Working with large causal graph with high arc_ratio

[kenneth-lee-ch]

I used a randomBN to generate a bn with n=64 and arc_ratio=3. Then I add 16 latent variables with 1 latent variable always connected with the target variable and use causalModel to turn that bn into causalModel object. When I loop through all possible queries and use causallmpact to compute the interventional distribution, I found that a large number of queries give me this error:

Exception: [pyAgrum] several evidence/CPT are incompatible together (proba=0): some evidence entered into the Bayes net are incompatible (their joint proba = 0)

How I can prevent this from happening?

The original bn CPT is non-zero, but then the calculation seems to blow up in the causalImpact process.

[kenneth-lee-ch]

Please ignore the above. I realize there is a zero-entry on my CPT.