Question

epsilon = (b)^(1-t) this is used to represent the smallest number that machine can store.. can anyone explain it clearly.....??

Answer 1

i saw that like 3 years ago, i dont remember quite well, but i looked this to guides, read them, and both looked good for me. And from what i read it isnt what u stated.

http://www.cecm.sfu.ca/organics/papers/corless/u/html/u.html
http://en.wikibooks.org/wiki/Floating_Point/Epsilon

Hope this help u. GL.

Answer 2

I assume you are talking about the smallest positive floating point number that can be stored using the standard 32-bit or 64-bit binary floating point data formats. And I assume you only are interested in *normalized* floating point numbers. Then if e is the smallest exponent permitted in that format (which will be a negative number) then the smallest positive floating point number is 2**(e-1)