So, I want to minimize L_1 plus n over L_1. And I get to choose L_1. Now, I could
differentiate this, set it to zero, and go crazy. Or, I could realize that, I mean, that's
not hard. But, that's a little bit too fancy for me. So, I could say, well, this is clearly
best when L_1 is small. And this is clearly best when L_1 is large. So, there's a
trade-off there. And, the trade-off will be roughly minimized up to constant factors
when these two terms are equal. That's when I have pretty good balance between
the two ends of the trade-off. So, this is up to constant factors. I can let L_1 equal n
over L_1, OK, because at most I'm losing a factor of two there when they happen to
be equal.