Thought this thread could use a little more rigor
Let \(\displaystyle \epsilon > 0\).
First things first: If f and g are any positive real-valued functions,
We can find a \(\displaystyle N_1 \in \mathbb{N}\) such that \(\displaystyle |x_n-\alpha| < f(\epsilon)\) whenever \(\displaystyle n>N_1\) [1]
We can find a \(\displaystyle N_2 \in \mathbb{N}\) such that \(\displaystyle |y_n-\beta| < g(\epsilon)\) whenever \(\displaystyle n>N_2\) [2]
If \(\displaystyle \alpha = \beta\) then Let \(\displaystyle N = \max\{N_1, N_2\}\) with \(\displaystyle f=g=\epsilon\).
If \(\displaystyle n > N\) then [1] and [2] both hold. Therefore \(\displaystyle |\max\{x_n,y_n\} - \alpha| = |\max\{x_n,y_n\} - \max\{\alpha, \beta\}| < \epsilon\)
Now assume WLOG, \(\displaystyle \alpha > \beta\). You wish to show \(\displaystyle \max\{x_n,y_n\} \to \alpha\) as \(\displaystyle n \to \infty\).
Suppose \(\displaystyle f=g=\min\{\frac{\alpha-\beta}{2}, \epsilon\}\)
Let \(\displaystyle N = \max \{N_1,N_2\}\) as above.
Then from [1] & [2] with our choice above for f and g, we get: (I'll leave the details and algebra to you)
\(\displaystyle y_n < \beta + \frac{\alpha-\beta}{2}= \alpha - \frac{\alpha-\beta}{2} < x_n\)
Thus \(\displaystyle n > N \implies x_n = \max\{x_n,y_n\}\)
The result follows.