On a problem in combinatorial geometry (Q5905326)

Let \(f(S)\) be the ratio of the area of a largest (nondegenerate) triangle determined by the points of a finite set \(S\) to that of a smallest, and \(f(n)=\inf_sf(S)\), where the infimum is taken over all planar, noncollinear sets \(S\) of cardinality \(n\). It is known that \(f(3)=f(4)=1\), and \(f(5)=(\sqrt5+1)/2\); it is clear (by taking \(S_0\) to be a set of \(n\) points equally spaced and evently distributed on two parallel lines) that \(f(n)\leq[\frac{1}{2}(n-1)]\). Using the interesting theorem of \(E\). Sas that the ratio \(\rho\) of the area of a convex set \(C\) to that of a triangle contained in \(C\) having maximal area satisfies the inequality \(\rho\leq 4\pi/3\sqrt3< 2.4184\), the authors prove that \(f(n)=[(n-1)/2]\) for \(n> 37\), and that, moreover, for even \(n\geq38\), if \(f(S)=f(n)\) then \(S\) is affinely equivalent to the set \(S_0\) mentioned above. It is conjectured that \(f(n)=[(n-1)/2]\) also for \(5< n\leq 37\), but in this range other extremal configurations besides \(S_0\) are possible. Several other excellent unsolved problems are stated.