MQMF (version 0.1.0)

negLLM: negLLM -ve log-normal likelihoods for multiple index time-series

Description

negLLM we have negLL and negLL1 for use when using -ve log-likelihoods to fit surplus production models that only have a single index of relative abundance, but there are many fisheries that have more than one index of relative abundance. negLLM is for those cases that have multiple (M) time-series of indices. It is used in conjunction with simpspmM and spmCE.

Usage

negLLM(pars, funk, logobs, indat, index = "cpue", harvpen = TRUE, ...)

Arguments

pars

the log-transformed parameter starting points. For a surplus production model these are r, K, Binit (if initial depletion is likely, otherwise omit this and it will be set =K inside the function), then as many sigma values as there are time-series of indices; these are the associated standard deviations of the log-normal residuals.

funk

the function that generates the predicted cpue values. for multiple time-series in a SPM use simpspmM

logobs

the log-transformed observed cpue columns in indat, the data needed by funk tranferred inside the ...

indat

the fisheries data used in the analysis

index

the prefix of the columns of each of the indices, defaults to cpue

harvpen

default = TRUE, which sets a penalty1 on each of the implied harvest rates to ensure we do not get harvest rates > 1.0

...

the continuation ellisis to allow the transfer of other arguments required by funk

Value

a single scalar as the -ve log-likelihood of the input data

Examples

# NOT RUN {
 data(twoindex)
 fish <- as.matrix(twoindex)
 pars <- log(c(0.04,155000,0.4,0.3))
 bestSP <- nlm(f=negLLM,p=pars,funk=simpspmM,indat=fish,
             schaefer=TRUE,logobs=log(fish[,c("cpue1","cpue2")]),
             steptol=1e-06,harvpen=TRUE)
 outfit(bestSP,digits=5,title="negLLM example") #optimum solution
 answer <- plotspmmod(bestSP$estimate,indat=fish,
                      plotprod=TRUE,maxy=3.4)
# }