September 12 2005 9:38:45.712 AM DIVISION: Demonstrate 16 ways to compute 1/3 using double precision. The results, stored as a table, are: 1 / 3 1 / 3.0 1 / 3.0E+00 1 / 3.0D+00 1.0 / 3 1.0 / 3.0 1.0 / 3.0E+00 1.0 / 3.0D+00 1.0E+00 / 3 1.0E+00 / 3.0 1.0E+00 / 3.0E+00 1.0E+00 / 3.0D+00 1.0D+00 / 3 1.0D+00 / 3.0 1.0E+00 / 3.0E+00 1.0D+00 / 3.0D+00 The matrix of results: 0.0000000000000 0.33333334326744 0.33333334326744 0.33333333333333 0.33333334326744 0.33333334326744 0.33333334326744 0.33333333333333 0.33333334326744 0.33333334326744 0.33333334326744 0.33333333333333 0.33333333333333 0.33333333333333 0.33333333333333 0.33333333333333 Defects might be more obvious if we multiply by 3: B = 3 * A: 0.0000000000000 1.0000000298023 1.0000000298023 1.0000000000000 1.0000000298023 1.0000000298023 1.0000000298023 1.0000000000000 1.0000000298023 1.0000000298023 1.0000000298023 1.0000000000000 1.0000000000000 1.0000000000000 1.0000000000000 1.0000000000000 B = 3.0 * A: 0.0000000000000 1.0000000298023 1.0000000298023 1.0000000000000 1.0000000298023 1.0000000298023 1.0000000298023 1.0000000000000 1.0000000298023 1.0000000298023 1.0000000298023 1.0000000000000 1.0000000000000 1.0000000000000 1.0000000000000 1.0000000000000 B = 3.0E+00 * A: 0.0000000000000 1.0000000298023 1.0000000298023 1.0000000000000 1.0000000298023 1.0000000298023 1.0000000298023 1.0000000000000 1.0000000298023 1.0000000298023 1.0000000298023 1.0000000000000 1.0000000000000 1.0000000000000 1.0000000000000 1.0000000000000 B = 3.0D+00 * A: 0.0000000000000 1.0000000298023 1.0000000298023 1.0000000000000 1.0000000298023 1.0000000298023 1.0000000298023 1.0000000000000 1.0000000298023 1.0000000298023 1.0000000298023 1.0000000000000 1.0000000000000 1.0000000000000 1.0000000000000 1.0000000000000 DIVISION: Normal end of execution. September 12 2005 9:38:45.717 AM