On the other hand maybe the code never will write itself. The following
change to the Poisson
model in glmmadmb using the log(1_+y_i) transformation to get good
initial parameter estimates
converges nicely on this data set.
case 0: // Poisson
if (cph<2)
tmpl=-square(log(1.0+y(_i,1))-log(1.0+lambda));
else
tmpl= log_density_poisson(y(_i,1),lambda);
break;
case 1: // Binomial: y(_i,1)=#successes, y(_i,2)=#failures,