linear vs. log-linear approx., subs(z, [y], exp([y]))

[cassiopia_79]

Dear all,

is there any way I can write down my system of model equations (in levels, not deviations) in the mod file, and then substitute, at once, all variables by the exponential of the variables?

Basically, what I want is to get a log-linear approximation but without having to write down exp(x) for each variable x.

the Matlab command would be 'subs', and in particular, I think I could add a line in the 'model.ff' file, subs(z, [y], exp([y])), and then call the model.m file and that would do what I want..

however, since I'm currently writing a code not for personal use, it'd be great if there is some option to do that directly from compiling the mod file (so I don't need to explain whoever I send a code, what they need to modify somewhere else.. etc..)

for bigger models, the code in the mod file gets much less readible if there are the exp(x), exp(y), exp(z) around everywhere...!

if there is no way in Dynare v3, do you think v4 could include that as an option..?

Any help appreciated!