Zero-Weight Support Face
Abstract
Zero projection weight confines a positive matrix to the complementary support.
Theorem 1.1 (Zero projection weight exposes a support face).
Proof. Machine-checked in Lean as D5/S3/QuantumStates/ZeroWeightSupportFace.zero_weight_support_face (✓ std3). ∎
Source. Repository-derived.
Commentary.
Let rho be a positive semidefinite complex matrix and P a self-adjoint idempotent matrix. If the trace weight Tr(rho P) vanishes, then both one-sided products P rho and rho P vanish.
The proof first compresses rho by P. The compression is positive semidefinite and has zero trace, so Mathlib’s trace-zero theorem makes it zero. A positive factorization of rho then turns this into the two one-sided annihilations.
Expanding the complementary compression and using those annihilations gives rho = (I-P) rho (I-P). No trace-one normalization or finite-rank restriction on the projection is required.
References
- Truth anchor:
D5/S3/QuantumStates/ZeroWeightSupportFace.zero_weight_support_face