Python Generalized Halton module’s documentation

class ghalton.GeneralizedHalton(dim[, seed] | config)

Creates a generalized Halton generator. It takes at least one argument, either the dimensionality, or a configuration. When the dimensionality is given, an optional argument can be used to seed for the random permutations created. The configuration is a series of permutations each of n_i numbers, where n_i is the n_i’th prime number. In that last case, the dimensionality is infered from the number of sublists given.

Parameters:
  • dim (integer) – Dimensionality of the points to create
  • seed (integer) – Seed to create the permutations
  • config (list of lists of integers) – List of permutations to scramble the halton digits
get(n)

Retreive the n next points from this sequence. Each point is a list containing each value for each coordinates and the points are returned in a list of n elements even if n is 1.

reset()

Reset the generator to its initial state, i.e. before it generated the first point.

seed([config])

Seed the generator with a new seed or configuration. Seeding a generator automatically call reset().

Parameters:config (None, integer, or list of lists of integers) – The config to seed the sequencer
class ghalton.Halton(dim)

Creates a Halton generator of dimensionality dim. This is similar to creating a GeneralizedHalton sequence with the identity permutations.

get(n)

Retreive the n next points from this sequence. Each point is a list containing each value for each coordinates and the points are returned in a list of n elements even if n is 1.

reset()

Reset the generator to its initial state, i.e. before it generated the first point.

ghalton.EA_PERMS

Permutations described in [DeRainville2012] for the 100 first dimensions of the Generalized Halton sequence.

ghalton.PRIMES

Prime numbers lower than 10000.

[DeRainville2012]F.-M. De Rainville, C. Gagné, O. Teytaud, D. Laurendeau. Evolutionary optimization of low-discrepancy sequences. ACM Trans. Model. Comput. Simul., 22(2):1-25, 2012.