These benchmarks are about computing Reiter-style minimal diagnoses
for a given digital circuit. Each instance is encoded as a parallel 
circumscription. The <high, \emptyset>-minimal models correspond to 
minimal diagnosis (thus the abnormality atoms "ab" are minimized). 

The instances are produced as follows. First a random tree is
generated using the inverse Pruefer algorithm. The leaves of the tree 
corresponding to the inputs of the digita circuit, are assigned
random boolean values while intermediate nodes are assigned random
logical operations corresponding to the gates of the circuit. The
gate at root node of the tree produces the output of the circuit. The
value of the output is calculated and swapped in order to obtain
faulty behaviour for the circuit. 

The number of nodes in tree forming the digital circuit grows from 15 to 
30 and there are 100 random instances of each size. 


