I am given the feeling that there exists scenarios when this equality holds. Can anyone state/refer to the situations?
One case that I hear of is that for $2+1$ CFTs the entanglement entropy across a spatial disk is the same as the free energy of the CFT system on a $S^2 \times S^1$ (or $S^3$?)
What is the exact correct statement?
I am told that this paper http://arxiv.org/pdf/1102.0440v2.pdf "proves" the above but despite reading this paper a few times I can't locate this argument! (...may be I am missing something very badly!...)