SearcharxivSearch

arXiv subjects

Idan Pazi

Publications and source records attributed to Idan Pazi.

3 recordsLinked to original sources

Perturbed Families of Symmetric Interval Exchange Maps

A perturbed family of interval exchange maps (FIEMs) provides a natural two-\linebreak{}dimensional area-preserving extension of interval exchange maps, with each IEM parameterized by an action variable $y$. Such families arise, for example, as models for iso-energy return maps of perturbed pseudointegrable Hamiltonian impact systems. These maps inherit a time-reversal symmetry, motivating the study of symmetric FIEMs. In the unperturbed case, the dynamics are generically uniquely ergodic for almost every value of $y$, while a dense set of action values supports periodic intervals. Exploiting time-reversal symmetry, we characterize these intervals and show that symmetric periodic orbits correspond to their midpoints. Under perturbation, the action variable is no longer conserved and generically periodic intervals break into isolated elliptic or hyperbolic periodic orbits. For sufficiently small perturbations, symmetric periodic orbits persist and can be located by a one-dimensional search along symmetry lines. Associated bifurcations generating symmetric and asymmetric periodic orbits are described and connected to those of the standard map, viewed here as a perturbed family of two-interval exchange maps.

math.DS

Hovering Sets in Near Pseudo-Integrable Hamiltonian Impact Systems

The transition from rotational to discontinuous behavior of the return map of the perturbed oscillators-step system, a paradigm model for a perturbation of a pseudo-integrable Hamiltonian impact system, is studied. The form of the return map is derived, and a truncated form of this map is simulated and analyzed. For a set of parameters the existence of a hovering set, a set of non-resonant orbits that pass sometimes above the step and sometimes to its side, without ever impacting it, is established and quantified. Its destruction as the sign of the perturbation term is reversed is established.

nlin.CD

Unsupervised Scale-Invariant Multispectral Shape Matching

Alignment between non-rigid stretchable structures is one of the most challenging tasks in computer vision, as the invariant properties are hard to define, and there is no labeled data for real datasets. We present unsupervised neural network architecture based upon the spectral domain of scale-invariant geometry. We build on top of the functional maps architecture, but show that learning local features, as done until now, is not enough once the isometry assumption breaks. We demonstrate the use of multiple scale-invariant geometries for solving this problem. Our method is agnostic to local-scale deformations and shows superior performance for matching shapes from different domains when compared to existing spectral state-of-the-art solutions.

cs.CV