Theorem B.0.1 — acts as independently on every weak collective DFS (corrected: containment form of (9.11))
ProvedBaconThesis.WeakUniv.theorem_B_0_1Let qubits undergo weak collective decoherence, with decoherence-free subspaces , , spanned by the computational basis states of Hamming weight , and projectors . Let be the real Lie algebra generated by , where consists of the nearest-neighbour exchange interactions and the diagonal two-qubit operators , , . Then for every with and every -qubit matrix ,
That is, contains a copy of that acts on and annihilates every other , : every traceless Hamiltonian supported on is obtained from by real scalar multiplication, addition and commutators .
Exponentiating, the nearest-neighbour Hamiltonians enact every special unitary on each weak collective decoherence DFS separately, which makes these DFSs usable for universal encoded quantum computation.
Formalization Note Matrices are indexed by ; is the real Lie span (bracket ) of , .
Differences from the source:
- correction: Theorem 9.2.1's display (9.11) L = ⊕_{h=0}^n su(C(n,h)) → the containment su(C(n,h)) ⊕ 0 ⊆ L for each h; reason: the equality is false: for n = 2, i T^P₁₂ = i|00⟩⟨00| ∈ L acts nontrivially on the one-dimensional DFS₂(0) of Hamming weight 0, where su(1) = 0; the containment is what Theorem 9.2.1 states in words and Theorem B.0.1 states and Appendix B proves
- reading: "there exist sets of Hamiltonians H^n_{K_n} … acting as su(d_{K_n}) on the DFS … independently on this DFS only" → every traceless anti-Hermitian X supported on the DFS lies in L; reason: such a set acts as su(d) on the DFS and as zero elsewhere exactly when the real Lie algebra it generates, which lies in L, consists of all traceless operators supported on the DFS
- encoding: the label K_n = n − 2h and d_{K_n} (printed d_K = C(n, K) on p.241) → the Hamming weight h and the projector P_h; reason: same subspaces; the statement needs no dimension, and the true dimension is C(n, h)
- encoding: Hamiltonians with bracket i[·,·] → anti-Hermitian matrices in the real Lie span of iH; reason: X ↦ iX is a Lie algebra isomorphism; same instances
import Definitions.Def_BaconThesis_WeakUniv_lieAlg import Definitions.Def_BaconThesis_WeakUniv_dfsProj open scoped Matrix
attribute [local instance 100] LieRing.ofAssociativeRing
namespace BaconThesis.WeakUniv
/-- **Theorem B.0.1 (= Theorem 9.2.1, containment form of (9.11)): `H` acts as `su(dim DFS_n(h))`
independently on every weak collective decoherence DFS.**
Bacon, *Decoherence, Control, and Symmetry in Quantum Computers*, arXiv:quant-ph/0305025v1,
Theorem B.0.1, p.235 (PDF p.252): "For any n ≥ 2 qubits undergoing weak collective decoherence,
there exist sets of Hamiltonians H^n_{K_n} [obtained from H of Eq. (B.1) via scalar
multiplication, addition, and Lie-commutator] acting as su(d_{K_n}) on the DFS corresponding to
the eigenvalue K_n. Furthermore each set acts independently on this DFS only (i.e., with zeroes in
the matrix representation corresponding to their action on the other DFSs)." Theorem 9.2.1,
p.127 (PDF p.144), states the same in words ("acts independently as su(n_h) on DFS_n(h) … there
are elements in the Lie algebra which act only on DFS_n(h) and annihilate all other DFS_n(h′)").
For every `n ≥ 2`, every Hamming weight `0 ≤ h ≤ n` and every traceless anti-Hermitian `X` that
acts on `DFS_n(h)` only (`P_h X P_h = X`), `X` lies in the real Lie algebra `L` generated by
`i·H`. Equivalently, every traceless Hermitian Hamiltonian supported on `DFS_n(h)` is obtained
from `H` by real scalar multiplication, addition and the commutator `i[·,·]`.
**Formalization Note.** The displayed equality (9.11), `L = ⊕_h su(C(n,h))`, is false (for
`n = 2`, `i T^P_12 = i|00⟩⟨00| ∈ L` acts nontrivially on the one-dimensional `DFS₂(0)` of weight 0,
where `su(1) = 0`); the statement is the containment that both theorems assert in words and that
Appendix B proves. The book's label `K = n − 2h` (B.2) is replaced by the Hamming weight `h`, and
`d_K` by `dim DFS_n(h) = C(n, h)` (Appendix B, p.241, prints `C(n, K)`), which needs no number in the
statement. `su(d)` on the DFS is the set of traceless anti-Hermitian matrices supported on it. -/
theorem theorem_B_0_1 (n : ℕ) (hn : 2 ≤ n) (h : ℕ) (hh : h ≤ n)
(X : Matrix (Fin n → Fin 2) (Fin n → Fin 2) ℂ) (hX : Xᴴ = -X)
(hsupp : dfsProj n h * X * dfsProj n h = X) (htr : X.trace = 0) :
X ∈ lieAlg n := by sorry
end BaconThesis.WeakUniv
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.