Matrix Norm Inequality
Find the smallest positive real number $ C $ for which
\[\left\| \begin{pmatrix} 2 & 3 \\ 0 & -2 \end{pmatrix} v \right\| \le C \|v\|\]for all two-dimensional vectors $ v $.
Note that for a two-dimensional vector $ \mathbf{a}, $ $ \|\mathbf{a}\| $ is the magnitude of $ \mathbf{a} $.
- 1
- 2
- 3
- +
- 4
- 5
- 6
- -
- 7
- 8
- 9
- $\frac{a}{b}$
- .
- 0
- =
- %
- $a^n$
- $a^{\circ}$
- $a_n$
- $\sqrt{}$
- $\sqrt[n]{}$
- $\pi$
- $\ln{}$
- $\log$
- $\theta$
- $\sin{}$
- $\cos{}$
- $\tan{}$
- $($
- $)$
- $[$
- $]$
- $\cap$
- $\cup$
- $,$
- $\infty$
Solution
Let $ v = \begin{pmatrix} x \\ y \end{pmatrix} $. Then
\[\|v\| = \left\| \begin{pmatrix} x \\ y \end{pmatrix} \right\| = \sqrt{x^2 + y^2},\]and
\begin{align*}
\left\| \begin{pmatrix} 2 & 3 \\ 0 & -2 \end{pmatrix} v \right\| &= \left\| \begin{pmatrix} 2 & 3 \\ 0 & -2 \end{pmatrix} \begin{pmatrix} x \\ y \end{pmatrix} \right\| \\
&= \left\| \begin{pmatrix} 2x + 3y \\ -2y \end{pmatrix} \right\| \\
&= \sqrt{(2x + 3y)^2 + (-2y)^2} \\
&= \sqrt{4x^2 + 12xy + 13y^2},
\end{align*}so the given inequality becomes
\[\sqrt{4x^2 + 12xy + 13y^2} \le C \sqrt{x^2 + y^2},\]or
\[\sqrt{\frac{4x^2 + 12xy + 13y^2}{x^2 + y^2}} \le C.\]Thus, we can think of $ C $ as the maximum value of the expression in the left-hand side.
Maximizing the expression in the left-hand side is equivalent to maximizing its square, namely
\[\frac{4x^2 + 12xy + 13y^2}{x^2 + y^2}.\]Let $ k $ be a possible value of this expression, which means the equation
\[\frac{4x^2 + 12xy + 13y^2}{x^2 + y^2} = k\]has a solution in $ x $ and $ y $. We can re-write this equation as
\[(4 - k) x^2 + 12xy + (13 - k) y^2 = 0.\]For this quadratic expression to have a solution in $ x $ and $ y $, its discriminant must be nonnegative. In other words,
\[12^2 - 4 (4 - k)(13 - k) \ge 0,\]or $ 4k^2 - 68k + 64 \le 0 $. This inequality factors as $ 4(k - 1)(k - 16) \le 0 $. The largest value of $ k $ that satisfies this inequality is 16, so the value of $ C $ we seek is $ \sqrt{16} = \boxed{4} $. Note that equality occurs for
\[v = \begin{pmatrix} 1 \\ 2 \end{pmatrix}.\]