Suppose you had to evaluate
, given
. The straightforward approach would sum up
-- 10
multiplications 4 additions. Although multiplication and division
don't generate as much error at a single step as catastrophic
cancellation (the operations are done in double precision, and then
converted to float), the errors accumulate (and multiplication is
computationally expensive). A better approach is to re-write the
polynomial using Cramer's rule: