Julia version of Dirk Nuyens' Magic Point Shop.
See the Tutorial for instructions on how to use this package.
QMCGenerators.jl
implements quasi-random (low discrepancy) sequence generators. Quasi-random points are carefully coordinated in a dependent manner to more evenly fill
This package implements two flavors of quasi-random sequences: lattice rules and digital nets, both in base 2. Independent randomizations may be applied to base sequences via random shifts for lattices and either random digital shifts, random linear matrix scrambles, and/or random Owen scrambles for digital nets.
Notice the gaps and clusters of the above pseudo-random points compared to the more even coverage of quasi-random points. The implemented quasi-random sequences are extensible, meaning you may increase the sample size through powers of two to better fill the space. Above, each plot starts with
Monte Carlo methods approximate the true mean
for
where
The above example approximates
and
where
where
- Kuo, F. Y., & Nuyens, D. (2016). Application of quasi-Monte Carlo methods to elliptic PDEs with random diffusion coefficients: a survey of analysis and implementation. Foundations of Computational Mathematics, 16, 1631-1696.