txgraffiti.generators

txgraffiti.generators.ratios

Enumeration generators for ratio-based conjectures.

This module defines functions that generate inequality conjectures of the form target >= c * feature and target <= c * feature under Boolean hypotheses, based on numerical ratios observed in a dataset.

txgraffiti.generators.ratios.ratios(*args, **kwargs)

txgraffiti.generators.convex_hull

Convex-hull-based conjecture generator.

This module defines a generator that builds linear inequality conjectures of the form target ≥ RHS or target ≤ RHS by computing the convex hull of feature-target vectors restricted to a logical hypothesis.

txgraffiti.generators.convex_hull.convex_hull(*args, **kwargs)

txgraffiti.generators.optimization

Linear programming-based conjecture generator.

This module defines generators that formulate inequalities between numeric graph invariants using sum-of-slacks linear programs. Given a hypothesis and a list of feature properties, the LP computes the best linear upper and lower bounds on a target invariant, yielding inequalities suitable for conjecture generation.

Requires an LP solver (CBC or GLPK) available on the system path.

txgraffiti.generators.optimization.linear_programming(*args, **kwargs)