This question is based on @Heisenberg's Question with the same title.
I recently offered a bounty for a more detailed answer for the same question. These are the things I would like to be in the answer:
Can molecules with 'n' canonical structures be taken as an n-state system with coupling(At least as an approximation) ?
Suppose if it is possible. Consider a molecule with 2 canonical structures represented by $|A\rangle$ and $|B\rangle$ Without the coupling let the energies of the two states be $E_1$ and $E_2$ respectively. So the Hamiltonian is represented by:
$H=\pmatrix{E_1& 0\\0&E_2}$
Now with coupling the Hamiltonian in the most general case will be:
$H=\pmatrix{H_{11}& H_{12}\\H_{12}^*&H_{22}}$
The energy eigen-value with the lower energy in this case is:
$E_1 = \frac{H_{11}+H_{12}}{2} - \frac{\sqrt{(H_{11}-H_{22})^2 + 4H_{12}H_{12}^*}}{2}$
So my doubt is if the 2 states are symmetrical then $H_{11}$ and $H_{22}$ are equal right? Then the term inside the square-root $(H_{11}-H_{22})$ becomes zero, and it seems to me that when this term becomes zero, the energy splitting becomes the least.
And yet it is resonance with equivalent canonical structures that are more stable. Why?
Are molecules with equivalent resonance structures actually more stable? Please include references or experimental evidence if the it is true.
If molecules with equivalent resonance structures are in fact more stable than others, then does it mean that resonance in molecules cannot be approximated as a discrete n-state system even for ones with 2 canonical structures (Since there might be a contradiction with arguments in 2)) Can resonance be understood only using MOT or Huckel Theory ?
Note: Please add your answer in Heisenberg's question so that I can award you the bounty. You might understand why I'm having these doubts and the doubts itself if you read these chapter of the Feynman lectures on Physics Chapter 8 , Chapter 9 , Chapter 10