1 (a) Solve the quadratic equation
by evaluating the quadratic formula using three-digit decimal arithmetic and
rounding. The exact roots rounded to 6 digits are 0.0399675 and 12.3600.
(b) You should observe loss of significance in one of your computed zeros of
part (a). If we consider the 6-digit answers provided as the “true” ones, what is
this solution’s relative error? Following Definition 2.2 in Kharab & Guenther
(p. 33), to how many significant digits does your computed root approximate
the true one? Explain what aspect of the coefficients in the quadratic equation
to be solved led to this loss of significance.
(c) Show that the roots r1, r2 of a quadratic equation ax^2+bx+c = 0 satisfy r1r2 = c/a.
(d) While there is loss of significance in one solution of the quadratic
equation in (a),
it does not occur in the other. Write an Octave/Matlab function which accepts
as a vector [ a b c] the coefficients of a quadratic function f (x) = ax^2 + bx + c,
computes the root that has larger absolute value, and then obtains the other
root using the formula in part (c). You may assume that the roots are distinct
real numbers—i.e., that b^2 - 4ac > 0. (Hand in a printout of your function.)
(e) As in part (a), use 3-digit decimal arithmetic to solve the quadratic equation
whose exact roots rounded to 6 digits are 6.00000 and 6.40000. You should
again observe loss of significance. Does our method in part (d) for dealing with
loss of significance help here? If not, what aspect of this situation makes it
2.2.5 If x1, x2 are approximations to X1, X2 with errors show that the relative error
of the product X1X2 is approximately the sum
2 In class we showed that, under suitable assumptions on f and h, we have
where lies between x and x + h, and lies between x and x - h.
(a) What, exactly, are these assumptions on f and h?
(b) Supposing that f''' is suitably continuous, show that the truncation
may be replaced by a single term (Hint.
Use the IVT.) In what interval could we assume this new number resides?
9.1.3 Derive the following approximate formulas.
9.1.3 Use Taylor’s series to derive the following approximation formul for
the first derivative
of f .
9.1.9 Show that the approximation formula in Exercise 9.1.8 has an error of O(h^3).