via Darboux first integrals

We start by discussing the general concept of half-return maps for smooth vector fields defined on domains with boundary. Let XX be a smooth vector field defined on an open subset U⊂RnU \subset\mathbb{R}^{n}. Given a closed subset V⊂U\mathcal{V} \subset U with boundary Σ=∂V\Sigma= \partial\mathcal{V}, we define the domain of the half-return map of X∣VX|_{\mathcal{V}} at Σ\Sigma as the set

Σin={p∈Σ:∃ tp>0:φX(tp;p)∈Σ and φX((0,tp);p)⊂V˚},\Sigma_{\mathrm{in}} = \left\{p \in\Sigma: \exists\ t_p > 0 : \varphi_X(t_p;p) \in\Sigma\text{ and } \varphi_X((0,t_p);p) \subset\mathring{\mathcal{V}}\right\},

where φX(⋅;p)\varphi_X(\cdot;p) denotes the flow of XX through the point pp with φX(0;p)=p\varphi_X(0;p) = p and V˚\mathring{\mathcal{V}} denotes the interior of the set V\mathcal{V}. In this case, the map σ:Σin→Σ\sigma: \Sigma_{\mathrm{in}} \to\Sigma given by σ(p)=φX(tp,p)\sigma(p) = \varphi_X(t_p,p) is called the half-return map (or Poincaré half-map) of X∣VX|_{\mathcal{V}} at Σ\Sigma. We denote by Σout=σ(Σin)\Sigma_{\mathrm{out}} = \sigma(\Sigma_{\mathrm{in}}) its image. Whenever Σin\Sigma_{\mathrm{in}} is non-empty, we say that a half-return map is well-defined for X∣VX|_{\mathcal{V}}.

Half-return maps for linear vector fields in half-planes

We want to completely characterize the domains Σin\Sigma_{\mathrm{in}} and images Σout\Sigma_{\mathrm{out}} in the case that XX is a diagonalizable linear vector field and V\mathcal{V} is a closed half-plane.

We say that a planar linear vector field XX is diagonalizable if its Jordan canonical form is diagonal. In other words, there exists an affine coordinate change that transforms XX into X=λ1u ∂u+λ2v ∂vX = \lambda_1 u\,\partial_u + \lambda_2 v\,\partial_v, with λ1λ2≠0\lambda_1\lambda_2 \ne0. Up to an additional time rescaling t→t/λ1t \to t/\lambda_1, we see that any diagonalizable vector field is orbitally affinely equivalent to

XJ=u∂u−λv∂v,(2)X_J = u\partial_u - \lambda v\partial_v, \tag*{(2)}

and the number λ∈R∗\lambda\in\mathbb{R}^{*} is called the hyperbolicity ratio of XX.

In this setting, the half-return maps, also called Poincaré half-maps, have been extensively studied in the literature, mainly connected to the dynamics of piecewise smooth vector fields (see [5, 6, 12, 13]). A recent result, due to Carmona and Fernández-Sánchez [5], characterizes these maps as the Cauchy Principal Value of the integral of a rational function. Their construction is based on a Liénard canonical form. In Subsection 2.3, we will take a different approach, constructing the half-return map from the Darboux first integrals of diagonalizable linear vector fields. In order to do so, we need some preliminary results.

Let us initially consider the following simplified situation. Let XX be given in the orbital Jordan form (2) for some λ∈R∖[−1,0]\lambda\in\mathbb{R} \setminus[-1,0] and Σ\Sigma be given by the affine line f(u,v)=0f(u,v)=0, where f(u,v)=u+v−1f(u,v)=u+v-1. Then, consider the lines

ℓ1={u=0},ℓ2={v=0},Xf(u,v)≡0,\ell_1=\{u=0\},\qquad\ell_2=\{v=0\},\qquad Xf(u,v)\equiv0,

where XfXf denotes the Lie derivative of ff with respect to XX. We consider also the intersection points p=(λλ+1,1λ+1)=Σ∩{Xf≡0}p=\left(\frac{\lambda}{\lambda+1},\frac{1}{\lambda+1}\right)=\Sigma\cap\{Xf\equiv0\}, p1=(0,1)=Σ∩ℓ1p_1=(0,1)=\Sigma\cap\ell_1 and p2=(1,0)=Σ∩ℓ2p_2=(1,0)=\Sigma\cap\ell_2.

The following result characterizes the domain and image of the half-return maps of XX restricted to the half-planes Σ+={f(u,v)≥0}\Sigma^+=\{f(u,v)\geq0\} and Σ−={f(u,v)≤0}\Sigma^-=\{f(u,v)\leq0\}. In the statement, for any two points q0,q1∈Σq_0,q_1\in\Sigma, (q0,q1)(q_0,q_1) denotes the open segment of Σ\Sigma between q0q_0 and q1q_1. The notation (q0,∞)(q_0,\infty) indicates the infinite ray starting from q0q_0.

Lemma 2.1. In the above setting, the following holds:

  1. If λ>0\lambda>0 (saddle case):

    a) If V=Σ+\mathcal{V}=\Sigma^+, then Σin=∅\Sigma_{\mathrm{in}}=\varnothing;

    b) If V=Σ−\mathcal{V}=\Sigma^-, then Σin=(p1,p)\Sigma_{\mathrm{in}}=(p_1,p) and Σout=(p,p2)\Sigma_{\mathrm{out}}=(p,p_2).

  1. If λ<−1\lambda<-1 (node case):

    a) If V=Σ−\mathcal{V}=\Sigma^-, then Σin=∅\Sigma_{\mathrm{in}}=\varnothing;

    b) If V=Σ+\mathcal{V}=\Sigma^+, then Σin=(p2,p)\Sigma_{\mathrm{in}}=(p_2,p) and Σout=(p,∞)\Sigma_{\mathrm{out}}=(p,\infty).

The proof of this lemma is immediate from a standard analysis of linear vector fields and an application of barrier functions. Figure 1 illustrates the simplified situation. We now take on the more general setting:

The domain and image of the half-return map in the setting of Lemma 2.1

Figure 1. The domain Σin\Sigma_{\mathrm{in}} and image Σout\Sigma_{\mathrm{out}} of the half-return map in the setting of Lemma 2.1

  1. XX is a diagonalizable linear vector field;

  1. V={f≥0}\mathcal{V} = \{f \ge0\} is an affine half-plane with boundary Σ={f=0}\Sigma= \{f = 0\}, where ff is a non-zero affine function.

We say that p∈Σp \in\Sigma is an invisible fold of X∣VX|_{\mathcal{V}} if the lines Σ\Sigma and Xf≡0Xf \equiv0 intersect only at the point pp and additionally X2f(p)<0X^2f(p) < 0 (Figure 2).

An invisible fold

Figure 2. An invisible fold.

References

  1. [1]A. A. Andronov, A. A. Vitt, and S. E. Khaïkin, Theory of oscillators, Dover Publications, Inc., New York, 1987. Translated from the Russian by F. Immirzi, Reprint of the 1966 translation.
  2. [2]J. C. Artés, J. Llibre, J. C. Medrado, and M. A. Teixeira, Piecewise linear differential systems with two real saddles, Math. Comput. Simulation, 95 (2014), pp. 13–22.DOI
  3. [3]M. Bezziou, Z. Dahmani, and A. Khameli, On some double weighted fractional integral inequalities, Sarajevo J. Math., 15(28) (2019), pp. 23–36.
  4. [4]C. Buzzi, C. Pessoa, and J. Torregrosa, Piecewise linear perturbations of a linear center, Discrete Contin. Dyn. Syst., 33 (2013), pp. 3915–3936.DOI
  5. [5]V. Carmona and F. Fernández-Sánchez, Integral characterization for Poincaré half-maps in planar linear systems, J. Differential Equations, 305 (2021), pp. 319–346.DOI
  6. [6]V. Carmona, F. Fernández-Sánchez, and D. D. Novaes, Uniform upper bound for the number of limit cycles of planar piecewise linear differential systems with two zones separated by a straight line, Appl. Math. Lett., 137 (2023), pp. Paper No. 108501, 8.DOI
  7. [7]——, Uniqueness and stability of limit cycles in planar piecewise linear differential systems without sliding region, Commun. Nonlinear Sci. Numer. Simul., 123 (2023), pp. Paper No. 107257, 18.DOI
  8. [8]V. Carmona, F. Fernández-Sánchez, and D. D. Novaes, Bounding the number of limit cycles in piecewise linear differential systems: methodology and worked examples, 2025. https://arxiv.org/abs/2508.02321.
  9. [9]L. Chen and C. Liu, The discontinuous planar piecewise linear system with two nodes has at most two limit cycles, 2025. https://arxiv.org/abs/2506.07716.
  10. [10]A. Cima, F. Mañosas, and J. Villadelprat, Isochronicity for several classes of Hamiltonian systems, J. Differential Equations, 157 (1999), pp. 373–413.DOI
  11. [11]A. F. Filippov, Differential equations with discontinuous righthand sides, vol. 18 of Mathematics and its Applications (Soviet Series), Kluwer Academic Publishers Group, Dordrecht, 1988. Translated from the Russian.
  12. [12]E. Freire, E. Ponce, F. Rodrigo, and F. Torres, Bifurcation sets of continuous piecewise linear systems with two zones, Internat. J. Bifur. Chaos Appl. Sci. Engrg., 8 (1998), pp. 2073–2097.DOI
  13. [13]E. Freire, E. Ponce, and F. Torres, Canonical discontinuous planar piecewise linear systems, SIAM J. Appl. Dyn. Syst., 11 (2012), pp. 181–211.DOI
  14. [14]——, A general mechanism to generate three limit cycles in planar Filippov systems with two zones, Nonlinear Dynam., 78 (2014), pp. 251–263.DOI
  15. [15]S.-M. Huan and X.-S. Yang, On the number of limit cycles in general planar piecewise linear systems, Discrete Contin. Dyn. Syst., 32 (2012), pp. 2147–2164.DOI
  16. [16]P. Kumar, Inequalities involving moments of a continuous random variable defined over a finite interval, Comput. Math. Appl., 48 (2004), pp. 257–273.DOI
  17. [17]S. Li, C. Liu, and J. Llibre, The planar discontinuous piecewise linear refracting systems have at most one limit cycle, Nonlinear Anal. Hybrid Syst., 41 (2021), pp. Paper No. 101045, 14.DOI
  18. [18]J. Llibre, D. D. Novaes, and M. A. Teixeira, Maximum number of limit cycles for certain piecewise linear dynamical systems, Nonlinear Dynam., 82 (2015), pp. 1159–1175.DOI
  19. [19]J. Llibre and E. Ponce, Three nested limit cycles in discontinuous piecewise linear differential systems with two zones, Dyn. Contin. Discrete Impuls. Syst. Ser. B Appl. Algorithms, 19 (2012), pp. 325–335.
  20. [20]J. Llibre, M. A. Teixeira, and J. Torregrosa, Lower bounds for the maximum number of limit cycles of discontinuous piecewise linear differential systems with a straight line of separation, Internat. J. Bifur. Chaos Appl. Sci. Engrg., 23 (2013), pp. 1350066, 10.DOI
  21. [21]J. Llibre and C. Valls, Limit cycles of planar piecewise differential systems with linear Hamiltonian saddles, Symmetry, 13 (2021), p. Paper No. 1128.DOI
  22. [22]R. Lum and L. O. Chua, Global properties of continuous piecewise linear vector fields. Part I: Simplest case in R², International Journal of Circuit Theory and Applications, 19 (1991), pp. 251–307.
  23. [23]J. C. Medrado and J. Torregrosa, Uniqueness of limit cycles for sewing planar piecewise linear systems, J. Math. Anal. Appl., 431 (2015), pp. 529–544.DOI
  24. [24]D. S. Mitrinović, J. E. Pečarić, and A. M. Fink, Classical and new inequalities in analysis, vol. 61 of Mathematics and its Applications (East European Series), Kluwer Academic Publishers Group, Dordrecht, 1993.
  25. [25]Y. SHAO, S. LI, AND K. WU, Global phase portraits of planar piecewise linear refracting systems of saddle-saddle type, Nonlinear Anal. Real World Appl., 62 (2021), pp. Paper No. 103381, 15.DOI
  26. [26]O. SHISHA AND C. B. MEHR, On involutions, J. Res. Nat. Bur. Standards Sect. B, 71B (1967), pp. 19–20.DOI
  27. [27]G. ZAMPIERI, Involutions of real intervals, Ann. Polon. Math., 112 (2014), pp. 25–36.DOI

Paper details

vlnc:W_7gfqub7n
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Claude Opus 5 (bare), Opus 5.5

View statement
Looking for related papers…
Contents