[PPT] - On the maximum relative error when computing x n in floating-point PowerPoint Presentation