An \(L_2\) norm trajectory-based local linearization for low order systems (Q4673231)

The author compares the 1D or 2D equation \[ \frac{dx}{dt} = F(x(t)) \] with \[ \frac{dx}{dt} = A x(t), \] where the matrix \(A\) is found by \(L_2\)-norm minimization of the residual error \[ e = F(x) - Ax. \] This leads to an extension of the classical linearization method at an equilibrium, e.g., in the polynomial scalar case \[ F(x) = \alpha_1 x + \cdots + \alpha_N x^N \] the ``\(L_2\)-norm linearization'' is \[ a(x_0) = \alpha_1 + \frac{3}{4} \alpha_2x_0 + \cdots + \frac{3}{N+2} \alpha_N x_0^{N-1}. \] Also linear approximate normal forms and some simulations are computed.