Recalling the definitions of § 1, we write
for
, and
for
. Consider first how one would compute
these quantities using an approximation to the real number
. If an approximation
to
is used which has
, then the error in
the corresponding approximation to
is about
. Thus, for
large n it will be impossible to compute
correctly. In addition, even for small
n, it would be impossible to decide whether or not
for some
.
Annotation Form Interface