1 MATH 221 - October 9, 2015 Sample Test II. “No one can make you feel inferior without your consent.” Eleanor Roosevelt 1 1 0 −5 −4 1. Let A = 0 1 3 is B = 6 4 2 3 4 −2 −1 x1 the unique solution to A x2 = ~b = x3 3 −3 . Find AB and write down A−1 . Find 1 1 0 . What is B −1 ? Find B −1~x. 1 2 −1 3 2. Find the inverse of A = 4 0 2 by row reductions [ A | I ] ∼ [ I | A−1 ]. In the 0 −1 3 1 1 answer (page 256) to problem 3.6(3) (page 96) replace by . 4 2 " 3. The homogeneous solution to the system A~x = 1 2 3 2 4 7 # " # x1 1 is given by x2 = 3 x3 −2 ~xh = x2 1 where the x2 is arbitrary. 0 (a) Find a basis for CA . (b) What is the dimRA ? (c) What is rank(A)? (d) Find N (A). (see ~xh ) (e) What is dimN (A)? (f) What is dimN (AT )? (g) Are there any vectors ~b ∈ R2 so that the system A~x = ~b has no solution? " 4. Change the coefficient matrix A in #3 to A = # 1 2 3 . The homogeneous solution 2 4 6 −2 −3 reads ~xh = x2 1 + x3 0 , x2 and x3 are arbitrary. Answer (a)-(g) from #3. 0 1 1 5. Suppose that ~v1 = 2 1 −2 1 sume that Aw ~1 = A 3 0 (i) (a) The size of A is of CA 2 and ~v2 = 1 and that CA = span {~v1 , ~v2 }. Further, as−1 1 0 −5 = Aw ~2 = A ~ 1, w ~ 2 }. 0 = 0 and N (A) = span {w 0 3 . (b) The dimension of N (A) is and the dimension . (c) A harder arithmetic problem: # of columns of A = 2 . True or False: A~x = ~b is always solvable. + = dimN (A)+ dimCA 1 (ii) Is it possible to add ~v3 = 1 ∈ CA and leave everything else as stated? 0 0 (iii) Same as part (b) for ~v3 = 1 ∈ CA . 1 For (ii) and (iii), it is relevant to know whether the set V = {~v1 , ~v2 , ~v3 } is a linearly dependent or independent set of vectors. Answers 1. A−1 = B, B −1 = A and B −1~x = ~b. see Theorem (3.12) pages 77. 2. see page 256. (" 3. (a) CA = span 1 2 # " , 3 7 span #) (" = span 1 0 # " , 0 1 #) , (b), (c) dimRA = 2 = rankA, −2 (e) dimN (A) = 1 and (f) dimN (AT ) = 0 (g) No 1 0 2 since dim CA = 2 = dimR the problem is always solvable. (d) N (A) = ~xh = (" 4. (a) CA = span 1 2 #) (b), (c) dimRA = 1 = rankA, (d) N (A) = span (" T T (e) so dimN (A) = 2 and (f) dimN (A ) = 1, note that N (A ) = span " (g) If A~x = ~b has no solution unless ~b is a multiple of of CA −2 1 #) # 1 . 2 5. (i) The size of A is 3 × 4 . (b) The dimension of N (A) is 2 and the dimension 2 . (c) # of columns of A = 4 = dimN (A)+ dimCA 2 + 2 . A~x = ~b is not always solvable since dimN (AT ) = 2. (ii) Since ~v3 = (1/3)~v1 + (1/3)~v2 , the answers in (i) are unchanged. (iii) The set {~v1 , ~v2 , ~v3 } is linearly independent so there is one of two possibilities: First: the scenario is impossible. The nullspace of A is “too large” due to the Dimension Theorem (never underestimate a counting theorem) or Second: one could remove one vector from the nullspace of A, say N (A) = span {w ~ 1} then all is well. The dimension of N (A) is 1 and the dimension of CA 3 . The # of columns of A = 4 = dimN (A)+ dimCA 1 + 3 . A~x = ~b is solvable since dim CA = 3 so that dimN (AT ) = 0. −2 −3 1 , 0 0 1