Equation Model

[cpm827]

Hello! I have an equation model which i will like to convert into a dynamic Bayesian network, is this possible in GeNIe ?

[marek [BayesFusion]]

Not quite yet but please stay tuned -- we should have it later this year. At the moment, what you can do is create a static hybrid model (i.e., with equation nodes) that has variables at time t and then t-1 (assuming links of the first order). Once you have that, you can use your static model for generation of CPTs for your DBN. Please find enclosed two models that I used to implement a discrete dynamic predator-prey model involving Lotka-Volterra equations. I generated the CPTs in the static model and just copied and pasted them into the DBN. I hope this helps. If you give us some months, we will have DBNs that have equation nodes within them.
Cheers,

Marek

[ccorreaf]

Hello,

I would like to confirm whether temporal nodes can be defined using equations in PySmile. For illustration:
b{0} = a
b{t+1} = b{t} + c{t} + d{t}, where a, c, and d follow Uniform distributions.

While developing a model, I noted the function:
set_node_temporal_equations(node: int | str, equations: List[str])

During execution, an error occurred. Could you clarify the format in which these temporal equations should be provided? In Genie, a new line is added for each temporal equation

Is there an example available for reference?

Thank you very much.

[shooltz[BayesFusion]]

Support for continuous/hybrid dynamic Bayesian networks was added to PySMILE. It will be released as part of the next SMILE update. Private build is available for those users who need to work with equation nodes in DBNs.