By Todd F. Dupont and Itir Mogultay
Abstract: A symmetric error estimate (SEE) for a numerical method is a statement that in some norm the numerical approximation is a best fit, up to a constant factor. For example, for Galerkin methods for second order elliptic problems the simplist error estimate is an SEE in the H1 norm. There are various SEEs for parabolic prolems. Recently, some of these have been extended to Navier-Stokes problems. In certain cases of interest, the results degrade to almost-symmetric error estimates.