“…In this case, the nodes of the start grid are those of all subsequent grids. At these points, we estimate (see, e.g., ) the error: and the effective accuracy order: At each points ( x , t ), where the solution of the problem has bounded first derivatives in time and second derivative in the spatial variables, and the error estimate is asymptotically sharp as N , M → ∞ . The violation of convergence () indicates the loss of smoothness of the exact solution.…”