splitmerge_LSS_SDDS.hpp File Reference

Locality Sensitive Sampling (LSS) with SDDS Split-Merge sampler implementation. More...

#include "../utils/Sampler.hpp"

Go to the source code of this file.

Classes
class	SplitMerge_LSS_SDDS
	Locality Sensitive Sampling (LSS) with SDDS Split-Merge sampler. More...

Detailed Description

Locality Sensitive Sampling (LSS) with SDDS Split-Merge sampler implementation.

This file contains the implementation of the LSS Split-Merge algorithm with SDDS (Smart-split, Dumb-merge, Dumb-split, Smart-merge), an optimized variant of split-merge sampling that uses locality sensitive sampling for selecting anchor points. The SDDS strategy adaptively uses smart (with sequential allocation) or dumb (simple random) proposals depending on the move type and point similarity, balancing computational efficiency with mixing quality. LSS provides computational advantages for large datasets by leveraging similarity information while maintaining theoretical properties of split-merge algorithms.

Author

Filippo Galli

Date

2025