 Tempered Gibbs sampler

A tempered Gibbs sampler changes the coordinate-selection probabilities and compensates by drawing from modified complete conditionals. Its invariant law can differ from the desired target, so expectations under the target are recovered with explicit importance weights.