mixture4_hmc_kernel: Mixture of kernels with HMC synchronous coupling and RWM...
In jeremyhengjm/UnbiasedMultilevel: Unbiased estimators for discretized models

Either HMC kernel or RWM kernel is selected for MCMC iteration with a certain probability

mixture4_hmc_kernel(
  level,
  state_coarse1,
  state_coarse2,
  state_fine1,
  state_fine2,
  identical_coarse,
  identical_fine,
  tuning_coarse,
  tuning_fine,
  proposal_coupling
)

`level`	an integer that controls the probability distribution in a multi-level setting
`state_coarse1`	a list that contains vector of coordinates and log of the density of probability distribution for the first chain at the coarse level: "level - 1"
`state_coarse2`	a list that contains vector of coordinates and log of the density of probability distribution for the second chain at the coarse level: "level - 1"
`state_fine1`	a list that contains vector of coordinates and log of the density of probability distribution for the first chain at the fine level: "level"
`state_fine2`	a list that contains vector of coordinates and log of the density of probability distribution for the second chain at the fine level: "level"
`identical_coarse`	a boolean variable: "True" if chains at the coarse level coincide, "False" otherwise
`identical_fine`	a boolean variable: "True" if chains at the fine level coincide, "False" otherwise
`tuning_coarse`	list of parameters for HMC and RWM kernels (number os steps and stepsize for HMC, standard devation for RWM) for two chains at the coarse level
`tuning_fine`	list of parameters for HMC and RWM kernels (number os steps and stepsize for HMC, standard devation for RWM) for two chains at the fine level

list that contains updated states of the first and the second chains at the coarse level, updated states of the first and the second chains at the fine level, updated value of the flag "identical_coarse", updated value of the flag "identical_fine" and cost of computations

jeremyhengjm/UnbiasedMultilevel documentation built on Dec. 20, 2021, 11:03 p.m.