Normalization flow (Q6069643)

The author proposes a continuous normalization for Hamiltonian systems near a nonresonant elliptic singular point. The system under consideration has the following form: \[\dot{z} = i \partial_{\bar{z}} \hat{H} , \qquad \dot{\bar{z}} = -i \partial_{\bar{z}} \hat{H},\] with \( \hat{H} = \hat{H} (z, \bar{z})\), where \( z = (z_1,\dots, z_n)\) and \(\bar{z} = (\bar{z_1}, \dots, \bar{z_n})\) are independent coordinates on \(\mathbb{C}^{2n}\). The Hamiltonian function has the form \(\hat{H} = H_2 + \hat{H}_0\) with \(\hat{H}_0 = O_3(z, \bar{z})\) and \[H_2(z,\bar{z}) = \sum_{j=1}^{n} \omega_j z_j \bar{z}_j,\] where \((\omega_1, \dots, \omega_n)\) are real numbers and \(O_3\) means \(O(|z|^3 + |\bar{z}^3|)\). Assuming that the frequency \(\omega\) is nonresonant, there exists a formal canonical near-identity change of variables \((z, \bar{z}) \rightarrow (Z, \bar{Z})\) with \(d\bar{z} \wedge dz = d\bar{Z} \wedge dZ\) so that the new Hamiltonian takes the form \(H(z, \bar{z}) = N(Z_1\bar{Z}_1, \dots, Z_n\bar{Z}_n)\). The change of variables is formal because \(Z = Z(z,\bar{z})\), \(\bar{Z} = \bar{Z}(z, \bar{z})\), and \(N\) are power series in general. The problem of the convergence or divergence of the normalizing transformation assuming that \(\hat{H}\) is analytic is here a central question. With \(\mathcal{F}\) defined as the space of all power series in the variables \(z\) and \(\bar{z}\), the author studies a flow on \(\mathcal{F}\), called normalization flow. In the shift \[H_2 + \hat{H}_0 \rightarrow H_2 + \phi^\delta(\hat{H}_0)\] for \(\hat{H}_0 \in \mathcal{F}\), the normalization flow is \(\phi^\delta\) with \( \delta > 0\); it leads to a canonical change of variables. The flow \(\phi^\delta\) is determined by an ordinary differential equation in \(\mathcal {F}\) given by \(\partial_\delta {H_0} = - \{ {\xi H_0}, H_2 + H_0 \} \) with \(H_0 = \hat{H}\) when \(\delta = 0\) where \(\{ , \}\) is the Poisson bracket and \(\xi\) is a linear operator on \(\mathcal{F}\). The author argues that solutions of this equation map Hamiltonian functions to their normal forms.