loglik.Mtb: Negative log-likelihood for models M0, Mt, Mb, Mtb
In david-borchers/mt5751a: Functions for the first part of MT5751

Description Usage Arguments Details Value

Calculates the negative log-likelihood for models M0, Mt, Mb, Mtb.

1	loglik.Mtb(pars, chmat, prevcap = NULL, timecov = NULL)

`pars`	Parameter vector with the correct number of parameters in the correct order (see Details).
`chmat`	Binary capture history matrix, with one row per individual detected and one column per capture occasion.
`prevcap`	Binary matrix of same size as `chmat`, with 0 indicating that the individual has not been captured by that occasion, and 1 indicating that it has. Used by the function to decide what the model is, so if it is not null, the model is taken to be Mb (if `timecov` is NULL) or Mtb (if `timecov` is not NULL).
`timecov`	Matrix of same size as `chmat` indicating the time associated with the capture occasion. At present it is treated as a factor.

The function decides which model to use on the basis of which of prevcap and timecov are null. It assumes that pars has the appropriate number of parameters for the model, and in the appropriate order, as follows:

'M0' prevcap and timecov both NULL: pars must be (log(N),qlogis(p)).
'Mt' prevcap NULL, timecov not: pars must be (log(N),qlogis(p1)...qlogis(pK)), where there are K occasions.
'Mb' prevcap not NUL, timecov is NULL: pars must be (log(N),qlogis(p),log(b)), where log(b) is the change in log-odds due to previous capture.
'Mtb' Neither prevcap nor timecov are NULL: pars must be (log(N),qlogis(p1)...qlogis(pK),log(b)), where there are K occasions and log(b) is the change in log-odds due to previous capture.