Chaotic dynamical systems are typically characterized by irregular, broadband behavior thatobscures underlying structure in low-dimensional observations. In this work, we demonstratethat such apparent complexity can arise from projection effects rather than intrinsic randomness.Using the Lorenz system as a canonical example, we apply delay embedding, singular valuedecomposition, and dynamic mode decomposition to reconstruct and analyze the underlyingdynamics. We show that the reconstructed state space reveals coherent geometric structure,that the dynamics admit a low-rank representation, and that a small number of dominantmodes is sufficient to capture the essential behavior of the system.Furthermore, the DMD spectrum reveals a structured set of eigenvalues, indicating thatthe dynamics can be interpreted in terms of coherent spatiotemporal modes with well-definedfrequencies and growth characteristics.These results suggest that chaotic systems, when analyzed in appropriate representations,exhibit hidden low-dimensional coherence. This provides a unified computational frameworkfor identifying structured representations of complex dynamical systems and motivates furtherinvestigation into the role of representation in distinguishing chaos from underlying order.
Melanie Giel (Thu,) studied this question.