# Claim 2

**Result: verified.** The exact diagram engine extracts Pareto fronts
from actual nonzero monomials and matches every printed Theorem 4.1 triplet in
14 cells: SYM/ASYM ν=2 through s=4 and SYM/ASYM ν=4 through s=1. A one-power
mutation of every predicted q exponent is rejected.

The finite cells are exact rational computations. They are recorded as finite
checks, not misrepresented as an independent proof for all ν and all orders.

## Registered wording

Theorem 4.1 characterizes the leading and Pareto-optimal terms of the loss expansion for identity-tensor targets, organizing gradient-flow learning regimes into a polygon structure defined by scaling conditions on H, p, and sigma^2 (Section 4, Theorem 4.1).