Generic variance reduction in Monte Carlo samplers

2022-08-25 — 2022-08-25

Wherein a survey of Monte Carlo variance reduction methods is presented, and the question of generic applicability is examined through consideration of sample diversity, Rao-Blackwellization, and preliminary references.

Bayes
estimator distribution
Markov processes
Monte Carlo
probabilistic algorithms
probability
Figure 1

Is it meaningful to talk about “generic” Monte Carlo variance reduction strategies? Let’s create this notebook and see if it accretes some references.

For now, there are a couple of links about optimal sample diversity. There are other strategies available, such as Rao-Blackwellization.

1 Sample diversity

2 Rao-Blackwellization

TBD

3 References

Mariet. 2016. “Learning and enforcing diversity with Determinantal Point Processes.”
Roberts, and Rosenthal. 2014. “Minimising MCMC Variance via Diffusion Limits, with an Application to Simulated Tempering.” Annals of Applied Probability.
Sheikh, Phielipp, and Boloni. 2022. “Maximizing Ensemble Diversity in Deep Reinforcement Learning.”