MQMF (version 0.1.0)

parasympt: parasympt generates N vectors from a multi-variate normal

Description

parasympt generates N vectors from a multi-variate normal distribution for a surplus production model. This can be used when estimating the uncertainty around an spm fit, or when conducting projections from a model fit while attempting to account for uncertainty. Use of this function requires the mvnnorm package. It could be generalized to suit any model.

Usage

parasympt(bestmod, N)

Arguments

bestmod

the output from nlm containing the optimal parameters in log-space, and the hessian

N

the number of parameter vectors to be sampled from the multi- variate normal defined by the optimal parameters and the inverse of the hessian (the variance covariance matrix).

Value

an N x numpar matrix of parameter vectors

Examples

# NOT RUN {
  data(abdat)
  schf <- FALSE
  param <- log(c(r=0.3,K=11500,Binit=3300,sigma=0.05))
  bestmod <- nlm(f=negLL1,p=param,funk=simpspm,logobs=log(abdat$cpue),
                 indat=abdat,typsize=magnitude(param),iterlim=1000,
                 schaefer=schf,hessian = TRUE)
  out <- spm(bestmod$estimate,indat=abdat,schaefer=schf)
  matpar <- parasympt(bestmod,1000)
  head(matpar,15)
  pairs(matpar)
# }