prior.fn — prior.fn • PEcAn.priors

Prior fitting function for optimization

prior.fn(parms, x, alpha, distn, central.tendency = NULL, trait = NULL)

Arguments

parms: target for optimization
x: vector with c(lcl, ucl, ct) lcl / ucl = confidence limits, ct = entral tendency
alpha: quantile at which lcl/ucl are estimated (e.g. for a 95% CI, alpha = 0.5)
distn: named distribution, one of 'lnorm', 'gamma', 'weibull', 'beta'; support for other distributions not currently implemented
central.tendency: one of 'mode', 'median', and 'mean'
trait: name of trait, can be used for exceptions (currently used for trait == 'q')

Value

parms

Details

This function is used within `DEoptim` to parameterize a distribution to the central tendency and confidence interval of a parameter. This function is not very robust; currently it needs to be tweaked when distributions require starting values (e.g. beta, f)

Author

David LeBauer

Examples

if (FALSE) { # \dontrun{
  DEoptim(fn = prior.fn,
          lower = c(0, 0),
          upper = c(1000, 1000),
          x=c(2, 6, 3.3),
          alpha = 0.05,
          distn = 'lnorm')$optim$bestmem
} # }