Kaba wrote: > It holds that x / f(x) = 1 = y / f(y). Thus they are both in the > sublevel set where f(x) <= 1. The equation above shows that their convex > combination is in the same sublevel set. Multiplying both sides with > beta in R, beta >= 0 shows that this holds for all sublevel sets. > Therefore f is quasiconvex.
The result is true, but I think the proof given here is erroneous. Anyway, to anyone interested, I summarized the contents of this thread to my blog: