As an extreme example, for n=2 and using binary floating point, you probably will find the mathematically correct solution, but there's no way to tell numerically whether your answer is off by a fraction of your floating point's epsilon value.
That's probably not important to physicists who want to know an answer, but it is for mathematicians.
What I find very surprising is (from https://en.m.wikipedia.org/wiki/Thomson_problem):
"Numerical solutions for N=8 and 20 are not the regular convex polyhedral configurations of the remaining two Platonic solids, whose faces are square and pentagonal, respectively."
I would like to see the visualizations of the better solutions for n=20 (wikipedia links to the one for n=8) and/or hear a heuristic argument as to how that can happen.