[Developers] profile likelihood

[dave fournier]

It might be a good idea to revisit the profile likelihood code.

1.) extend it to RE models.

2.) change the method to augmented Lagrangian (at least as an option)

I think the  augmented Lagrangian method would be quite a bit faster
than straight penalty methods.


http://www.google.ca/url?sa=t&rct=j&q=augmented%20langrian&source=web&cd=2&sqi=2&ved=0CC8QFjAB&url=http%3A%2F%2Fen.wikipedia.org%2Fwiki%2FAugmented_Lagrangian_method&ei=n4PSTrDBC-LiiAK655z7Cw&usg=AFQjCNGSHeJWdFI1_y6dZHtju5NDlO7BPg&sig2=g4P70cmyJ70tLMT0Je41Ag&cad=rja