Example of predicting on newdata by rebuilding TMB object

lm.R

library(TMB)
compile("lm1.cpp")
dyn.load(dynlib("lm1")) # can't call it 'lm'!
set.seed(123)
x <- seq(0, 10, length.out = 10)
data <- list(Y = rnorm(length(x)) + x, x = x)
parameters <- list(a = 0, b = 0, logSigma = 0)
obj <- MakeADFun(data, parameters, DLL = "lm1")
opt <- nlminb(obj$par, obj$fn, obj$gr)
sdr <- sdreport(obj)
sdr
obj$report(obj$env$last.par.best)

# now give it new data that is longer and try to predict
nd <- data.frame(Y = rep(1, 11), x = rep(2, 11))
obj2 <- MakeADFun(nd, obj$env$parList(), DLL = "lm1")
obj2$fn(opt$par)
lp <- obj$env$last.par.best
obj2$report(lp)

lm1.cpp

#include <TMB.hpp>

template<class Type>
Type objective_function<Type>::operator() ()
{
  DATA_VECTOR(Y);
  DATA_VECTOR(x);
  PARAMETER(a);
  PARAMETER(b);
  PARAMETER(logSigma);
  Type nll = 0.0;
  Type sigma = exp(logSigma);
  vector<Type> mu(x.size());
  for (int i=0; i < x.size(); i++) {
    mu(i) = a + b * x(i);
    nll -= dnorm(Y(i), mu(i), sigma, true);
  }
  REPORT(mu);
  REPORT(sigma)
  return nll;
}