ComputerArithmetic,Part6-UniversityofCalifornia,Santa.pptVIP

May 2015 Computer Arithmetic, Function Evaluation Slide * 23.5 Use of Approximating Functions Convert the problem of evaluating the function f to that of function g approximating f, perhaps with a few pre- and postprocessing operations Approximating polynomials need only additions and multiplications Polynomial approximations can be derived from various schemes The Taylor-series expansion of f(x) about x = a is f(x) = ? j=0 to ? f (j) (a) (x – a) j / j! The error due to omitting terms of degree m is: f (m+1) (a + m(x – a)) (x – a)m+1 / (m + 1)! 0 m 1 Setting a = 0 yields the Maclaurin-series expansion f(x) = ? j=0 to ? f (j) (0) x j / j! and its corresponding error bound: f (m+1) (mx) xm+1 / (m + 1)! 0 m 1 Efficiency in computation can be gained via Horner’s method and incremental evaluation May 2015 Computer Arithmetic, Function Evaluation Slide * Some Polynomial Approximations (Table 23.1) ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Func Polynomial approximation Conditions ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– 1/x 1 + y + y 2 + y 3 + . . . + y i + . . . 0 x 2, y = 1 – x ex 1 + x /1! + x 2/2! + x 3/3! + . . . + x i /i ! + . . . ln x –y – y 2/2 – y 3/3 – y 4/4 – . . . – y i /i – . . . 0 x ? 2, y = 1 – x ln x 2 [z + z 3/3 + z 5/5 + . . . + z 2i+1/(2i + 1) + . . . ] x 0, z = x–1x+1 sin x x – x 3/3! + x 5/5! – x 7/7! + . . . + (–1)i x2i+1/(2i + 1)! + . . . cos x 1 – x 2/2! + x 4/4! – x 6/6! + . . . + (–1)i x2i /(2i )! + . . . tan–1 x x – x 3/3 + x 5/5 – x 7/7 + . . . + (–1)i x2i+1/(2i + 1) + . . . –1 x 1 sinh x x + x 3/3! + x 5/5! + x 7/7! + . . . + x2i+1/(2i + 1)! + . . . cosh x 1 + x 2/2! + x 4/4! + x 6/6! + . . . + x2i /(2i )! + . . . tanh–1x x + x 3/3 + x 5/5 + x 7/7 + . . . + x2i+1/(2i + 1) + . . . –1 x 1 ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– May 2015 Computer Arithmetic, Function Evaluat