The pictures below show evolution of pouzyry in the example mentioned in the first paper:

The original function f(x) is defined in a 7-dimensional hypercube, and is non-zero in a strip of radius 0.1 around the diagonal.

The green dots are the non-zero sample points, distributed randomly.

Only 200 sample points with non-zero f(x) were retained (N=200).

The minimization was run with 10 pouzyry (P=10) and R=0.1.

R is an upper bound of the resulting pouzyry *after* the minimization.

The red dots are centers, and the blue circles are effective radii, of the pouzyry.

The yellow dots show random sample points generated according to the probability distribution corresponding to the current configuration of pouzyry.

The pictures show a 2-dim projection of successive steps of the
minimization procedure.

One can see how intelligently the pouzyry find their positions.