Break-up of invariant surfaces in action-angle-angle maps and flows (Q5940278)

An integrable action-action-angle map on a three-dimensional manifold \(M=B\times S^1\) with action-action-angle coordinates \((I_1,I_2,\varphi)\), is one of the form \(I_j\to I_j'= F_j(I)=I_j\), \(\varphi\to \varphi'= h(I,\varphi)= \varphi+ \Omega(I)\). Here \(j=1,2\) and \(I=(I_1,I_2)\); moreover \(I_j\in [a_j,b_j]\) and \(B=[a_1,b_1] \times[a_2,b_2]\). Such systems are interesting per se and relevant in applications. Any manifold \(M_0=V \times S^1\) with \(V\to B\) is obviously invariant; invariant manifolds not of this form are called of resonant type; on these \(\Omega(I_1,I_2)= 2\pi p/q\) with \(p,q\) integers. These manifolds can only be zero- or one-dimensional. Here the author considers volume-preserving perturbations of an action-action-angle system, of the form \(F_j=I_j+ \varepsilon f_j (I,\varphi,\varepsilon)\), \(h= \varphi+\Omega (I)+\varepsilon g(I,\varphi, \varepsilon)\), with \(\Omega,F_j,h\) analytic in all their variables. He studies persistence of invariant manifolds, focusing mainly on resonant type ones and on their stability. The analysis is also extended to flows, and the relevant example of the extended standard map is studied. The author is also specially interested in the mechanism of resonance-induced dispersion, which allows for transport from one adiabatic surface to another; he claims that this transport is facilitated by the existence of saddle-foci of different stability type arising close to the intersection of resonant surfaces with surfaces to which the average component of the perturbation in action variables is tangent. Note, that the paper deals with action-action-angle systems, and not with action-angle-angle ones.