The Nelder-Mead default for optim exists because it is a good method for more than one argument to be optimized in the case when no derivatives are available. When one argument is optimized the Nelder-Mead method throws a warning, which makes for a bad experience given that we always have derivatives (thanks to TMB) and therefore should never be using Nelder-Mead anyways.
From @bbolker.
The Nelder-Mead default for
optim
exists because it is a good method for more than one argument to be optimized in the case when no derivatives are available. When one argument is optimized the Nelder-Mead method throws a warning, which makes for a bad experience given that we always have derivatives (thanks to TMB) and therefore should never be using Nelder-Mead anyways.