In this question, we see a proof that the largest eigenvalue of a stochastic matrix is equal to 1:
Proof that the largest eigenvalue of a stochastic matrix is 1
However, I think I've found a proof that every eigenvalue of a stochastic matrix is equal to 1. Can you tell me where my proof is wrong?
Proof: Suppose ${\bf r}$ is an eigenvector of the column stochastic matrix $M$ (i.e. $M{\bf r} = \lambda {\bf r}$ for some $\lambda$), and assume without loss of generality that the entries of ${\bf r}$ sum to $1$. Then $$M{\bf r} = \begin{bmatrix} M_{11}\\ M_{21}\\ \vdots\\ M_{n1} \end{bmatrix} r_1 + \begin{bmatrix} M_{12}\\ M_{22}\\ \vdots\\ M_{n2} \end{bmatrix} r_2 + \dots + \begin{bmatrix} M_{1n}\\ M_{2n}\\ \vdots\\ M_{nn} \end{bmatrix} r_n$$
Since $M$ is column stochastic, each column must sum to $1$, so the sum of the entries in $M {\bf r}$ is just $1 \cdot r_1 + 1 \cdot r_2 + \dots + 1 \cdot r_n = 1$. Therefore, $\lambda$ must be $1$, and $M$ can only have one eigenvalue.
Thanks!