Skip to content

QMC Optimizer

Introduction

Quasi-Monte Carlo (QMC) Optimizer uses low-discrepancy sequences instead of random sampling. This provides better coverage of the parameter space compared to pure random sampling, often leading to faster convergence.

Usage Example

When to Use QMC Optimizer

Best for: - Better space coverage than random search - Integration and sampling problems - When you want deterministic "random" sequences - High-dimensional parameter spaces

Parameters: - qmc_type: Type of QMC sequence ("sobol", "halton") - scramble: Whether to scramble the sequence

QMC Sequence Types

Sobol Sequences

  • Excellent space-filling properties
  • Good for most optimization problems
  • Default choice for QMC

Halton Sequences

  • Simpler construction
  • Can suffer from correlation in higher dimensions
  • Good for lower-dimensional problems