Path integral in coherent states

1. Sep 28, 2010

topper

Hey,

there is something I don't really understand about the path integral (functional integral) formalism in QFT:

Why do you need to introduce a coherent-state representation of the Dirac fields in order to evaluate their path integral?
Where is the crucial point why it doesn't work like in the scalar field case, where you have the operators in the correlation function on the one side and just scalar functions under the integral on the other side?

I hope someone gets my point :)

Thanks,
topper