{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

Math 128A - Spring 2000 - Brown - Final

# Math 128A - Spring 2000 - Brown - Final - FRI 18:01 FAX...

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 03/23/2001 FRI 18:01 FAX 6434330 MOFFITT LIBRARY 001 Math 128A — Final Exam Spring 2000 — Nate Brown 1) (IOpts) Assume M is a constant, N is a function of h and M = N_(h) + h2 + h3 + 5,4 + - - - _ Use Richardson’s~ extrapolation to ﬁnd a function N and constants K3, K4, . . . such that M = N(h) +K3h3 +K4h4+v - - . (Hint: Replace h with 2h.) 2) (10pts) Use a divided difference table to ﬁnd a cubic polynomial Whose graph passes through the points (~1,2), (0,1), (1,0) and (2, 5). 3) (20pts) U‘se Taylor’s theorem to prove that if f E C2[a, b]1 21:0 E [a, b] and h > 0 is such that me + h E [a, b] then there exists .5 E {\$0, \$0 + h] such that mm) = 1/h(f(a:a + h) — ma) 7 (f”(€)/2)h. 4) Let f(:r) = \$2 + Let Qn be the Newton polynomial and H2n+l be the Hermite polynomial which interpolate f at distinct points \$0, . . . , :5“ E [2, 3]. a) (lOpts) Find an n such that |f(a:) w Qn(m)| < 10‘6 for all m E [2, 3]. b) (lOpts) Find an n such that |f(z) — H2n+1(32)| < 10*6 for all m E [27 5) (20pts) Use the ﬁrst Lagrange interpolating polynomial (at the endpoints) to prove the simple Trapezoidal rule. That is, prove that if f E (72 [(1, b] then there exists 5 E [0,, b] such that b —a —a f mumbg (f(a)+f(b))+(512)3f”(€)- (Use the identities ff %dm = f" gate = “Ta and f:{mia)(meb)dm = (1:52.) 130 6)LetA= 010. 203 a)(10pts) Find elementary matrices 241,142 such that AZAIA is upper trian- gular. _I. 1 _\ b)(10pts) Let b: 2 . Find a vector (2 and an upper triangular matrix 3 A such that if 5; is a solution to A a: c then E is also a solution to A E: b. 03/23/2001 FRI 18:02 FAX 6434330 MOFFITT LIBRARY 7) a)(10pts) If T E MAR), —C‘E IR” describe an iterative technique for solving .mx the ﬁxed point type problem 5r. : T E + b)(10pts) Describe the Jacobi method for solving linear systems. 3 2 U c)(5pts) Will the Jacobi method converge if A = f} 2 1 ? 1 1 4 8)(10pts) Assume A E MAR) is invertible and we wish to solve the linear system A In general, the Gauss—Sgdel‘ method applied to this system will not converge. Find a linear system B a): c with the properties that ii) if a: is a solution to the equation B 5:? then 55 is also a solution to the equation A Biz; and ii) the Gauss-Seidel method will converge for the equation B 33:3. 9) Assume y(t) is the unique solution to the initlal value problem :11: 7 f(t,i), ogtgb, y(a):a. Leth=(b—a)/n,ti:a+ih(1Sign), mg :0: and Luz-+1 = 10.; + h.q5(ti,wi, h), 0 S i S n, i 1, for some function Q5. a)(5pts) Deﬁne the local truncation error. b)(5pts) Deﬁne consistency. c)(5pts) Deﬁne convergence. d)(20pts) If f(t,y) : tgeis ~ 37:33], prove that the modiﬁed Euler method will converge. (Recall that for this method wit]. : at; + it/2[f(ti,w1;) + fﬂti_l_1,w,.; + hf(t¢,‘1l}i)).) 10) a)(5pts) Let {ya} be a convergent sequence of numbers with limit y. Deﬁne what it means for {yn} to converge quadratically (Le. with order of convergence 2). b)(25pts) Assume f E lea,b], 79 E [(1, b], Hp) : 0 and _f’(p) gt 0. Prove that there exists 6 > 0 such that for any initial guess .170 E [p — 6, p+6], N ewton’s method will converge (at least) quadratically to 73. 002 ...
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online