As far as i can see, the solution provided here is a little bit different from the normal approach, where two R5/2 resistances would be used and the common node voltage would be gathered from their common connection. The above presented approach would make sense to me if two buffers were used in order to repeat the signal and then divide it on the two 47K resistors. But why use such a configuration, where the non-inverting input is tied to the output? I have never seen such thing and by simulating this i got a strange result.
Any thoughts? Thanks.