Rev Language Reference
mvDPPValueScaling - Scaling move applied to individual categories of a Dirichlet process mixture
Operates on draws from a Dirichlet process prior (DPP) on mixtures of [RealPos](https://revbayes.github.io/documentation/RealPos.html) distributions, i.e., distributions defined over non-negative real numbers.
Usage
Arguments
| x : | RealPos[] (<stochastic> pass by reference) |
| The variable on which this move operates. | |
| weight : | RealPos (pass by value) |
| The weight determines the relative frequency with which this move will be attempted. For details, see the description of the 'moveschedule' parameter on the documentation page for 'mcmc()'. | |
| Default : 1 | |
| tuneTarget : | Probability (pass by value) |
| The acceptance probability targeted by auto-tuning. | |
| Default : 0.44 |
Details
In Dirichlet process mixtures, the number of categories (= clusters) is not specified beforehand but inferred from the data, and can range anywhere from 1 to the total number of elements (= observations). The move takes the current number of categories and simultaneously updates the value of every category using the scaling move with a scaling factor (lambda) of 1.
Example
# Here, we draw from a DP mixture for 3 elements, where every element
# is a non-negative real number drawn from an exponential distribution
x ~ dnDPP(1, dnExp(1), 3)
# Next, we add the move. Note that without moves other than
# mvDPPValueScaling, only the values of the categories will be updated:
# the total number of categories and the assignment of elements to
# categories will be determined by the initial draw.
moves[1] = mvDPPValueScaling(x, weight=1)
monitors[1] = mnScreen(x, printgen=1)
mymodel = model(x)
mymcmc = mcmc(mymodel, monitors, moves)
mymcmc.run(generations=50)
Edit this page