No label defined (Q4944262)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 1417964
Language Label Description Also known as
default for all languages
No label defined
    English
    No label defined
    scientific article; zbMATH DE number 1417964

      Statements

      0 references
      0 references
      19 March 2000
      0 references
      multilevel method
      0 references
      spectral method
      0 references
      numerical examples
      0 references
      method of characteristics
      0 references
      periodic solution
      0 references
      convection-diffusion problem
      0 references
      stability
      0 references
      error estimates
      0 references
      The authors consider the linear unsteady convection-diffusion problem that involves a function \(u(x,t)\) from \(]0,2\pi[^d\times \mathbb{R}^+\) into \(\mathbb{R}\): \[ {\partial u\over\partial t}+ a.\nabla u-\nu\Delta u= f, \] where \(\nu\) is the viscosity, \(f\) is the forcing term and \(a\) is some divergence-free vector field. The initial conditions are that \(u(x,0)= u_0(x)\) for all \(x\in ]0,2\pi[^d\) and \(u\) is \(2\pi\)-periodic in all space variables. The spatial discretization uses the space \(S_M\) of all trigonometric polynomials of degree \(\leq M\) in each variable. A two-level decomposition is introduced by considering another parameter \(m\), \(0< m< M\) and writing \(S_M= S_M= S_m+ (I- P_m)S_M\) where \(P_m\) is the \(L^2\)-projection onto \(S_m\). The numerical integration then relies on two different interpolation opeators and the approximate solution is obtained as the sum of a term in \(S_m\) and a term in \((I- P_m)S_M\). The two components are advanced in time using different time steps. The authors investigate the stability fo this scheme and derive error estimates. These indicate that the high-frequency term can be integrated with a larger time-step. The paper contains a few numerical tests for 1D-problems. It is found that the two-level method allows a significant gain in computing time with respect to the classical method. The accuracy is always better than the one of the classical method based on \(S_m\) and close to the one based on \(S_M\).
      0 references

      Identifiers