cast from type double to integer

[Maia Sosa Kapur]

I have a matrix of derived quantities inside my model which do not (explicitly) depend on parameter values.

I would like to round these to the nearest integer, stash that output separately, and use the result to index within a different matrix. I currently have a static, externally-generated lookup table but would like to use the "live" values for accuracy.

So far I have tried:

 int temp_ind = int(matrix_of_doubles(x,y,z)); 

 return_val = other_table(temp_ind); // must be an integer

And received error: invalid cast from type 'CppAD::AD<double>' to type 'int'. If I substitute a raw double for  matrix_of_doubles(x,y,z)  (e.g. 3.14159) this error doesn't come up. I suspect TMB is thinking that matrix_of_doubles is deeply related to differentiated parameters, and therefore shouldn't be rounded?

For what it's worth, I'm trying to use the predicted lengths-at-age to look up selectivity values associated with that length bin. In my model, the predicted lengths in question are deterministic (no estimation of growth parameters). 

Is there a way to store the rounded value of a derived double, with no intention of modifying the original quantity? 

Thanks

MK

[Maia Sosa Kapur]

Thanks to help from Dr. W. Stockhausen, the appropriate syntax is indeed straightforward:

int temp_ind = CppAD::Integer(matrix_of_doubles(x,y,z)); 

Thanks all.