rbern_constr_v: Random Draws of Bernoulli Variables Subject to Global...

In sumtxt/consilium: Estimating multivariate probit models with partial observability in R

Description

rbern_constr_v draws a vector of Bernoulli variables subject to global constraints such as the number of 1's in each draw.

Usage

rbern_constr_v(x, R, V, z, qweights, pweights, qvote, pvote, adapt, rate, step,
  verbose)

Arguments

x	(required) the vector of probabilities.
R	(required) how many 1's does the the final draw have to have?
V	(required) how many of the first V entries of a draw have the be 1?
z	(required) if z==1 all constrains are obeyed, if z==0 all constraints are not.
qweights	(required) integer numbers. The first weights attached to a draw.
pweights	(required) integer numbers. The second weights attached to a draw.
qvote	(required) the smallest sum of qweights for all entries that are non-zero in a draw before the draw is accepted. If qvote=0 the pweights are ignored.
pvote	(required) the smallest sum of pweights for all entries that are non-zero in a draw before the draw is accepted. If qvote=pvote the pweights are ignored.
adapt	(required) 0 deactivates adaption algorithm 2 in Marbach (2016), 1 enables it.
rate	(required) schedule for the adaption with algorithm 2
step	(required) ε-parameter in algorithm 2
verbose	(required) use 0 for no output; use any positive natural number to report progress at each verbose-th iteration.

Details

The function is part of the Gibbs sampler used for BayesPMP(). The function runs in C++. The function draws a random vector of Bernoulli variables and checks the global constraints. If the draw does not obey the constraints a new draw is obtained until the constraints are met.

Value

vector of draws.

Examples

 ## Not run: 
 # Can be used to simulate vote choices of 3-member committee that has passed a proposal with majority rule
 
 # Vote choice probabilities 
 p <- c(0.2,0.3,0.4)
 R <- 2
 V <- 0 
 z <- 1 
 pweights <- qweights <- rep(1,3)
 qvote <- pvote <- 0
 adapt <- 1
 rate <- 200 
 step <- 0.05
 verbose <- 0
 
 rbern_constr_v(x=p, R=R, V=V, z=z, qweights=qweights, pweights=pweights, qvote=qvote, 
   pvote=pvote, adapt=adapt, rate=rate, step=step, verbose=verbose)
 
## End(Not run)
 

sumtxt/consilium documentation built on May 30, 2019, 8:38 p.m.