A question about how to start ab initio simulations

[ftft]

I want to simulate STM images of some compound composed of 3 molecules A, B and C..
Which is the right way to do the simulations?
*Use the known equilibrium geometries of A, B and C as input for simulating the compound? or
*First optimize the geometries of A, B and C independently and then use their optimized geometries in simulating the compound
I have tried both ways and there were tiny differences in the resulting compound structures.. Still, I need to check for academic accuracy..

Thanks in advance

 

[ModestyKing]

What do you mean "known" equilibrium geometries? Were those found by other peoples' simulations? Was it through experiment? What is the chance the experimental geometries are a little off due to the experimental method used?
When doing the ab initio optimizations of A, B, and C, did different methods, basis sets, functionals produce different results?

Those are some questions to think about. As for me, I generally take the experimentally confirmed equilibrium geometry, optimize it with my program, and if I get worryingly different results, I try a different method for the optimization.