Magic & McTrivia, great work! It is always fascinating if a design "by construction" also turns out as optimum. Genius at work
There is little to add from my side, as an old Dell PC can not compete with McTrivia's impressive server farm. Nevertheless, I can confirm that Magic's solution is indeed at least a local optimum, and I have not found anything better. However, if Magic's solution is indeed optimum, then I claim that there are infinitely many optimum solutions!
I tried to visualize the constellation of the points and the corresponding polyhedron with the applet on my home page:
http://www.aleakybos.ch/sph_codes.htm : choose PackAnti from the drop down menu (packing optimzation criterion, antipodal = parallel faces) and 16 points.
Point 0 is the North pole (A1 is Magic's description), red lines correspond to the minimum distance between any two points. It can be seen that the points on the second layer (4,5,6,7 or B1-B4) all have minimum distance to the North pole and also to one of their two neighbors on the same layer (4-5 and 6-7). Layer 3, the equatorial plane, is interesting. Of the 6 points, there are two pairs with minimum distance to each other (e.g. 3-10) and to two points in the upper layer (3-5 and 10-7). However, there are also two points (1 and 9 or D1 and D2 in Magic's picture) that are so called rattlers: the circles on these points do not touch any other circle! This means that point 1 (and its antipode 9) can be moved inside a small area (about 1.4% of the spheres' radius) as long as the distance to its neighbours is not below the minimum distance, hence there are indeed infinitely many designs, Although I have not found a design that beats Magic's, I found many that are equally good
However, it makes sense to allocate points 1 (and 9) such that the design is symmetric, so kudos to Magic once more!
No worry, Tom, no need to produce infinitely many designs, one will do :laughing:
Given those symmetries, it is actually possible to express the coordinates analytically. Assuming that Theta is the elevation of the first plane (same notation as Magic) and Beta is the smaller azimuth angle between two points on the first plane (4-5 or B1-B2), it can be shown that
cos(Beta)=(cos(Theta)-cos^2(Theta))/(1-cos^2(Theta))
and
cos(Theta)=sin(Theta)*sin((Theta+Beta)/2)
Solving these two equations numerically yields
Theta=49.62703013°
Beta=66.85149262°
This relates to Magic's design as follows:
Second and 4th layer: Phi=+/- Beta/2 and +/- (180°-Beta/2)
Third layer, points C: Phi=+/- (90°-Theta/2) and +/- (90°+Theta/2)
So much for now.
Alea Kybos