custom_multiType: get the likelihood of a custom model custom_multiType allows...
In cboettig/wrightscape: wrightscape

Description Usage Arguments Details Examples

get the likelihood of a custom model custom_multiType allows the user to indicate which regimes will share which parameters, making all submodels of the global model possible.

1 2	custom_multiType(data, tree, regimes, par, indices, make_bm = NA, ...)

`data`	the trait data
`tree`	the phylogenetic tree in ouch format
`regimes`	the regimes in ouch format
`par`	a list of initial parameters, structured according to indices (see details)
`indices`	a list with vectors alpha_i, sigma_i, and theta_i. Each vector contains the index number of the parameter to be used for that regime – repeated index numbers indicate a parameter that is the same for both regimes. The entries in the vector are in order of the regimes and must be of length n_regimes.
`make_bm`	the regimes which should have their alpha value set to BM. the default value of NA will avoid forcing any to zero
`...`	extra options that are passed to the optimizer routine (optim).

Consider this example: Regimes are numbered 1 to 7. 1-6 follow modelspec global alpha/sigma, indep theta and regime 7 follows modelspec alpha=0, theta, shares a global sigma. Also imagine a parameter list, par <- (alpha, sigma, theta1,2,3,4,5,6,7). Then: indices <- list(alpha_i = rep(1,7), sigma_i = rep(2,7), theta_i = 3:(7+2)) make_bm <- c(1) # makes the 7th alpha value set to zero always

## Define ouch's "hansen" function the custom way
## (equiv to using multiTypeOU withmodel_spec global gobal indep)
require(wrightscape)
require(ouch)
data(bimac)
tree <- with(bimac,ouchtree(node,ancestor,time/max(time),species))

par <- c(1, 1, 3, 3, 3) # init guesses for alpha, sigma, theta_1, theta_2, theta_3
# parameter mapping is created by indices:
indices <- list(alpha_i = c(1,1,1), sigma_i=c(2,2,2), theta_i = c(3,4,5))
ou3 <- custom_multiType(log(bimac[['size']]), tree, bimac[['OU.LP']], par, indices)