Creating a node which sums the results of parent nodes' states

The front end.
Post Reply
clairebre
Posts: 2
Joined: Thu Sep 30, 2021 6:06 am

Creating a node which sums the results of parent nodes' states

Post by clairebre »

I am trying to figure out how to create a node which gives a result which is the sum of the 'dies' probabilities in its parent nodes. The parent nodes are all discrete and have two states 'lives' or 'dies'.

Any advice would be appreciated. Thanks, Claire
marek [BayesFusion]
Site Admin
Posts: 430
Joined: Tue Dec 11, 2007 4:24 pm

Re: Creating a node which sums the results of parent nodes' states

Post by marek [BayesFusion] »

I'm afraid that this may be tricky. Being able to refer to the posterior probabilities of the parent nodes would amount to using the solution of a system of equations in the definition of some of the variables. I would suggest that you make all parent nodes and the child node continuous and then reflect on how to capture the number of deaths rather than probability of death. The child node could then be a simple sum of the parents. Please note that definitions of nodes have to be made in terms of states of the parents and not their probabilities.
I hope this helps,

Marek
nzsk
Posts: 24
Joined: Thu Aug 29, 2019 8:27 am

Re: Creating a node which sums the results of parent nodes' states

Post by nzsk »

I have a related problem - I have a node that sums the values of five parent nodes (all continuous variables). However, I want to exclude zeros from the parent nodes, hence I set the minimum interval for each parent node to 1e-12. This means that for some parent nodes, I obtain between 2-30% of valid simulation values. This seems to produce correct results for these parent nodes. However, when I then try to sum all values for these five parent nodes in a single child node, I don't generate any valid samples. I get the message 'Error (-5): Update Beliefs failed. I don't understand why this is the case. I increased the number of simulations to 100 000 and the minimum number of valid samples in a parent node is 2500, others have a lot more simulated samples. Why is it that the child node can't generate enough valid samples? I set inference to 'Reject out-of-bounds and invalid samples', otherwise the zeros are included among the samples. Is there any other way how to exclude the zeros from the posterior distribution of a parent node for calcuations in a child node?
marek [BayesFusion]
Site Admin
Posts: 430
Joined: Tue Dec 11, 2007 4:24 pm

Re: Creating a node which sums the results of parent nodes' states

Post by marek [BayesFusion] »

I will be happy to look at this problem. Would you be willing to share the piece of our model that produces this behavior?
Cheers,

Marek
Post Reply