merge_H_Q: Merge latent state profile matrix (H_star) and Q matrix
In oslerinhealth/rewind: Reconstructing Etiology with Binary Decomposition

Description Usage Arguments Value Examples

This function

merges identical rows in H: pseudo-cluster —> scientific clusters;
merges partner latent states (identical non-zero columns in H); merge the corresponding rows in Q by pmax. Even though identifiability condition may prevent non-identical columns in H, because we specify a maximum latent state dimension m_max, MCMC may produce identical columns in some iterations.

merge_H_Q(
  H_star_redun,
  mylist,
  t,
  Q,
  VERBOSE = FALSE,
  z_pseudo = NULL,
  skip_Q = FALSE
)

`H_star_redun`	A binary matrix of rows `t_max+3` by `m_max`; It may be redundant because it has rows corresponding to empty pseudo-clusters, and if the matrix has at least one non-zero element, the zero columns indicate inactive latent states
`mylist`	Cluster labels; A vector of length `t_max+3` comprised of integers that need not be consecutive.
`t`	The number of non-empty pseudo-clusters
`Q`	The `m_max` by `L` Q matrix
`VERBOSE`	Default to `FALSE`: no print of the merged matrices; Otherwise set to `TRUE`.
`z_pseudo`	Default is `NULL`. A vector of pseudo-cluster indicators to be merged into scientific cluster indicators `z_sci`.
`skip_Q`	Default is `FALSE` - merge the rows of Q that correspond to partner states; Set to `TRUE` if `Q` is given in which case only rows of `H_star_redun` will be merged (keep the zero columns in H_star_redun, and not merging partner states).

A list comprised of two elements named: H_star_merge and Q_merge and z_sci if z_pseudo is provided.

# the third latent state is inactive among non-empty pseudo-clusters;
# the 2nd and 4th latent states are partners.
# merge 1st and 2nd pseudo-clusters.
 H_star_redun <- matrix(c(0,1,0,1,
                          0,1,0,1,
                          1,0,0,0,
                          0,0,0,0,
                          0,0,0,0,
                          0,0,0,0,
                          0,0,0,0),nrow=7,ncol=4,byrow=TRUE)
mylist <- c(1,2,3,5,0,0,0)
t <- 4
Q <- simulate_Q(4,100,p=0.1)
z_pseudo <- c(1,1,3,5,2,2,2,1,3,3,5,1,2,2,1)
# Q is unknown and sampled
merge_H_Q(H_star_redun,mylist,t,Q,TRUE,z_pseudo)
# Q is known:
merge_H_Q(H_star_redun,mylist,t,Q,TRUE,z_pseudo,TRUE)