Symplectic dynamical low rank approximation of wave equations with random parameters (Q2216487)

The authors of the work focused on second-order wave equations with random parameters such as acoustic/elastic waves with random speed/source terms and whose applications include seismology. The goal of the work is to extend the so-called Dynamic Low Rank (DLR) method which was originally designed for parabolic problems to second-order wave equations. They proposed a Symplectic Dynamical Low Rank (SDLR, a combination of symplectic order reduction and DLR) or Symplectic Dynamically Orthogonal (DO) method for wave equations with random parameters rewritten in Hamiltonian form to preserve the underlying geometric structure of the system. Deterministic reduced basis that is symplectic and orthonormal is used in the construction of the method due to challenges in finding a characterization of the manifold of approximate solutions and parameterization of its tangent speed at each point. It is stated that the approximate solution constructed by the method preserves the (approximated) mean energy and continuously adapts in time to the structure of the solution. Numerical examples are for various cases of random parameters and/or initial conditions are shown and several validation tests are carried out.