Learning the Intrinsic Dimensionality of Fermi-Pasta-Ulam-Tsingou Trajectories: A Nonlinear Approach using a Deep Autoencoder Model
This work addresses the challenge of analyzing complex dynamical systems for researchers in physics and nonlinear dynamics, though it is incremental as it applies an existing deep learning method to a specific model.
The authors tackled the problem of determining the intrinsic dimensionality of high-dimensional trajectories in the Fermi-Pasta-Ulam-Tsingou model using a deep autoencoder, finding that trajectories lie on a 2D manifold in weakly nonlinear regimes and increase to 3D at a symmetry-breaking transition.
We address the intrinsic dimensionality (ID) of high-dimensional trajectories, comprising $n_s = 4\,000\,000$ data points, of the Fermi-Pasta-Ulam-Tsingou (FPUT) $β$ model with $N = 32$ oscillators. To this end, a deep autoencoder (DAE) model is employed to infer the ID in the weakly nonlinear regime ($β\lesssim 1$). We find that the trajectories lie on a nonlinear manifold of dimension $m^{\ast} = 2$ embedded in a $64$-dimensional phase space. The DAE further reveals that this dimensionality increases to $m^{\ast} = 3$ at $β= 1.1$, coinciding with a symmetry breaking transition, in which additional energy modes with even wave numbers $k = 2, 4$ become excited. Finally, we discuss the limitations of the linear approach based on principal component analysis (PCA), which fails to capture the underlying structure of the data and therefore yields unreliable results in most cases.