A Sobol sequence is a quasi-random sampling method that distributes points evenly across a multidimensional search space. Small gaps between points improve coverage when generating candidate solutions or an initial population. section 7.1
Use a Sobol sequence when:
- You want high diversity with low computational cost during population initialization. section 7.1
- You need broader search-space coverage than pseudo-random sampling typically provides. section 7.1
Generate two-dimensional samples in Python
Install sobol_seq and call sobol_seq.i4_sobol_generate(2, N) to create N points in two dimensions. Plot the resulting coordinates beside pseudo-random, Box–Muller, central-limit-theorem, and Latin-hypercube samples to compare dispersion. section 7.1
!pip install sobol_seq import sobol_seq P_sobel=sobol_seq.i4_sobol_generate(2,N)
Inspect the compact Python listing for generating a two-dimensional Sobol sample. section 7.1
