Forslagsfordeling
Forslagsfordeling refers to the probability distribution from which candidate states are drawn in a Markov Chain Monte Carlo (MCMC) algorithm. The design of a good proposal distribution is crucial for the efficiency of many sampling methods, especially the Metropolis–Hastings algorithm.
In the Metropolis–Hastings framework, starting from a current state \(x\), a new state \(y\) is proposed by
The choice of proposal impacts both convergence speed and computational cost. A proposal that is too narrow
In practice, the proposal distribution is tuned empirically or through theoretical criteria such as optimal acceptance