Penny-packing and two-dimensional codes (Q1262539): Difference between revisions

Let \(P_ 1,...,P_ n\) be the centres of n plane circular discs with diameter d, and let \(\bar P=n^{-1}\Sigma P_ i\) be their centroid. The problem is to choose points \(P_ i\) \((i=1,...,n)\) so that \((1)\quad \| P_ i-P_ j\| \geq d\) \((i,j=1,...,n\), \(i\neq j)\) \((P_ 1,...,P_ n\) is called an n-point packing) and so that (2) the second moment \(U=d^{-2}\Sigma \| P_ i-\bar P\|^ 2\) is minimal, where \(\| \|\) is the Euclidean norm. These packings are also minimal- energy two-dimensional codes. The authors develop and discuss three algorithms for the construction of good packings. One of them, which proceeds by adding one point at a time according to a minimum-principle, appears to produce optimal packings for infinitely many values of n. A table is given of the best packings known for \(n\leq 500\). Several conjectures are proposed; e.g., it is conjectured that for \(n\neq 4\) every optimal packing is part of a hexagonal lattice.