MLE: A convenience function to carry out maximum likelihood...
In rettopnivek/mle: Convenience functions for maximum likelihood estimation

A convenience function to carry out maximum likelihood estimation using either optim or nlm.

1
2
3

MLE(dat, mle_fn, start, priors = NULL, gr_fn = NULL, method = NULL,
  control = list(), hessian = T, alpha = 0.9544997, emStop = 20,
  nRep = 1, parNames = NULL)

`dat`	the data to be fitted; can be in a variety of forms as long as it is compatible with the `mle_fn` function that is passed in.
`mle_fn`	a function to compute the sum of the log-likelihoods for a vector of parameters `prm`, the data `dat`, and control parameters `sum` and `priors`.
`start`	either a vector of initial starting values for the parameters, or a function (which takes no parameters) used to generate a dispersed set of starting values.
`gr_fn`	an optional function to compute the gradients for the parameters being estimated. See `optim`.
`method`	the optimization algorithm to use. See `optim`. If `NULL`, a step-by-step approach is used, where viable estimates and standard errors are sought by first trying the 'BFGS' method, then the 'BFGS' method using initial estimates from a 'Nelder-Mead' approach, then the 'nlm' approach, and finally the 'Nelder-Mead' algorithm.
`control`	a named list where options controlling the optimization can be specified. See `optim` and `nlm`. The lower and upper limits for the 'Brent' method can also be specified in this list.
`hessian`	a logical value; if true, the hessian matrix is estimated, allowing the derivation of standard errors and confidence intervals.
`alpha`	the width of the interval for the confidence intervals around parameters. Defaults to +/- two standard deviations.
`emStop`	the number of iterations used when attempting to generate viable starting values.
`nRep`	the number of times to repeat estimation using dispersed starting values. Only applicable when a function to generate starting values is provided.
`parNames`	an optional character vector giving a set of labels for the parameters being estimated.