3 ms·
I think that tends to produce a distribution clustered at the poles. At high latitudes, the impact of the longitude is much smaller than it would be around the
by cs_ 9y ago
I think that tends to produce a distribution clustered at the poles. At high latitudes, the impact of the longitude is much smaller than it would be around the equator. Mathworld has a nice write up that visualizes the effect and details more of the reasoning for the Gaussian method.
[0] http://mathworld.wolfram.com/SpherePointPicking.html http://mathworld.wolfram.com/SpherePointPicking.html
- theoh 9y agoYes, it is literally analogous to choosing points from the set of intersections of lines of integral latitude and longitude, which is not what you want. The specialness of the Gaussian has been mentioned here before: https://news.ycombinator.com/item?id=9446126 https://news.ycombinator.com/item?id=9446126