So we want to be able to sample from the messages right? Since that's essentially a mixture of \psi_st terms, those terms should probably be normalisable with one of the variables blocked, this is fundamentally the first integrability requirement of Sudderth.
Think about this more (potentially discuss with Arnaud) and come back to this.
So we want to be able to sample from the messages right? Since that's essentially a mixture of
\psi_st
terms, those terms should probably be normalisable with one of the variables blocked, this is fundamentally the first integrability requirement of Sudderth.Think about this more (potentially discuss with Arnaud) and come back to this.