§3.3 9. Consider the vectors cos(x + α) and sin(x) in C[−π, π]. For what values of α will the two vectors be linearly dependent? Give a graphical interpretation of your answer. Solution: Let us first define a matrix A as: sin(x) cos(x + α) A= cos(x) −sin(x + α) We know from theorem 3.3.3 that these vectors will be linearly independent when det(A) 6= 0. Therefore we need to consider the values of α that will make det(A) = 0. We know: det(A) = −sin(x)sin(x + α) − cos(x)cos(x + α) = −(cos(x)cos(x + α) + sin(x)sin(x + α)) = −cos(x − (x + α)) = −cos(α) Since −cos(α) = 0 when α is an odd multiple of π/2, we will consider those values of α. Clearly when α is an odd multiple of π/2, cos(x) is shifted to the left or right by an odd multiple of π/2. When this occurs cos(x + α) becomes equivalent to sin(x) or −sin(x). sin(x) and sin(x) are clearly linearly dependent becuase the system: c1 sin(x) + c2 sin(x) = 0 has the nontrivial solution c1 = 1 and c2 = −1. Similarly, sin(x) and −sin(x) are linearly dependent becuase the system c1 sin(x) − c2 sin(x) = 0 has the nontrivial solution c1 = 1 and c2 = 1. Thus, both vectors are clearly linearly dependent when α is an odd multiple of π/2 (Note that the graphical interpretation is above). 13. Prove that any nonempty subset of a linearly independent set of vectors {v1 , ..., vn } is also linearly independent. Solution: I will prove the contrapositive. 1 If we have a nonempty subset of a set of vectors {v1 , ..., vk }, k ≤ n, and this subset is linearly dependent, then we have that for some c1 , c2 , ..., ck , c1 v1 + c2 v2 + ... + ck vk = 0 where c1 , c2 , ..., ck do not all equal 0. Now, let’s look at the set of vectors {v1 , ..., vn } from which we have the aforementioned subset. This set has at least the members contained in our subset. Therefore, it has values of c1 , c2 , ..., ck , ..., cn , k ≤ n, other than c1 = c2 = ... = ck = ... = cn = 0 that will make the equation c1 v1 + c2 v2 + ... + ck vk + ... + cn vn = 0 Since we have proven the contrapositive of our original statement, we have that our original statement is also proven, which is our desired result. 14. Let A be an m × n matrix. Show that if A has linearly independent column vectors, then N (A) = {0} Solution: I will call the column vectors of A: a1 , a2 , . . . an . Since the column vectors of A are linearly independent we know: c1 a1 + c2 a2 + . . . + cn an = 0 only has the solution c1 = c2 = . . . = cn = 0 by definition. Recall that N (A) = {x Rn : Ax = 0}. This means that for any x in N (A) we must have Ax = 0. Let us represent any x Rn by x = (c1 , c2 , . . . cn )T where ci is a scalar. We already know that: c1 a1 + c2 a2 + . . . + cn an = 0 implies c1 = c2 = . . . = cn = 0. Then, by the hint given in the book we know that: Ax = c1 a1 + c2 a2 + . . . + cn an . Thus, we know that Ax = 0 implies that x = (0, 0, . . . 0)T which clealry means that N (A) = {0}. 15. Let {x1 , ..., xk } be linearly independent vectors in Rn , and let A be a nonsingular n × n matrix. Define yi = Axi for i = 1, ..., k. Show that {y1 , ..., yk } are linearly independent. Solution: We have that {x1 , ..., xk } are linearly independent vectors, so we then know that 2 the n×n matrix x1 ... xk is invertible. We also have that A is nonsingular, so we know that A is invertible. Therefore, the product A x1 ... xk is invertible as well. Simplifying the product, we obtain Ax1 ... Axk which is equal to the matrix y1 ... yk . Since we have that all yi = Axi for i = 1, ..., k can be written as an n × n invertible matrix, we also have that the column vectors {y1 , ..., yk } are linearly independent, our desired result. 3