Unordered event picking for BDD risk analysis

Binary Decision Diagrams (BDDs) have become the next quantification evolution for risk analysis. Their accuracy and re-quantification speed have sparked much research and development. However, the ability to create them diminishes as the size of the models gets larger and larger. When using BDDs for risk analysis, the pre-process of event ordering is the primary factor in the BDD building speed and efficiency. Before building the BDD if one event ordering is determined, the size of the resulting BDD could be 100 times larger than if a different event ordering were to be used. Much of the research into BDDs for risk analysis is done on finding a better heuristic for determining a more optimal event ordering. Currently most BDDs used for risk analysis are ordered BDDs, where the ordering is determined before the BDD creation. This pre-ordering was very useful for some tasks when using BDDs for circuit evaluation. Since the idea of using BDDs for risk analysis came from their use in circuit evaluation, many of the methods and ideas were inherited. However, having ordered BDDs for risk analysis is not necessary for most tasks. In this paper a method of using unordered BDDs is explored. This method allows the BDD to choose the events that it wants to use next as it is being created. This may allow for a more optimal method of BDD creation by removing the restrictions created by ordered BDDs, thus allowing BDDs to be effective for larger models.