Optimal-Space Filling Design

Generates an optimally-sampled distribution of points in arbitrary dimensions with almost-uniform distribution. Simpler version of MAXIMIN.

Syntax

x = osf(n, p)

x = osf(n, p, iterations)

x = osf(n, p, iterations, cycles)

Description

x = osf(n, p, i, j) returns a distribution of 'n' points in 'p' dimensions with a quasi-uniform distribution seeking an optimally-space sampled design. The algorithm performs 'i' cycles of 'j' MAXIMIN iterations per cycle (maximising the statistical minimum distance between points). This algorithm can be regarded as a simpler and much faster version of the MAXIMIN FileExchange submission, when the design space is cartesian and no seed generation constraints are needed.

For reference,

https://uk.mathworks.com/matlabcentral/fileexchange/108374-uniformly-distributed-points?s_tid=srchtitle

See also MAXIMIN, LHSDESIGN, LHSNORM, UNIFRND.

Examples

x = osf(40, 2);
x = osf(50, 5);
x = osf(70, 3, 5, 400);