Skip to content

3.4-final-2012-strang-b

Final Exam, 2012 (strang) · Section 3.4: Independence, Basis, and Dimension Source: finalexam_2.pdf, p. 85

Problem

8 (16 pts.) In this TRUE/FALSE problem, you should circle your answer to each question.

(a) Suppose you have 101 vectors \(v_1, v_2, \dots, v_{101} \in \mathbb{R}^{100}\).

  • Each \(v_i\) is a combination of the other 100 vectors: TRUE FALSE
  • Three of the \(v_i\)'s are in the same 2-dimensional plane: TRUE FALSE

(b) Suppose a matrix \(A\) has repeated eigenvalues 7, 7, 7, so \(\det(A - \lambda I) = (7 - \lambda)^3\).

  • Then \(A\) certainly cannot be diagonalized (\(A = S\Lambda S^{-1}\)): TRUE FALSE
  • The Jordan form of \(A\) must be \(J = \begin{bmatrix} 7 & 1 & 0 \\ 0 & 7 & 1 \\ 0 & 0 & 7 \end{bmatrix}\): TRUE FALSE

(c) Suppose \(A\) and \(B\) are \(3 \times 5\).

  • Then \(\mathrm{rank}(A + B) \leq \mathrm{rank}(A) + \mathrm{rank}(B)\): TRUE FALSE

(d) Suppose \(A\) and \(B\) are \(4 \times 4\).

  • Then \(\det(A + B) \leq \det(A) + \det(B)\): TRUE FALSE

(e) Suppose \(u\) and \(v\) are orthonormal, and call the vector \(b = 3u + v\). Take \(V\) to be the line of all multiples of \(u + v\).

  • The orthogonal projection of \(b\) onto \(V\) is \(2u + 2v\): TRUE FALSE

(f) Consider the transformation \(T(x) = \int_{-x}^{x} f(t)dt\), for a fixed function \(f\). The input is \(x\), the output is \(T(x)\).

  • Then \(T\) is always a linear transformation: TRUE FALSE
Solution (a) FALSE and FALSE. The 101 vectors in \(\mathbb{R}^{100}\) are linearly dependent, so at least one \(v_i\) is a combination of the others — but not necessarily each one: take \(v_1 = e_1, v_2 = e_1, v_3 = e_2, \dots, v_{101} = e_{100}\); then \(e_2\) is not a combination of the other 100 vectors, whose span misses the \(e_2\) direction. Also, three of the \(v_i\)'s need not lie in a common 2-dimensional plane: for example take the 100 basis vectors \(e_1, \dots, e_{100}\) plus their sum \(e_1 + \cdots + e_{100}\); any three of these are independent, so no three lie in a 2-dimensional plane. (b) FALSE and FALSE. \(A = 7I\) has eigenvalues 7, 7, 7 and is diagonalized by any \(S\), so the repeated eigenvalue does not force non-diagonalizability. The Jordan form could be \(J = \mathrm{diag}(7, 7, 7)\) (or one block \(J_3(7)\) with \(S = I\) for the given matrix \(J\)), so it need not be the single Jordan block. (c) TRUE. Every column of \(A + B\) is a combination of columns of \(A\) and \(B\), so \(C(A + B) \subseteq C(A) + C(B)\), and \[ \mathrm{rank}(A + B) = \dim C(A + B) \le \dim(C(A) + C(B)) \le \mathrm{rank}(A) + \mathrm{rank}(B). \] (d) FALSE. The determinant is not subadditive: with \(A = \mathrm{diag}(2, 1, 1, 1)\) and \(B = \mathrm{diag}(0, 1, 1, 1)\), \[ \det(A + B) = \det(\mathrm{diag}(2, 2, 2, 2)) = 16 > 3 = \det(A) + \det(B). \] (e) TRUE. The projection of \(b\) onto the line through \(u + v\) is \[ \frac{(3u + v)\cdot(u + v)}{\|u + v\|^2}\,(u + v) = \frac{3 + 1}{2}\,(u + v) = 2u + 2v, \] using \(u \cdot v = 0\) and \(\|u\| = \|v\| = 1\). (f) FALSE. \(T\) is not linear in general: for \(f(t) = t^2\), \[ T(x) = \int_{-x}^{x} t^2\,dt = \frac{2x^3}{3}, \] which is not linear in \(x\). (For some \(f\) — e.g. constants — \(T\) happens to be linear, but not always.)

Keywords: true/false, linear dependence, rank, projection