The independence number $α$ of a finite simple graph lies between the polynomial-time computable Havel-Hakimi residue $R$ and annihilation number $a$. The Caro-Wei sum $W$ gives another lower bound and the known upper bound $α\le(Δ+1)W/2$ for maximum degree $Δ\ge1$. This upper bound does not by itself control the larger quantity $a$. This note proves $a\le(Δ+1)W/2$ for $Δ\ge1$, with equality attained, for example, by regular graphs of positive degree and even order. Together with $R\ge W$, this bound shows that the interval from $R$ to $a$ improves or matches both Caro-Wei endpoints. Its sharp worst-case endpoint ratio remains $(Δ+1)/2$, even among connected graphs. For maximum degree at least three, the bound also proves the inequality $α\ge(a+R)/Δ$ conjectured by TxGraffiti, an automated conjecturing program. A direct calculation handles connected graphs of maximum degree two. Together these cases resolve the conjecture for every connected graph with at least three vertices. The single edge is the only nontrivial connected exception. Equality holds for exactly eight connected graphs up to isomorphism. For maximum degree at least three, the conjectured lower bound never exceeds the residue, so it gives no stronger independence estimate.
ML-KEM and ML-DSA compute forward number-theoretic transforms on secret polynomials. Their fixed multipliers, or twiddles, are a documented fault-attack target. This paper characterises the linear map from a secret polynomial to the difference between correct and faulted transform outputs. Each selected twiddle takes a fixed changed field value at every use during one call; zero and nonzero replacements are covered. For one replacement, rank equals the number of butterflies sharing its twiddle, halving by layer from 128 to 2 for ML-KEM and to 1 for ML-DSA. With one replacement per layer, known fault positions, replacement values and a raw transform difference determine every coefficient except the constant and linear coefficients for ML-KEM, or the constant for ML-DSA. Lean 4 certifies this result for both 256-coefficient shapes over fields of characteristic other than two, with nonzero original twiddles. Those coefficients cancel from the difference for every allowed fault set and replacement choice. Seed storage motivates comparing a known correct public key with a released faulty key from the same seed. If only one ML-KEM secret-polynomial transform is affected and all other computations are unchanged, the decoded key difference is a known matrix column times the transform difference. An invertible column entry reconstructs the latter. For ML-DSA, reconstruction and exact candidate counts from paired compressed keys remain unestablished.