Nikulchev's Jet Space Method Beats Takens in Chaos Reconstruction

Evgeny Nikulchev's new paper challenges a 40-year-old pillar of nonlinear time series analysis: Takens' theorem. While Takens embedding guarantees only topological equivalence, it often distorts metric and group properties—especially symmetries crucial for understanding chaotic systems. Nikulchev proves that by moving from delay-coordinate space to jet space (the space of the signal and its derivatives), one can exactly preserve the original system's Lie algebra of symmetries. The theoretical result is backed by numerical experiments on the Lorenz and Rössler attractors, using variational elastic energy and correlation dimension as quantitative metrics. In some cases, jet-space reconstruction yields more accurate invariants than even direct projections of the true system.

This breakthrough offers a practical, coordinate-invariant method for classifying strange attractors and detecting hidden attractors—a long-standing challenge in chaos theory. For AI and ML practitioners, it suggests better ways to embed time series for tasks like forecasting, anomaly detection, and system identification. The paper (arXiv:2606.24929) is 18 pages with 2 figures, and its implications reach beyond pure mathematics into any field where chaotic dynamics must be faithfully reconstructed from limited data.