Normal form construction for nearly-integrable systems with dissipation (Q691238)

From MaRDI portal





scientific article; zbMATH DE number 6111438
Language Label Description Also known as
English
Normal form construction for nearly-integrable systems with dissipation
scientific article; zbMATH DE number 6111438

    Statements

    Normal form construction for nearly-integrable systems with dissipation (English)
    0 references
    0 references
    0 references
    30 November 2012
    0 references
    Nearly integrable Hamiltonian systems can be modeled by a Hamiltonian function as \(H(y,x,t)=h_{00}(y)+\epsilon h_{10}(y,x,t)\) where \(y\in R^\ell\), and \( (x,t)\in T^{\ell+1}\), \(T\) is the torus and \(\epsilon\) is a small parameter. If \(\epsilon=0\) then the system is integrable with frequency \(\omega(y)\equiv \frac{\partial h_{00}(y)}{\partial y}\). For \(\epsilon\neq 0\), the Hamilton equations are \({\dot x}=\omega(y)+\epsilon h_{10,y}(y,x,t),\;{\dot y}=-\epsilon h_{10,x}(y,x,t)\). For a nearly integrable dissipative system, the authors' equations are \({\dot x}=\omega(y)+\epsilon h_{10,y}(y,x,t)+\mu f_{01}(y,x,t),\;{\dot y}=-\epsilon h_{10,x}(y,x,t)+\mu(g_{01}(y,x,t)-\eta)\) where \(\mu\in R_+\) is the dissipative parameter and \(\eta \in R^{\ell}\) is called the drift vector. Here the dissipative term \(\mu\) is assumed to be small. The authors show how to achieve a normal form transformation \((Y,X,t)=\Xi^{(N)}(y,x,t)\) to a suitable order \(N\) such that the system becomes \({\dot X}=\Omega_d (Y)+ h.o.t.,\;{\dot Y}=h.o.t\), where \(\Xi\) is of order not greater than \(N\) as a polynomial in \(\epsilon,\mu\) and \(X=x\), \(Y=y\) when \(\epsilon=\mu=0\). This transformation is possible for \(\omega(y)\) irrational and if \(\eta\) satisfies some compatibility properties. The construction is complicated, but the authors provide good motivation for their work and the exposition is very clear. They illustrate their normal form calculation with the example \({\dot x}=y-\mu(\sin(x-t)+\sin(x)),\;{\dot y}=-\epsilon(\sin(x-t)+\sin(x))-\mu(y-\eta)\). They construct the normal form for this example, make a comparison with numerical integration, and determine the region of validity for the normal form.
    0 references
    dissipative system
    0 references
    normal form
    0 references
    non-resonant motion
    0 references
    nearly-integrable system
    0 references

    Identifiers