Data-driven inference of Hopf normal form representations from oscillatory time series
Shinsuke Koyama, Ryota Kobayashi · 2026
Read Data-driven inference of Hopf normal form representations from oscillatory time series by Shinsuke Koyama, Ryota Kobayashi (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-26847.txt.
Introduction
Shinsuke Koyama and Ryota Kobayashi infer the hidden dynamics of rhythmic systems directly from noisy recordings. Oscillations run through brains, hearts, and daily biological cycles. The Hopf normal form is the universal low-dimensional description near the onset of oscillation. The authors embed that form in a probabilistic model of the data. Latent states and system parameters are estimated together.
The method recovers natural frequency, growth rate, and phase robustly, even far from the onset point and under strong noise. Combined with Gaussian process regression, it also rebuilds how phase and amplitude respond to nudges. Tests on the van der Pol oscillator, a classic self-sustaining rhythm, beat existing phase-based and regression methods clearly. The work links normal-form theory with statistical inference for messy real signals.
Catalog record
| CREATOR | Shinsuke Koyama, Ryota Kobayashi |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.26847
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Data-driven inference of Hopf normal form representations from oscillatory time series Shinsuke Koyama, Ryota Kobayashi Abstract We introduce a data-driven framework that maps noisy oscillatory time series directly onto the Hopf normal form, enabling inference of underlying dynamics without knowledge of governing equations. By embedding the normal form in a probabilistic state-space model, the method jointly infers latent states and system parameters, yielding robust estimates of the natural frequency, Floquet exponent, and asymptotic phase even far from the bifurcation point and under strong noise. Combined with complex Gaussian process regression, the approach further reconstructs phase and amplitude sensitivity functions from data. Benchmarks on the van der Pol oscillator demonstrate substantially improved accuracy and noise robustness compared with existing phase-based and regression methods. This work establishes a direct bridge between normal-form theory and statistical inference, providing a general and practical route to low-dimensional descriptions of oscillatory dynamics in complex systems. I Introduction Rhythmic activity is a ubiquitous phenomenon observed across a wide range of biological systems. Representative examples include oscillatory dynamics in cortical networks of the brain ( 3 ; 38 ) , circadian rhythms in mammals ( 21 ) , cardiac and respiratory cycles in humans ( 27 ; 16 ; 15 ) , and locomotor patterns such as animal gait ( 4 ; 13 ; 9 ) . Owing to their repetitive and self-sustained nature, such phenomena are commonly modeled as limit cycle oscillators ( 33 ; 45 ) , which provide a fundamental mathematical framework for describing stable periodic behavior. Studying limit cycle oscillators is essential for understanding the dynamics of oscillatory systems, as well as for enabling the inference and control of their behavior in both natural and artificial settings. A canonical mechanism underlying the emergence of limit cycle oscillations is the Hopf bifurcation. In this scenario, a stable fixed point characterized by a pair of complex-conjugate eigenvalues loses stability as a system parameter varies, leading to the birth of a stable limit cycle from infinitesimal perturbations. Because of its structural robustness and universality, the Hopf bifurcation serves as a central theoretical paradigm for explaining the onset of rhythmic behavior in diverse physical and biological systems. To analyze the dynamics near bifurcation points in a unified framework, nonlinear dynamical systems can often be simplified through an appropriate change of variables into a normal form ( 11 ; 40 ; 19 ) . The normal form framework provides a systematic way to study dynamics near bifurcation points by reducing complex behavior to a tractable low-dimensional system, while preserving the qualitative properties of the original system through topological equivalence. In particular, systems undergoing a Hopf bifurcation are locally topologically equivalent to a cubic normal form, which captures the essential mechanism responsible for the emergence and stabilization of a limit cycle. The normal form framework has advanced our theoretical understanding of nonlinear dynamics ( 12 ) . However, its application to the quantitative analysis of observed data remains relatively limited. While a coordinate transformation to the normal form is mathematically guaranteed to exist near bifurcation points, the explicit functional form of this transformation is generally unknown. As a result, identifying normal form representations directly from data remains a challenging problem. Recently, a data-driven framework has been proposed to infer dynamical systems near a Hopf bifurcation from estimated phase and amplitude response curves ( 44 ) . Nevertheless, inferring representations consistent with the normal form from noisy observations remains challenging, particularly in the presence of substantial observation noise. The primary objective of this article is to bridge normal form theory and data-driven analysis. Specifically, we propose a method for inferring representations consistent with the Hopf normal form from noisy observations of oscillatory activity governed by unknown dynamical equations. This representation enables the extraction of intrinsic dynamical properties of the underlying system in a data-driven manner. We achieve this by embedding the Hopf normal form within a state-space modeling framework ( 8 ; 14 ) , which enables the inference of coordinates consistent with the normal form. The proposed framework consists of two components. The first component is based on a state-space approach that estimates model parameters, including the natural frequency and the nontrivial Floquet exponent, from noisy oscillatory time-series data. This approach additionally assigns a latent state, representing phase and amplitude, to each observation. The second component extends these pointwise estimates by interpolating the complex-valued latent states over unobserved regions of the observation space using complex-valued Gaussian process regression. In particular, this approach enables the estimation of phase- and amplitude-sensitivity functions. Compared with existing approaches, the proposed framework exhibits improved robustness to observation noise while accurately inferring the underlying dynamical properties. The main contributions of this study are summarized as follows: 1. We propose a data-driven method for constructing representations consistent with the Hopf normal form from noisy observational time-series data, without requiring prior knowledge of the governing equations. 2. We develop a state-space–based model for robust estimation of system parameters and latent phase–amplitude states under observation noise. 3. We infer intrinsic dynamical properties, including phase and amplitude sensitivity functions, from the inferred representations. The remainder of this article is organized as follows. Section II briefly reviews the Hopf normal form. Section III presents the proposed methodology, with detailed derivations and algorithmic procedures provided in the Appendix. Section IV demonstrates the performance of the proposed approach using synthetic data generated from a van der Pol oscillator. Finally, Section V concludes the article with a summary of the main findings and a discussion of potential extensions and implications. II Hopf normal form In this section, we briefly review the Hopf normal form, which serves as the foundation for the proposed method developed in the subsequent sections. Consider an autonomous dynamical system described by the vector field \frac{d\bm{y}(t)}{dt}=F(\bm{y}(t)),\quad\bm{y}(t)\in\mathbb{R}^{D},\quad(D\geq 2), (1) and assume that the systesm possesses a stable periodic orbit arising from a fixed point via a supercritical Hopf bifurcation. In a neighborhood of the Hopf bifurcation point, the system can be transformed, via an appropriate smooth change of variables, into a canonical normal form known as the Stuart–Landau equation ( 17 ; 22 ) . For the system ( 1 ), we consider the following canonical form in the vicinity of the Hopf bifurcation point: \frac{d}{dt}\begin{pmatrix}x_{1}\\ x_{2}\end{pmatrix}=\begin{pmatrix}\alpha x_{1}-\omega x_{2}-\alpha x_{1}\left(x_{1}^{2}+x_{2}^{2}\right)\\ \omega x_{1}+\alpha x_{2}-\alpha x_{2}\left(x_{1}^{2}+x_{2}^{2}\right)\end{pmatrix}, (2) where \omega>0 denotes the angular frequency and \alpha>0 determines the rate of attraction toward the limit cycle. Introducing polar coordinates (r,\theta) , the system ( 2 ) can be rewritten as \displaystyle\frac{d\theta}{dt} \displaystyle=\omega, (3) \displaystyle\frac{dr}{dt} \displaystyle=\alpha(r-r^{3}). (4) These equations show that the system admits a stable periodic orbit on the unit circle, along which the motion is a uniform rotation with constant angular velocity \omega . The parameter \alpha governs the exponential convergence of trajectories toward the limit cycle and is directly related to the Floquet exponent of the limit-cycle by \lambda=-2\alpha . According to Hopf bifurcation normal form theory, the original system ( 1 ) is locally topologically equivalent to ( 2 ) in a neighborhood of the bifurcation point, ensuring that the qualitative dynamical behavior of the system is preserved under the transformation. In Appendix A , we review the conventional phase–amplitude representation of limit-cycle oscillators ( 17 ; 39 ; 22 ; 29 ;