HW8_solutions

HW8_solutions - Math 167 homework 8 solutions November 27,...

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

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Math 167 homework 8 solutions November 27, 2010 5.3.2: Bernadelli studied a beetle "which lives three years only, and propagates in its third year." They survive the first year with probability 1 , and 2 the second with probability 1 , and then produce six females on the way out: 3 0 0 6 Beetle matrix A = 1 0 0 . 2 1 0 3 0 Show that A3 = I, and follow the distribution of 3000 beetles for six years. Since 0 2 0 A2 = 0 0 3 , 1 0 0 6 we have A3 = A2 A = I. After 6 years the population distribution for the 3000 beetles remains unchanged. This follows from A6 = I. 5.3.4: Suppose each "Gibonacci" number Gk+2 is the average of the two previous numbers Gk+1 and Gk . Then Gk+2 = 1 (Gk+1 + Gk ): 2 Gk+2 = 1 (Gk+1 + Gk ) 2 Gk+1 = Gk+1 is Gk+2 Gk+1 =A Gk+1 Gk . (a) Find the eigenvalues and eigenvectors of A. The matrix A= 1 1 2 1 2 1 0 has eigenvalues 1 = 1 and 2 = - 1 with eigenvectors 2 1 1 respectively. (b) Find the limit as n of the matrices An = Sn S -1 . Since 2 1 1 0 1 1 3 3 An = , (-1)n 1 -1 1 -2 0 2n 3 3 it follows that lim A = n and 1 -2 , 1 1 1 0 0 0 n 1 -2 2 3 1 3 1 3 -1 3 = 2 3 2 3 1 3 1 3 . 2 (c) If G0 = 0 and G1 = 1, show that the Gibonacci numbers approach 3 . Since Gn+1 Gn =A Gn Gn-1 = A2 Gn-1 Gn-2 = = An G1 G0 , it follows from part (b) that lim Gn+1 Gn = lim A n n G1 G0 n = 2 3 2 3 1 3 1 3 1 0 = 2 3 2 3 , and thus limn Gn = 2 . 3 5.3.8: Suppose there is an epidemic in which every month half of those who are well become sick, and a quarter of those who are sick become dead. Find the steady state for the corresponding Markov process 1 1 0 dk+1 dk 4 1 sk+1 = 0 3 2 sk . 4 wk+1 0 0 1 2 wk 2 The eigenvalues for the Markov matrix are are 1 = 1, 2 = 3/4, and 3 = 1/2 with eigenvectors 1 -1 1 x1 = 0 , x2 = 1 , and x3 = -2 , 0 0 1 respectively. Thus the steady state solution is d s = c1 x1 , w where c1 is the first component of the solution c of the linear system 1 -1 1 c1 d0 0 c2 = s0 . x1 x2 x3 c = 1 -2 0 0 1 c3 w0 Solving this system gives c1 = d0 + s0 + w0 , and thus d = d0 + s0 + w0 , s = 0, and w = 0. 5.3.10: Find the limiting values of yk and zk (k ) if yk+1 = .8yk + .3zk zk+1 = .2yk + .7zk y0 = 0 z0 = 5. Also find formulas for yk and zk from Ak = Sk S -1 . .8 .3 . The matrix A has eigenvalues 1 = 1 and 2 = .5 with Let A = .2 .7 eigenvectors -1 3 , and 1 2 respectively. It follows that Ak = and thus 1 5 3+ 2- 1 2k-1 1 2k-1 3- 2+ 3 2k 3 2k , 1 3 3 y y = lim Ak 0 = z0 z k 5 2 2 3 3 0 . = 2 5 0 -1 u (eigenvalues i and -i) 1 0 goes around in a circle: u = (cos t, sin t). Suppose we approximate du/dt by forward, backward, and centered differences F, B, C: 5.3.16: The solution to du/dt = Au = (F) un+1 - un = Aun or un+1 = (I + A)un (this is Euler's method). (B) un+1 - un = Aun+1 or un+1 = (I - A)-1 un (backwards Euler). 1 (C) un+1 - un = 1 A(un+1 + un ) or un+1 = (I - 1 A)-1 (I + 2 A)un . 2 2 Find the eigenvalues of I + A, (I - A)-1 , and (I - 1 A)-1 (I + 1 A). For which 2 2 difference equation does the solution un stay on a circle? (F): The eigenvalues of (I + A) are 1 = 1 + i and 2 = 1 - i. Note that the eigenvalues imply that (I + A) cannot be an orthogonal matrix. The magnitude of un must grow with n, thus the un 's are not on a circle. 1 (B): The eigenvalues of (I - A)-1 are 1 = 2 (1 + i) and 2 = 1 (1 - i). The 2 eigenvalues imply that the magnitude of un is decreasing with n. Thus the un 's are not on a circle. 1 (C): The eigenvalues of (I - 1 A)-1 (I + 2 A) are 1 = .6+.8i and 2 = .6-.8i. 2 Note that the matrix is an orthogonal matrix, thus the magnitude of the un 's is constant. This means that the un 's stay on a circle. 5.3.22: What are the limits as k (the steady states) of the following? .4 .2 .6 .8 Note that the matrix A= k 1 0 , .4 .2 .6 .8 k 0 1 , .4 .2 .6 .8 k . .4 .2 .6 .8 has the eigenvalues 1 = .2 and 2 = 1 with eigenvectors -1 1 respectively. It follows that Ak = 1 4 -1 1 1 3 .2k 0 0 1 4 -3 1 1 1 , and 1 3 , and thus k lim Ak = 1 4 -1 1 1 3 0 0 0 1 1 0 0 1 -3 1 1 1 1 4 1 4 1 3 1 3 , . = 1 4 1 1 3 3 . In particular, k lim Ak = and k lim Ak = 5.4.2: For the matrix A= write the general solution to du/dt = Au, and the specific solution that matches u(0) = (3, 1). What is the steady state as t ? (This is a continuous Markov process; = 0 in a differential equation corresponds to = 1 in a difference equation, since e0t = 1.) The eigenvalues of A are 1 = 0 and 2 = -2 with eigenvectors x1 = 1 1 and x2 = 1 -1 , -1 1 1 -1 respectively. Thus the general solution u(t) is given by u(t) = c1 e1 t x1 + c2 e2 t x2 = c1 + c2 e-2t c1 - c2 e-2t , where c1 , c2 R. If u(0) = (3, 1) is the initial condition, then c1 = 2 and c2 = 1, and so the particular solution is u(t) = The steady state solution is t 2 + e-2t 2 - e-2t 2 2 . lim u(t) = . 5.4.4: If P is a projection matrix, show from the infinite series that eP I + 1.718 P. 5 As a projection matrix, P satisfies the relation P 2 = P , which implies P n = P for all n 2. It follows that eP = I + =I =I =I =I P P2 P3 Pn + + ++ + 1! 2! 3! n! P P P P + + + ++ + 1! 2! 3! n! 1 1 1 1 + + ++ + P + 1! 2! 3! n! 1 2 1 1 13 1n + -1 + 1 + + + ++ + 1! 2! 3! n! + -1 + e1 P I + 1.718 P. P 5.4.6: The higher order equation y + y = 0 can be written as a first-order system by introducing the velocity as y another unknown : d dt y y = y y = y -y . If this is du/dt = Au, what is the 2 by 2 matrix A? Find its eigenvalues and eigenvectors, and compute the solution that starts from y(0) = 2, and y (0) = 0. The matrix is 0 1 . A= -1 0 It has the eigenvalues 1 = i and 2 = -i with eigenvectors x1 = 1 i and x2 = i , 1 respectively. For the initial conditions y(0) = 2 and y (0) = 0, the solution is given by y(t) y (t) = 1 2 1 i i 1 eit 0 0 e-it 1 -i -i 1 2 0 = 2 cos t -2 sin t . 5.9: What happens to the Fibonacci sequence if we go backward in time, and how is F-k related to Fk ? The law Fk+2 = Fk+1 + Fk is still in force, so F-1 = 1. 6 Since the law Fk+2 = Fk+1 + Fk is assumed to hold true for all integers k, the relation Fk+1 Fk 1 1 =A , where A := , Fk-1 1 0 Fk holds true for all integers k. In particular, replacing k by -k and multiplying from the left by A-1 it follows from the above relation that F-k F = A-1 -(k-1) , F-(k+1) F-k This in turn implies that F-(k+1) F-k = A-1 k k = 1, 2, 3, . . . . F0 , F-1 k = 1, 2, 3, . . . , 1+ 5 2 where F0 = 0 and F-1 = 1. Since the matrix A has the eigenvalues 1 = and 2 = 1-2 5 with eigenvectors x1 = 1 1 and x2 = 2 , 1 1 1 respectively, the matrix A-1 has the eigenvalues 1 = 1 and 2 = 2 with the same eigenvectors x1 and x2 . Just as we had done in class for the "forward" Fibonacci recursion, it follows that 1 F-k = (1 )k - (2 )k , 5 Since 1 = - it follows that 1 1- 5 F-k = (-1)k 2 5 k k = 1, 2, 3, . . . . 1+ 5 2 1- 5 2 and 2 = - , k - k 1+ 5 2 1- 5 2 k 1+ 5 1 = (-1)k+1 2 5 - 7 , k = 1, 2, 3, . . . . Comparing this with the formula k 1 1+ 5 Fk = - 2 5 F-k = (-1)k+1 Fk 1- 5 2 k for the Fibonacci numbers, which we had derived in class, it follows that for all k = 0, 1, 2, . . . . , k = 0, 1, 2, . . . , Can you find a time T at which the solution u(T ) is guaranteed to return to the initial value u(0)? The matrix A has eigenvalues 1 = 0, 2 = i 2, and 3 = -i 2. It follows that 1 0 0 0 S -1 , eAt = S 0 ei 2t 0 0 e-i 2t where S is a matrix the columns of which are eigenvectors corresponding to 1 , 2 and 3 . We wish to find a T sothat eAT = eA0 = I. The last relation is satisfied if, and only if, ei 2T = e-i 2T = 1. All T > 0 with this property are given by 2j T = , j = 1, 2, 3, . . . . 2 5.17: (a) Find the eigenvalues and eigenvectors of A = 0 4 . 1 0 4 The eigenvalues of A are 1 = 1 and 2 = -1 with eigenvectors x1 = respectively. 8 4 1 and x2 = 4 -1 , 5.10: Find the general solution to du/dt = Au if 0 -1 0 A= 1 0 -1 . 0 1 0 (b) Solve du/dt = Au starting from u(0) = (100, 100). The solution u(t) is given by u(t) = c1 et x1 + c2 e-t x2 where c1 and c2 are determined by 4 4 1 -1 c1 100 = . c2 100 It follows that c1 = 125/2 and c2 = -75/2, and thus u(t) = 25 2 20et - 12e-t 5et + 3e-t . (c) If v(t) = income to stockbrokers and w(t) = income to client, and they 1 help each other by dv/dt = 4w and dw/dt = 4 v, what does the ratio v/w approach as t ? Since 4c1 et + 4c2 e-t v(t) , = u(t) = c1 et x1 + c2 e-t x2 = c1 et - c2 e-t w(t) it follows that v(t) 4c1 et + 4c2 e-t = lim = 4, t w(t) t c1 et - c2 e-t lim provided that c1 = 0. 9 ...
View Full Document

This note was uploaded on 03/18/2012 for the course MAT MAT 167 taught by Professor Rolandw.freund during the Fall '10 term at UC Davis.

Ask a homework question - tutors are online