Skip to content

8.1-final-2001-spring-conflict

Final Exam (Conflict), 2001 (spring-conflict) · Section 8.1: The Idea of a Linear Transformation Source: finalexam_1.pdf, p. 75

Problem

5. (20pts.) Let \(U\) and \(V\) be vector spaces.

(a) Define the kernel and image of a linear transformation \(T : U \to V\).

(b) Show that the kernel of \(T\) is a subspace of \(U\).

(c) Let \(T\) be a linear transformation from \(U\) to \(V\) and let \(u_1, \dots, u_k\) form a basis of Ker \(T\). The following steps help you to show that if \(u_1, \dots, u_k, u_{k+1}, \dots, u_n\) form a basis of \(U\), then \(Tu_{k+1}, \dots, Tu_n\) form a basis of Im \(T\). So assume that \(u_1, \dots, u_k, u_{k+1}, \dots, u_n\) are a basis of \(U\).

i. Argue that \(Tu_{k+1}, \dots, Tu_n\) are elements of Im \(T\).

ii. Show that any element of Im \(T\) can be expressed as a linear combination of \(Tu_{k+1}, \dots, Tu_n\).

iii. Show that \(Tu_{k+1}, \dots, Tu_n\) are linearly independent.

(d) Deduce a formula relating the dimensions of \(U\), Ker \(T\) and Im \(T\).

Solution (a) \(\text{Ker } T = \{u \in U : Tu = 0\}\), the set of vectors that \(T\) maps to zero. \(\text{Im } T = \{T(u) : u \in U\}\), the set of all outputs; it is a subspace of \(V\). (b) \(T(0) = 0\), so \(0 \in \text{Ker } T\). If \(u, w \in \text{Ker } T\), then \(T(u + w) = Tu + Tw = 0 + 0 = 0\), and for any scalar \(c\), \(T(cu) = c\,Tu = 0\). So Ker \(T\) is closed under addition and scalar multiplication: a subspace of \(U\). (c) i. Each \(Tu_j\) is an image \(T(u_j)\) of an element of \(U\), hence lies in Im \(T\). ii. Any element of Im \(T\) is \(Tu\) for some \(u = c_1u_1 + \dots + c_nu_n\). Then \(Tu = c_1Tu_1 + \dots + c_kTu_k + c_{k+1}Tu_{k+1} + \dots + c_nTu_n = c_{k+1}Tu_{k+1} + \dots + c_nTu_n\), because \(Tu_1 = \dots = Tu_k = 0\). So \(Tu_{k+1}, \dots, Tu_n\) span Im \(T\). iii. If \(d_{k+1}Tu_{k+1} + \dots + d_nTu_n = 0\), then \(T(d_{k+1}u_{k+1} + \dots + d_nu_n) = 0\), so \(d_{k+1}u_{k+1} + \dots + d_nu_n \in \text{Ker } T\), hence it is a combination of \(u_1, \dots, u_k\): \(d_{k+1}u_{k+1} + \dots + d_nu_n - e_1u_1 - \dots - e_ku_k = 0\). Since \(u_1, \dots, u_n\) is a basis of \(U\), all coefficients are zero; in particular \(d_{k+1} = \dots = d_n = 0\). So \(Tu_{k+1}, \dots, Tu_n\) are linearly independent. (d) From (c), \(\dim \text{Im } T = n - k = \dim U - \dim \text{Ker } T\), i.e. \[ \dim U = \dim \text{Ker } T + \dim \text{Im } T \] (the rank-nullity theorem).

Keywords: kernel, image, linear transformation, rank-nullity theorem