If all eigenvalues of a transition matrix have magnitude greater than 1, the dynamic system converges to a steady state.

False

False. Convergence needs every eigenvalue's magnitude below 1: each application scales its eigen-direction by $\lambda$, so $|\lambda| > 1$ makes that component grow without limit and the system diverges. The largest magnitude decides stability, which is why it alone is worth computing.