[ADMB Users] Diverging likelihood and maximum gradient component

[dave fournier]

This commonly occurs when you are fitting a normal distribution and
the model can fit observations with a common variance perfectly.
It is like falling into a black hole (I assume).