Expressing the analysis domains with meshes is common in simulations for fluid dynamics and structural mechanics. The times for calculations and precisions depend on the meshes, so suitable meshes are needed. For phenomena without proper directions of changes, isotropic meshes are preferable. Our aim is improving 2-dimensional triangular mesh isotropy for such simulations. We report inserting vertices into and optimizing meshes generated by an algorithm DistMesh proposed by Persson and Strang [18]. Then we verify the improvements with smoothing after optimization comparing with some results for uniform and adaptive meshes respectively, and show the advantages of our proposed method.
We are concerned with an efficient numerical solution of linear equations at each time-stepping of the trapezoidal rule applied to a system of linear ordinary differential equations (ODEs) with a constant coefficient matrix of large dimension. We do not assume that the matrix is symmetric. Hence numerical solutions in the family of BiCG method are sought. The present paper describes a method to reuse Krylov subspaces in the BiCGSTAB process over a number of computational steps. It can suppress increase of the memory usage as well as reduce the total number of BiCGSTAB iterations. Numerical examples depict its efficiency.