In a system approach, I need to define several events, corresponding to the nodes of a fail tree for example.
In the current version, I can only define the events separately (thanks to the getMarginal() method, in relation with the Ticket #112), but in this case, if I have correlated inputs, the correlation will be lost in the different Events, and my global probability of failure is wrong.
For example, if I try to calculate the probability of failure of "myEvent" (see below) thanks to Monte Carlo simulations, it will be wrong if there is correlation between the inputs...
vect = RandomVector(myDistribution)
G = RandomVector(EtatLimite, vect)
G1 = G.getMarginal(0)
G2 = G.getMarginal(1)
G3 = G.getMarginal(2)
threshold = 0.0
myEventG1 = Event(G1, ComparisonOperator(LessOrEqual()), threshold)
myEventG2 = Event(G2, ComparisonOperator(LessOrEqual()), threshold)
myEventG3 = Event(G3, ComparisonOperator(LessOrEqual()), threshold)
myEvent = ((myEventG1 or myEventG2) and myEventG3
Is there a method to solve this problem?
--
Emmanuel