I want to use a mixture model with an added constraint that the number of samples from each mixture component should be approximately equal. E.g. in the MWE below, I know that half the samples should come from each of the two components, and I ensure that this constraint is not violated too much by …

[image] erik-werner: I’m presenting some work based on this model to an audience with lots of Stan users next week, so it would be nice to know whether this model would be possible to run in Stan, where IIUC it’s impossible to sample from discrete variables? But it seems like it would be very di…

Can you marginalize a mixture model where the draws from the different components are not independent?

Questions v5

iavicenna March 1, 2024, 10:32am 4

So you want the weights to be exact? Then you can set w=[1/2, 1/2] in the code I have posted above?

Topic		Replies	Views
Marginal likelihood for distributions with discrete variables v3	12	1511	November 14, 2022
How would you go about marginalizing over discrete parameters in PyMC? Questions	1	1584	May 27, 2020
Condition Categorical Variable on Bernoulli Parents Questions	16	2340	May 14, 2019
Marginalizing out a categorical variable Questions	18	2241	April 1, 2021
Where did the indicator go in the customed likelihood? Questions	1	390	December 9, 2019

Can you marginalize a mixture model where the draws from the different components are not independent?

Related topics