“…To accomplish this, an overlap penalty function was constructed that varied from a cost penalty for points deep within ellipsoids to a cost benefit for points at the edge of their containing ellipsoids. Given a point {i, j, k}, an ellipsoid specified by the semiaxes {a, b, c}, and an ellipsoid position (centroid) {x, y, z}, then an ellipsoid function (E) is specified as follows: [5] The overlap between a given ellipsoid and its neighbors is evaluated in terms of a cost of ␣ (by analogy with energy), an overlap encouragement factor (), a zero penalty ( ), and the ellipsoid function already defined, as follows: [6] In these simulations, ϭ 1.0, ϭ 0.95, and the cost is equal to zero when E ϭ and equal to Ϫ when E ϭ 1. As run in these simulations, ␣ varies quadratically from a cost penalty of almost Ϫ10 at an E of zero to a cost benefit of ϩ1 at an E of 1.…”