A fundamental problem in the study of dynamical systems is to ascertain whether the e fect of a perturbation on an integrable Hamiltonian system accumulates over time and leads to a large e fect (instability) or it averages out (stability). Instabilities in nearly integrable systems, usually called Arnold diffusion, take place along resonances and by means of a framework of partially hyperbolic invariant objects and their homoclinic and heteroclinic connections.

The goal of this project is to develop new techniques, relying on the role of invariant manifolds in the global dynamics, to prove the existence of physically relevant instabilities and homoclinic phenomena in several problems in celestial mechanics and Hamiltonian Partial Di ferential Equations.

The N-body problem models the interaction of N puntual masses under gravitational force. Astronomers have deeply analyzed the role of resonances in this model. Nevertheless, mathematical results showing instabilities along them are rather scarce. This ERC project plans to develop a new theory to analyze the transversal intersection between invariant manifolds along mean motion and secular resonances to prove the existence of Arnold di usion. This theory will also be applied to construct oscillatory motions.

Several Partial Di erential Equations such as the nonlinear Schrodinger, the Klein-Gordon and the wave equations can be seen as infi nite dimensional Hamiltonian systems. Using dynamical systems techniques and understanding the role of invariant manifolds in these HamiltonianPDEs, two type of solutions will be studied: transfer of energy solutions, namely solutions that push energy to arbitrarily high modes as time evolves by drifting along resonances; and breathers, spatially localized and periodic in time solutions, which in a proper setting can be seen as homoclinic orbits to a stationary solution.