(of the general form variance(f)≲E[∣∣gradient(f)∣∣2]variance(f) \lesssim \mathbf{E}[||gradient(f)||^2]variance(f)≲E[∣∣gradient(f)∣∣2], see here)
See this video for simple derivation of the Poincare inequality for Boolean functions and also funcitons from the hypercube to the reals. This is like a discrete Isoperimetric inequality