Closed olgakuznetsova closed 3 years ago
We need this restrictions for the computation of the MLE but the partition is already computed in the gaussianRing itself because we need it to create the matrices K,L,P. So it seemed that it was too restrictive to not allow this in the ring itself.
I must say that I am super confused about all this. I can't find in Lauritzen reference (the mixed graph paper) where the partition Carlos wrote in our background document comes from.
This is highly connected to https://github.com/Macaulay2/Workshop-2020-Warwick/issues/149 because to document the kind of gaussianRings mixedGraphs we are building we need a good understanding of how restrictive is this partition. Plus I think we should say something in the intro of GraphicalModelsMLE about exactly for which king of graphs we are giving a meaningful MLE.
Hi @roserhp @lukeamendola Did you have a chance to discuss this yesterday?
No, we could only discuss solverMLE. Let's do it on Thursday!
Check #166
@roserhp we have probably discussed this but now I forgot, sorry: why are these warnings and not errors?
https://github.com/Macaulay2/Workshop-2020-Warwick/blob/efa4e5df2c544d671f85c2ee144b26a254c12491/AlgebraicStatistics/MLE/StatGraphs.m2#L266
https://github.com/Macaulay2/Workshop-2020-Warwick/blob/efa4e5df2c544d671f85c2ee144b26a254c12491/AlgebraicStatistics/MLE/StatGraphs.m2#L272