[ADMB Users] inner maxg = 0 infinite loop in ADMB-RE

[dave fournier]

I can't stand the idea of revisitng this model. However there is
a general approach to fitting models where one variable (here the 
biomass) must be >0 and you are removing something from it (here the 
catch)  whihc may be greater than the biomass for bad values of the
model parameters.

start with

                 B - C

this is the same as

                  B*(1-C/B)

which to first order is the same as

                  B/(1+C/B)

and works for C>B

you can get it more accurate  by splitting up the time step as it were
to get

               B/(1+C/nB)^n


You would think they would know this stuff in the univeristies by now.