Mass Functions

A mass function is a basic belief assignment over propositions in a frame.

m = frame.mass({
    A: 0.4,
    B: 0.2,
    A | B: 0.4,
})

Mass values must be non-negative and sum to one by default. Tiny floating-point drift near one is normalized, while real sum mismatches are still rejected.

Keys may be Proposition objects, atom names, string expressions, or iterables of atom names:

frame.mass({"A": 0.4, "B": 0.2, "A | B": 0.4})
frame.mass({A: 0.4, B: 0.2, A | B: 0.4})

Inspecting masses

m.items()
m.focal()
m.to_dict()
m.total_mass
m.conflict

items() returns (proposition, value) pairs sorted by proposition label. focal() returns only propositions with non-zero mass. to_dict() is useful for display, serialization, and examples.

m.conflict is the mass assigned to the empty proposition. It is usually zero for a valid source, but it can appear after unnormalized combination on a constrained model.

Belief measures

m.mass(A)
m.belief(A)
m.plausibility(A)
m.commonality(A)

mass(A) returns the direct assigned mass.
belief(A) sums masses of propositions contained in A.
plausibility(A) sums masses of propositions intersecting A.
commonality(A) sums masses of propositions that contain A.

For DST, belief(A) <= plausibility(A) gives the usual lower and upper support interval for A.