Closed rmnldwg closed 10 months ago
To wrap this up a little more:
Unilateral
should request transition probabilities from the Node
class, and send new parameters to the Edge
class.Node
class should request the parameters to compute its transition probability from the incoming Edge
classes, and send the computed probability to the Unilateral
class.Edge
class should receive new spread probabilities from the Unilateral
class, and send then to the Node
class it points to upon request.That way, everything is nicely encapsulated and modular. For example, if we want to change the parametrization, we would only need to modify the implementation in the Unilateral
class.
This is solved with the latest update. What has not been implemented so far is the splitting of the Node class into tumor node and LNL node (we can discuss this on Wednesday). Additionally I sampled both a trinary and binary model and achieved full conformity with the former model before the adaptations. --> Seems like the adaptations are fully functional, but tests still need to be developed for the new code. (To be done after discussion whether the implementation is fine)
Currently, the
Unilateral
class handles passing spread probabilities to theNode
instances to help it compute itstrans_prob
. But theNode
class should be able to do that itself based on its incomingEdge
instances.