Create a control structure for running EM algorithms
em.control(ctrl = NULL, ...)
nplot | how often to plot wages of comparaison to model0 |
---|---|
ncat | how often to log information |
maxiter | maximum number of iterations |
model_var | whether to allow for flexible variance (default is TRUE) |
est_Amb | TBD |
cstr_type | defines the type of constraints on the means for estimation |
cstr_val | TBD |
tol | tolerance for stopping the EM |
tau | posterior likelihood to use instead of computing them using initial parameters |
model0 | model to compare estimation to when plotting |
presp=1e-9 | TBD |
rel_weight | setting a particular weight for stayers versus movers (default is 1) |
est_rho | vector of TRUE/FALSE that states which rho should be estimated |
rho_in_diff | whether to estimate rho in level of in differences |
dprior | Dirichlet prior for proportions (default 1.01) |
nfirms_to_update=10, | number firms to update in probabilistic approach |
proba_include | what terms to include in the liklihood for probabilistic approach (default to = c(1,1,1,1)) |
check_lik | whether to check the likelihood at each updating parameter |
stochastic | whether to use stochastic EM instead of straight EM |
fixb | when TRUE, imposes fixed interactions in different time periods |
fixm | when TRUE, levels are not updated, only variances and proportions |
deps=1e-50 | TBD |
file_backup_prefix | TBD |
sd_floor= | floor imposed on all standard deviations (default is 1e-10) |
posterior_reg | term added to posterior probablities (this is to deal with numerical issues, default is 1e-9) |
textapp | text to show in logging |
sdata_subsample | share of the stayers to use in estimation |
sdata_subredraw | whether to redraw the subsample of stayers |
vdec_sim_size | size to use in simulation |
stayer_weight | weight attributed to stayers in joint estimation |
est_rep= | number of starting values for EM |
est_nbest | number of best starting values to select from using connectedness |