Floating point system error estimation in terms of machine precision epsilon -
i'm doing question ask me derive upper error bound in terms of machine precision episilon computer estimation of express (x/z)*y given fl(fl(x)*fl(y)) = (x*y)(1-delta)
my approach fl(fl(fl(x)/fl(z))fl(y)) = xy/z(1-delta1)(1-delta2)(1-delta3)(1-delta4)/(1-delta5), don't know (1-delta5) since don't know how cancel term.
Comments
Post a Comment