Xavier Holt edited Vague_Outline_Slice_Sampling_Slice__.md  over 8 years ago

Commit id: a14a895d451d20180f4598331e3dc265e5aea157

deletions | additions      

       

## Slice Sampling  Slice sampling is a non-deterministic method used to sample from arbitrary curves. Specifically it is a slice sampling belongs to the family of  Markov chain Monte-Carlo (McMC) method. methods.  The general principle behind such methods is to construct a Markov chain that has the desired distribution as an equilibrium solution. They are used to obtain numeric approximations from multi-dimensional integrals for which no closed-form solution is readily available. We can then use the samples to construct estimations of statistics on the underlying distribution (mean, mode, variance etc.) by simply calculating the equivalent sample-statistic. That is, we could estimate  Slice sampling in particular possesses several very desirable properties. It allows us to sample from integrals that are _proportional_ to our underlying distribution. That is, there is no assumption that our probability mass sums to one. The intractability of the denominator above is of no concern -- we simply proceed using the scaled posterior.