compilerbitch

I'm alluding to 'independent attribute' vs. 'relational attribute' models. Basically, an independent attribute model (of a circuit, say) would model all values independently of each other with respect to nondeterminism, whereas in a relational attribute model values can depend on each other. A simple example would be classical bits vs. qubits, where k classical bits can exist in 2^k states, but k qubits can be constrained to only possibly exist in a subset of the 2^k classical states -- in this case, the qubit model could formally be regarded as a relational attribute model, and the classical bit model could be regarded as an independent attribute model.