1: \begin{proof}

2: The intuition is that $\vct{b}_s$ is close to $\vct{b}$, which is close to $\vct{x}$.  Rigorously,

3: $$

4: \enorm{ \vct{x} - \vct{b}_s }

5: 	\leq \enorm{ \vct{x} - \vct{b} } + \enorm{ \vct{b} - \vct{b}_s }

6: 	\leq 2 \enorm{ \vct{x} - \vct{b} }.

7: $$

8: The second inequality holds because $\vct{b}_s$ is the best $s$-sparse approximation to $\vct{b}$.  In particular, the $s$-sparse vector $\vct{x}$ is a worse approximation.

9: \end{proof}

10: