- This topic has 0 replies, 1 voice, and was last updated 4 years, 6 months ago by
.
Sometimes when you generate a DTT model, you will get a message indicating that one or more of your tests contain a node in an Indeterminate State:
This thread will describe how Indeterminate states come to be, how to identify them, and how to eliminate them to ensure the proper logical health of your Cause Effect Model.
Indeterminate states come to be when Mask constraints are not fully ‘tied’ to the nodes in the Mask constraint. In essence, a state of ‘Mask’ is passed into one of the effect nodes, and the effect node isn’t sure how to handle a state of ‘Mask’ so it returns an output of Indeterminate. Remember that you can predict the state of an effect node by looking at it’s truth table. For an ‘AND’ effect node with two Inputs, the truth table looks like this:
So long as the only possible states for the Inputs are TRUE and FALSE, we’ll always be able to predict the output state. What happens when introduce the possibility of an Input state of Mask? Let’s say Input 2 has a Mask constraining it. The truth table now look like this:
As we see in rows 2 and 3, Input 2 has a state of Mask. When Input 1 is FALSE (row 3), we’re able to predict the Output to be FALSE (any FALSE input into an AND gate results in an output of FALSE). Input 1 with a state of TRUE and Input 2 with a state of Mask (row 2) is where we run into a problem. We’re not able to predict the output, and thus we have an Indeterminate state.
As mentioned above, when you generate a model that contains Indeterminate states, a message will display in the Model Review Matrix (MRM). You can use the MRM to identify the tests impacted by this Indeterminate state as well as the node in this state. In the example above, the node ‘N6’ is Indeterminate in Test 3:
Now that we know what causes Indeterminate states and how to identify them, we can look at how to eliminate them so that we are always able to predict what the output should be.
The sure fire way to avoid Indeterminate states from occurring is to always ‘tie’ the constrained nodes in a Mask relationship back to the constraining nodes. In other words, use Links to create a relationship between the Masked and Masking nodes. Here’s an example:
Notice the relationships (Links) between N4 & N7 and N5 & N7. These Links ensure we’ll never have a situation where N4 (Input 1) is Masked and N5 is TRUE resulting in N7 in Indeterminate state.
We’ve done something similar with N4 and its Links to N9, incorporating the OR node N8 between the Input and Output nodes.
- You must be logged in to reply to this topic.