From: Earnie B. <ear...@ya...> - 2002-09-19 12:59:15
|
"Stephen M. Webb" wrote: > > On September 18, 2002 05:06 pm, Oscar Fuentes wrote: > > > > Once we accept that floating point numbers are just approximations to > > real numbers, we should accept that a result is "correct" if it is > > "good enough". The definition of "good enough" depends on the kind of > > problem and calculations you are performing. On some fields of > > Physics, a 10 digit result is overspecified, as the involved measures > > and constants have just four or five digits of precision. > > My slide rule gives only 3 digits of precision. Slide rules were good > enough for decades. Perhaps all budding computer scientists should be > required to master the use of the slide rule before being allowed to > program with floating-point numbers. > I have one still in the original box with the original case if one desires to have one, perhaps I could arrange for a personal auction. ;^O Earnie. |