Perfect foresight, nonlinear to linear+ ZLB

I have a model in a .mod file and the equilibrium conditions are written in their full nonlinear form.

I would like to do perfect foresight simulations using a linear version of the same model except that I still enforce the ZLB constraint.

Can I do this without having to write a new .mod file with linear versions of all the equilibrium conditions?

There is an option called linear_approximation (see here in the reference manual) which will replace the non linear model by its first order approximation around the steady state (replacing the nonlinear equations by the Jacobian evaluated at the deterministic steady state), but obviously this would also drop any sign constraint on the endogenous variables if you code these constraints with max, min or inequality operators in the model block.

I did not try, but you may achieve what you want by specifying the constraints as slackness conditions (mcp equation tags) and using the lmmcp algorithm. Unfortunately this is not allowed with the current version of Dynare. If feasible, I will open a Github issue for that.

So the only currently available approach is to provide the linearized version of the model.