Difference between `mle` and `optimize` functions in R I am trying some examples in R about maximum likelihood estimation, and it seems that we can use both the optimize function of the "stats" package and the mle function of the "stats4" package for this task?
Is it fair to say that optimize is a more generic version of mle?
One thing I notice is the the mle function requires an initial seed value (intelligent guesses), but the optimize function does not. Why is this?
 A: The docs for mle say:

The optim optimizer is used to find the minimum of the negative log-likelihood. An approximate covariance matrix for the parameters is obtained by inverting the Hessian matrix at the optimum.

So, in fact, mle uses optim internally.  Looks like mle does some after the fact work to give the caller back data in an object that has some convenience methods attached for tasks specific to maximum likelihood estimation.
As far as I know, optim also requires initial values.  From the optim docs:

par:
  Initial values for the parameters to be optimized over.

is the first argument to optim.
The optimize function, on the other hand, can only handle one variable optimization problems, so is considerably less general than optim.  It does not use a gradient based approach

The method used is a combination of golden section search and successive parabolic interpolation

Here's a reference for that method.  Since it is not gradient based, it does not need a starting point, instead working by "chopping up" the interval into smaller and smaller pieces, each of which bound a local extrema.
