Action minimizing invariant measures for positive definite Lagrangian systems (Q910781)

From MaRDI portal

This is the item page for this Wikibase entity, intended for internal use and editing purposes.
scientific article; zbMATH DE number 4140874
Language Label Description Also known as
default for all languages
No label defined
    English
    Action minimizing invariant measures for positive definite Lagrangian systems
    scientific article; zbMATH DE number 4140874

      Statements

      Action minimizing invariant measures for positive definite Lagrangian systems (English)
      0 references
      0 references
      1991
      0 references
      Let \(M\) be a compact smooth manifold and let \(L: TM\times {\mathbb{R}}\to {\mathbb{R}}\) be a time dependent Lagrangian of class \(C^ 2\) which is periodic of period one in the time variable (i.e. \(L(\xi,z+1)=L(\xi,t))\), satisfies the Legendre condition (in local coordinates \(L_{\dot x\dot x}>0)\), and has superlinear growth along the fibers of \(TM\) (i.e. \(L(\xi,t)/\| \xi \| \to +\infty\) as \(\| \xi \| \to +\infty\), for \(\xi\in TM\), \(t\in {\mathbb{R}})\). Let \(\Phi_ L\) denote the associated Euler-Lagrange flow. Assume that \(\Phi_ L\) is complete. For every \(\Phi_ L\)-invariant probability measure \(\mu\) on TM\(\times {\mathbb{R}}/{\mathbb{Z}}\) define the average action as \(A(\mu)=\int L\, d\mu\) and the rotation vector \(\rho (\mu)\in H_ 1(M,{\mathbb{R}})\) by \(\langle[\lambda],\rho (\mu)\rangle=\int \lambda\, d\mu\), where \(\lambda : TM\to {\mathbb{R}}\) is a closed 1-form on \(M\) and \([\lambda]\in H^ 1(M,{\mathbb{R}})\) is its cohomology class. The rotation vector \(\rho\) (\(\mu)\) exists whenever \(A(\mu)<\infty.\) Theorem: For every \(h\in H_ 1(M,{\mathbb{R}})\), there exists a \(\Phi_ L\)- invariant probability measure \(\mu\) such that \(A(\mu)<\infty\) and \(\rho (\mu)=h\). Moreover, \(A\) takes a minimum value \(\beta(h)\) on the set of \(\Phi_ L\)-invariant probability measures \(\mu\) such that \(A(\mu)<\infty\) and \(\rho (\mu)=h\). The function \(\beta\) is convex and has superlinear growth. For \(c\in H^ 1(M,{\mathbb{R}})\), let \({\mathfrak M}_ c\) denote the set of \(\Phi_ L\)-invariant probability measures \(\mu\) which minimize \(A_ c(\mu)=A(\mu)-\langle,\rho (\mu)\rangle\). Let \(\text{supp }{\mathfrak M}_ c\) denote the support of \({\mathfrak M}_ c\). Let \(\pi : TM\times {\mathbb{R}}/{\mathbb{Z}}\to M\times {\mathbb{R}}/{\mathbb{Z}}\) denote the projection. Theorem. Supp \({\mathfrak M}_ c\) is compact. The restriction of \(\pi\) to supp \({\mathfrak M}_ c\) is injective. The inverse mapping from \(\pi (\text{supp } {\mathfrak M}_ c)\) to \(\text{supp }{\mathfrak M}_ c\) is Lipschitz. An application of these results is given to \(C^ 1\) small perturbations to a Hamiltonian system having a KAM torus satisfying a positive definiteness condition in the normal direction. Also, it is shown how basic results concerning area preserving twist maps follow from these results.
      0 references
      Lagrangian system
      0 references
      Euler-Lagrange flow
      0 references
      probability measure
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references