Dear Sirs,
Many thanks I dislike the claim that algebraicly equivalent expressions can somehow not be numerically equivalent. What is probably meant is that a calculator, and even a computer, will make rounding errors that become apparent when computing the value of an expression by two different methods. There is a lovely discussion of it (with nice examples) in the first two editions of CALCULUS by James Stewart (in the appendix entitled "Lies My Calculator Told Me"). He claims that the owner's manual that comes with a calculator often describes limitations. Otherwise, one finds such information in books on numerical methods. Chris
