The interactive evaluation of the ID-decide-test does not end with the same expected utility as the normal evaluation

[Manuel Luque]

The Maximum Expected Utility of the ID-decide-test according to the decision tree or the “inference mode” appears to be 9.3289. However, when using the interactive mode of arc-reversal in the GUI, it ends with a utility slightly different: 9.40.

[Iago París]

I haven’t been able to replicate the 9.40 result. However, there was an issue when absorbing a node implies merging utility children (as in the case provided) which produced random wrong results in the interactive mode. That error is now solved so please, @Manuel Luque , test again to check if it is now working correctly. For me it is.

[Iago París]

Solved if no further notice