Model checking demonstrates that a formally stated theory holds under a given interpretation; the theory is built from constants, functions and predicates stated in predicate logic or as data patterns, and the pattern model the network embodies is tested against the inputs. The check is only valid if no model is left unchecked. It can prove relationships among sets that obey a relation, for instance that every pair of letter images in an alphabet lies more than a threshold distance apart, which demonstrates the existence or absence of adversarial inputs.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.