set_smc_options: Set SMC compute options

Description

Sets the SMC compute options to be used in update_mallows.BayesMallows().

Usage

set_smc_options(
  n_particles = 1000,
  mcmc_steps = 5,
  resampler = c("stratified", "systematic", "residual", "multinomial"),
  latent_sampling_lag = NA_integer_
)

Value

An object of class "SMCOptions".

Arguments

n_particles: Integer specifying the number of particles.
mcmc_steps: Number of MCMC steps to be applied in the resample-move step.
resampler: Character string defining the resampling method to use. One of "stratified", "systematic", "residual", and "multinomial". Defaults to "stratified". While multinomial resampling was used in steinSequentialInferenceMallows2023;textualBayesMallows, stratified, systematic, or residual resampling typically give lower Monte Carlo error Douc2005,Hol2006,Naesseth2019BayesMallows.
latent_sampling_lag: Parameter specifying the number of timesteps to go back when resampling the latent ranks in the move step. See Section 6.2.3 of Kantas2015BayesMallows for details. The \(L\) in their notation corresponds to latent_sampling_lag. See more under Details. Defaults to NA, which means that all latent ranks from previous timesteps are moved. If set to 0, no move step is applied to the latent ranks.

Lag parameter in move step

The parameter latent_sampling_lag corresponds to \(L\) in Kantas2015BayesMallows. Its use in this package is can be explained in terms of Algorithm 12 in steinSequentialInferenceMallows2023BayesMallows. The relevant line of the algorithm is:

for \(j = 1 : M_{t}\) do
M-H step: update \(\tilde{\mathbf{R}}_{j}^{(i)}\) with proposal \(\tilde{\mathbf{R}}_{j}' \sim q(\tilde{\mathbf{R}}_{j}^{(i)} | \mathbf{R}_{j}, \boldsymbol{\rho}_{t}^{(i)}, \alpha_{t}^{(i)})\).
end

Let \(L\) denote the value of latent_sampling_lag. With this parameter, we modify for algorithm so it becomes

for \(j = M_{t-L+1} : M_{t}\) do
M-H step: update \(\tilde{\mathbf{R}}_{j}^{(i)}\) with proposal \(\tilde{\mathbf{R}}_{j}' \sim q(\tilde{\mathbf{R}}_{j}^{(i)} | \mathbf{R}_{j}, \boldsymbol{\rho}_{t}^{(i)}, \alpha_{t}^{(i)})\).
end

This means that with \(L=0\) no move step is performed on any latent ranks, whereas \(L=1\) means that the move step is only applied to the parameters entering at the given timestep. The default, latent_sampling_lag = NA means that \(L=t\) at each timestep, and hence all latent ranks are part of the move step at each timestep.