mcmc_autocorr_plots

MCMC auto-correlation

Module

mcmc_autocorr_plots

Description

• Fields

  • burn_in percent [0, 1] to burn of the number of samples.

  • functional the currently supported functionals are size and graph score.

  • thinning tells the module to only considering every graph at a given interval length.

  • active should be set to false to deactivate.

  • lags the maximum number of lags after thinning.

This module plots the auto-correlation of a functional of the graphs in a MCMC trajectory.

Fig. 43 Auto-correlation of the scores in trajectory of order MCMC

Fig. 44 Auto-correlation of the scores in trajectory of order MCMC for differnt algorithm seeds and parameter settings.

Example

[
  {
    "id": "omcmc_itsample-bge",
    "burn_in": 0.5,
    "thinning": 1,
    "lags": 50,
    "functional": [
      "score",
      "size"
    ],
    "active": true
  }
]