Secondary Hopf Bifurcation: Theory and Dynamics

• Secondary Hopf bifurcation is the transition where a periodic orbit loses hyperbolicity due to a conjugate pair of Floquet multipliers crossing the unit circle, thus generating an invariant torus.
• The phenomenon is characterized by strict local criteria such as nonresonance up to order 4 and the sign of the first Lyapunov coefficient, distinguishing between supercritical and subcritical cases.
• Its study extends to nonautonomous systems, codimension-two bifurcations, and generalized torus structures, providing practical insights for diverse dynamical models.

Searching arXiv for papers on secondary Hopf bifurcation and closely related terms. Secondary Hopf bifurcation is the bifurcation in which a pre-existing periodic orbit loses hyperbolicity because a conjugate pair of Floquet multipliers crosses the unit circle, producing an invariant closed curve in a Poincaré section and, by saturation under the flow, a two-dimensional invariant torus in phase space (Pereira et al., 14 Jul 2025). In this standard autonomous sense, it is the continuous-time counterpart of the Neimark–Sacker bifurcation for maps (Pereira et al., 14 Jul 2025). The term is also used more loosely in adjacent settings, including nonautonomous skew products, delay equations near Hopf–Hopf points, and fluid or synchronization problems where a nontrivial oscillatory state loses stability to more complicated motion. Across these contexts, a persistent distinction is that between primary Hopf bifurcation, meaning equilibrium $\to$ periodic orbit, and secondary Hopf bifurcation, meaning periodic orbit $\to$ torus or quasiperiodic modulation (Das et al., 2013, Pereira et al., 14 Jul 2025).

1. Classical meaning and local structure

In the setting of a one-parameter family of autonomous ODEs

$\dot x = F(x,\mu),$

with a family of periodic orbits $\gamma_\mu$, a local Poincaré section $\Sigma$ produces a family of maps

$p \mapsto \Pi(p,\mu).$

A secondary Hopf bifurcation is the situation in which a fixed point $p^*(\mu)$ of $\Pi(\cdot,\mu)$, corresponding to the periodic orbit $\gamma_\mu$, loses hyperbolicity because a conjugate pair of Floquet multipliers crosses the unit circle (Pereira et al., 14 Jul 2025). The paper "Normal Hyperbolicity in Secondary Hopf Bifurcations" states the bifurcation hypotheses as follows: there is a curve $p^*(\mu)$ of fixed points,

$\to$0

and the eigenvalues $\to$1 of

$\to$2

satisfy the conditions

$\to$3

$\to$4

$\to$5

$\to$6

These conditions specify a single critical complex pair on the unit circle, nonresonance up to order $\to$7, and transversal crossing (Pereira et al., 14 Jul 2025).

The first Lyapunov coefficient $\to$8 of the Poincaré map distinguishes the supercritical and subcritical cases: $\to$9 gives the supercritical case, while $\dot x = F(x,\mu),$0 gives the subcritical case (Pereira et al., 14 Jul 2025). The torus exists when

$\dot x = F(x,\mu),$1

This is the standard local criterion quoted in that paper (Pereira et al., 14 Jul 2025).

The invariant object created at bifurcation is a closed invariant curve $\dot x = F(x,\mu),$2. Saturating that curve by the flow yields a two-dimensional invariant torus $\dot x = F(x,\mu),$3 for the full system (Pereira et al., 14 Jul 2025). One angular direction comes from the phase along the original periodic orbit, and the second comes from the invariant closed curve in the section (Pereira et al., 14 Jul 2025). This is the canonical geometric meaning of secondary Hopf bifurcation in continuous time.

2. Difference from primary Hopf bifurcation

The distinction from primary Hopf bifurcation is explicit in several of the cited works. The paper "Super-Critical and Sub-Critical Hopf bifurcations in two and three dimensions" is fundamentally about Hopf bifurcation of a fixed point: the loss of stability of an equilibrium when a complex-conjugate pair of eigenvalues crosses the imaginary axis, producing or destroying a limit cycle (Das et al., 2013). It states plainly that it does not analyze a secondary Hopf bifurcation in the usual sense of a Hopf/Neimark–Sacker bifurcation of a periodic orbit leading to a torus or quasiperiodicity (Das et al., 2013).

Its normal-form prototype is

$\dot x = F(x,\mu),$4

with Hopf point at $\dot x = F(x,\mu),$5, supercritical if $\dot x = F(x,\mu),$6, subcritical if $\dot x = F(x,\mu),$7 (Das et al., 2013). At this level, the equilibrium changes stability and a limit cycle is born or dies; no torus is involved (Das et al., 2013). The same restriction is emphasized in the Kuramoto–Daido paper "A Hopf bifurcation in the Kuramoto-Daido model," which proves a Hopf bifurcation of the incoherent equilibrium and explicitly notes that it does not study a secondary Hopf bifurcation in the strict sense (Chiba, 2016).

This distinction matters dynamically. A supercritical primary Hopf bifurcation creates a stable small-amplitude periodic orbit, which can then potentially undergo secondary instabilities; a subcritical primary Hopf bifurcation involves an unstable small-amplitude periodic orbit and often hysteresis or coexistence, changing what later bifurcations are dynamically accessible (Das et al., 2013). This suggests that primary criticality helps determine the dynamical setting from which a secondary Hopf bifurcation may later arise.

3. Normal hyperbolicity and stability of the bifurcating torus

A central recent result is that the torus created in a secondary Hopf bifurcation is normally hyperbolic (Pereira et al., 14 Jul 2025). The main theorem in (Pereira et al., 14 Jul 2025) states:

The invariant torus provided by the standard secondary Hopf theorem is normally hyperbolic. It has, exactly and respectively, $\dot x = F(x,\mu),$8 and $\dot x = F(x,\mu),$9 stable directions in the subcritical and supercritical case.

Here $\gamma_\mu$0, where $\gamma_\mu$1 and $\gamma_\mu$2 count the noncritical multipliers of the Poincaré map inside and outside the unit circle at the bifurcation point (Pereira et al., 14 Jul 2025). Thus:

• in the subcritical case $\gamma_\mu$3, the torus has exactly $\gamma_\mu$4 stable directions;
• in the supercritical case $\gamma_\mu$5, the torus has exactly $\gamma_\mu$6 stable directions (Pereira et al., 14 Jul 2025).

The proof combines two ingredients already available in the literature. First, the invariant closed curve born in the Neimark–Sacker bifurcation of the Poincaré map is normally hyperbolic (Pereira et al., 14 Jul 2025). Second, Fenichel’s thin-surface-section theory lifts that normally hyperbolic invariant curve to a normally hyperbolic invariant torus for the flow (Pereira et al., 14 Jul 2025). The relevant return time is

$\gamma_\mu$7

This normal hyperbolicity is not an additional assumption; it is deduced from the classical secondary-Hopf hypotheses (Pereira et al., 14 Jul 2025). A plausible implication is that the torus is not merely locally invariant but belongs to the robust class of normally hyperbolic invariant manifolds, with persistence under perturbations and stable or unstable manifolds in the standard Fenichel–Hirsch–Pugh–Shub sense, which is exactly how the paper motivates its averaging application (Pereira et al., 14 Jul 2025).

4. Nonautonomous analogues and generalized tori

The paper "A model for the nonautonomous Hopf bifurcation" studies what Arnold called the two-step nonautonomous Hopf bifurcation scenario (Anagnostopoulou et al., 2013). It does not use the terminology “secondary Hopf bifurcation” explicitly, but its second bifurcation point is closely analogous to a secondary Hopf or torus bifurcation in a forced or skew-product setting (Anagnostopoulou et al., 2013).

The model is a skew product

$\gamma_\mu$8

with

$\gamma_\mu$9

After passage to projective polar coordinates, the system becomes a double skew product

$\Sigma$0

where

$\Sigma$1

The bifurcation thresholds are

$\Sigma$2

in the deterministic case, and

$\Sigma$3

in the random case (Anagnostopoulou et al., 2013).

Between the two thresholds, the attractor is not yet a disk but a line segment in a projectively selected direction (Anagnostopoulou et al., 2013). After the second threshold, the upper bounding radius becomes strictly positive for all projective angles, every fiber fills out to a topological disk, and the boundary is the new invariant torus-like object (Anagnostopoulou et al., 2013). In the deterministic setting,

$\Sigma$4

This is the paper’s “generalised torus” (Anagnostopoulou et al., 2013).

The analogy with secondary Hopf is explicit: a previously stable lower-dimensional invariant object loses transverse stability, a torus-like invariant set is born, and in the deterministic quasiperiodic forcing case with $\Sigma$5, the post-bifurcation set can literally be a two-torus (Anagnostopoulou et al., 2013). The differences are equally explicit: the bifurcating object is not a periodic orbit of an autonomous system, the torus is fiberwise and driven by a base system, and in the random case the full generalized torus is a pullback attractor while the forward attractor may remain only a random two-point set embedded in the torus (Anagnostopoulou et al., 2013).

5. Codimension-two routes: Hopf–Hopf and multiple oscillatory modes

A major route to torus birth and more complicated secondary oscillatory behavior is through codimension-two Hopf–Hopf bifurcation. The neutral functional differential equation paper (Niu et al., 2014) studies a nonresonant Hopf–Hopf singularity in a class of NFDEs. At the critical parameter value, the linearization has two distinct pairs of purely imaginary eigenvalues,

$\Sigma$6

with nonresonance

$\Sigma$7

(Niu et al., 2014). Its cubic normal form on the center manifold leads to amplitude equations

$\Sigma$8

$\Sigma$9

After rescaling, this becomes the standard double-Hopf amplitude system

$p \mapsto \Pi(p,\mu).$0

(Niu et al., 2014). In the van der Pol example with extended delay feedback, the realized unfolding is case VIa, and the corresponding bifurcation diagram contains parameter regions with periodic solutions, quasi-periodic solutions on 2D tori, quasi-periodic solutions on 3D tori, and chaotic behavior after torus breakdown (Niu et al., 2014).

The diffusive Leslie–Gower predator–prey system with two delays gives a closely related picture (Du et al., 2018). Its double Hopf point occurs when two switching curves intersect in the $p \mapsto \Pi(p,\mu).$1-plane, producing two distinct pairs of imaginary eigenvalues (Du et al., 2018). The center-manifold reduction yields the amplitude system

$p \mapsto \Pi(p,\mu).$2

from which the local dynamics include periodic orbits, quasi-periodic solutions on a 2-torus, quasi-periodic solutions on a 3-torus, and strange attractors (Du et al., 2018). This is not a secondary Hopf bifurcation of a pre-existing periodic orbit in the narrow codimension-one sense, but it is a standard codimension-two route to the same torus-type and multi-frequency phenomena.

The related predator–prey model with fear effect reaches the same conclusion: it studies a nonresonant Hopf–Hopf bifurcation at the positive equilibrium, derives the normal form on the center manifold, and identifies quasi-periodic motion on a 2-torus near the Hopf–Hopf point, together with numerical evidence for three-torus and strange-attractor behavior under further parameter variation (Duan et al., 2018). This suggests that, in delay and reaction–diffusion systems, “secondary oscillatory behavior” is often organized by equilibrium-level Hopf–Hopf interactions rather than only by isolated torus bifurcations of a single periodic orbit.

6. Extensions, variants, and limits of the term

Several papers in the corpus delimit the scope of the phrase “secondary Hopf bifurcation” by contrast.

The Navier–Stokes paper "Hopf bifurcation of a non-parallel Navier-Stokes flow" studies what it explicitly calls a secondary Hopf bifurcation of a non-parallel flow: the base state is already a nontrivial steady square-eddy solution, and a branch of small-amplitude time-periodic solutions issues from it when dissipation parameters vary (Chen, 2024). The steady state is the stream function

$p \mapsto \Pi(p,\mu).$3

and the critical eigenvalue problem is

$p \mapsto \Pi(p,\mu).$4

Under the paper’s simplicity and transversality assumptions, a unique local branch of periodic solutions exists in a neighborhood of the critical state (Chen, 2024). This use of “secondary Hopf” differs from the torus-birth meaning of (Pereira et al., 14 Jul 2025): here the bifurcation is steady state $p \mapsto \Pi(p,\mu).$5 time-periodic flow, but the base state is itself already a nontrivial equilibrium rather than the trivial state.

The complexified Kuramoto paper "Hopf-Induced Desynchronization" is relevant in yet another sense (Lee et al., 20 Jun 2025). It analyzes a detached branch of synchronous locked states and reports that local numerical eigenvalue analysis suggests destabilization through a Hopf bifurcation of that pre-existing branch. The paper is careful: the abstract says “Local numerical analysis suggests that the transition is linked to a Hopf bifurcation destabilizing synchrony” (Lee et al., 20 Jun 2025). This is secondary relative to the synchronous family, but the Hopf identification is numerical rather than a rigorous local theorem, and no torus or post-Hopf branch is classified (Lee et al., 20 Jun 2025).

In discrete dynamics, the paper on heterodimensional cycles via homoclinic tangencies studies a genuine Neimark–Sacker bifurcation of long-period periodic orbits $p \mapsto \Pi(p,\mu).$6 in a three-dimensional diffeomorphism (Tomizawa, 19 May 2025). The critical multipliers satisfy

$p \mapsto \Pi(p,\mu).$7

and the first Lyapunov coefficient is

$p \mapsto \Pi(p,\mu).$8

This is the map analogue of secondary Hopf in the strict dynamical-systems sense, although the paper’s focus is the construction of heterodimensional cycles rather than invariant circles as such (Tomizawa, 19 May 2025).

By contrast, the review "Twenty Hopf-like bifurcations in piecewise-smooth dynamical systems" treats planar nonsmooth equilibrium-to-limit-cycle transitions, not torus birth or quasiperiodicity (Simpson, 2019). Its relevance is contrastive: in this literature, “Hopf-like” means stationary solution to limit cycle, whereas classical “secondary Hopf” means periodic orbit to torus (Simpson, 2019).

7. Conceptual synthesis

Across these works, three distinct usages emerge.

First, in the classical autonomous smooth sense, secondary Hopf bifurcation means a Neimark–Sacker bifurcation of the Poincaré map of a periodic orbit, creating an invariant torus. The decisive local data are a single critical pair of Floquet multipliers on the unit circle, transversality

$p \mapsto \Pi(p,\mu).$9

nonresonance up to order $p^*(\mu)$0, and a nonzero first Lyapunov coefficient $p^*(\mu)$1 (Pereira et al., 14 Jul 2025).

Second, in forced, random, or skew-product settings, the closest analogue is the creation of a generalized torus at a second bifurcation threshold, as in Arnold’s two-step nonautonomous Hopf scenario (Anagnostopoulou et al., 2013). Here the invariant object is torus-like but intrinsically nonautonomous.

Third, in codimension-two oscillatory settings such as Hopf–Hopf bifurcation, the direct object of study is often an equilibrium with two critical frequency pairs, but the unfolding naturally contains the torus and multi-frequency dynamics that motivate the phrase “secondary oscillation” (Niu et al., 2014, Du et al., 2018, Duan et al., 2018). In these cases the reduced amplitude systems

$p^*(\mu)$2

or

$p^*(\mu)$3

organize the transition from one-frequency oscillation to invariant tori and, in some examples, to higher-dimensional tori and chaos (Niu et al., 2014, Du et al., 2018).

A plausible implication is that “secondary Hopf bifurcation” is best treated as a precise local codimension-one torus bifurcation when the object is a periodic orbit and a Poincaré map is available, but as a broader torus-creating or multi-frequency organizing mechanism in nearby literatures. The papers considered here consistently preserve the distinction between that strict meaning and the wider family of oscillatory transitions to which it is often compared.