loglik: Log-likelihood function

In erinclancey/matepref: A Maximum Likelihood Based Method to Estimate the Strength of Mate Preference

Description

This function is the log-likelihood function for the mate preference model described by Clancey, Johnson, Harmon, and Hohenlohe (2021) (see equation 2.6 in manuscript).

Usage

loglik(theta, xp, yp, x, y)

Arguments

theta	Vector of the model parameters (alpha, delta, mx, my, sx, and sy). Note that mx and my are the means of the X and Y traits, respectively, and sx and sy are the standard deviations of the X and Y traits, respectively.
xp	Vector of paired observations of X.
yp	Vector of paired observations of Y.
x	Vector of unpaired observations of X.
y	Vector of unpaired observations of Y.

Value

The value of the log-likelihood function.

Source

Clancey, E., Johnson, T. R., Harmon, L. J., and Hohenlohe, P. A. Estimation of the strength of mate preference from mated pairs observed in the wild. Unpublished manuscript.

Examples

# simulate data
mydata <- simdata(0.5, 0, 50, 52, 1, 1, 100, 50, 50)
# set starting values using unpaired observations for means and standard deviations
theta <- c(0, 0, mean(c(mydata$xp, mydata$x)), mean(c(mydata$yp, mydata$y)), sd(c(mydata$xp, mydata$x)), sd(c(mydata$yp, mydata$y)))
# compute MLEs and Hessian (set fnscale = -1 to maximize rather than minimize)
fit <- optim(theta, loglik, method = "L-BFGS-B", control = list(fnscale = -1),
  lower = c(0, 0, -Inf, -Inf, 0, 0), hessian = TRUE,
  xp = mydata$xp, yp = mydata$yp, x = mydata$x, y = mydata$y)
# estimates and standard errors
data.frame(mle = fit$par, se = sqrt(diag(solve(-fit$hessian))))

erinclancey/matepref documentation built on Dec. 20, 2021, 6:36 a.m.