On page 75 in Sutskever's thesis http://www.cs.utoronto.ca/~ilya/pubs/ilya_sutskever_phd_thesis.pdf
In equation (7.5) setting $a_0=1$,
$a_{t+1} = (1+\sqrt{4 a_t^2 + 1})/2 $
The author said, "to understand the sequence $a_t$ we note that the function $x \rightarrow (1+ \sqrt{4x^2+1})/2$ quickly approaches $x \rightarrow x + 0.5 $ from below as $x \rightarrow \infty$,"
then he claims
"so $a_t \approx (t+4)/2 $ for large $t$ "
I couldn't figure out how he made the deduction here. It is not at all obvious to me how one might even guess the express. If we brute-force it we could probably guess $t/2$ but the $t+4$ term seems to come from nowhere at the large number limit.
