Eqs. (8.41)–(8.43) — collective decoherence decomposition with
ProvedBaconThesis.Collective.eq_8_41_8_43Let be the collective operators on qubits, and let be the complex algebra they generate together with the identity . Let run over , down to ( even) or ( odd). Then there are multiplicities and a unitary
such that
- (8.43) for ,
where is the standard -dimensional representation of ; 2. (8.42) for every ,
- (8.41) an operator on the qubits belongs to if and only if
that is, , the whole algebra and not only a subalgebra of it.
This is the structure theorem of strong collective decoherence: information stored in the degeneracy factor of a fixed is untouched by every error in , so each carries a decoherence-free subsystem of dimension (Definition 8.6.1).
Formalization Note with ; the multiplicity formula is stated in with factorial arguments and . The unitary is a matrix between the two index sets with and .
Differences from the source:
- reading: A ≅ ⊕J I{n_J} ⊗ M_{2J+1} (8.41), "corresponding to the decomposition" S_α = ⊕J I{n_J} ⊗ S_α(2J+1) (8.43) → one unitary W such that W S_α W† is the block form (8.43) and W A W† ranges exactly over the block operators ⊕_J I ⊗ X_J for A in the algebra; reason: the isomorphism of (8.41) is the one implemented by the change of basis of (8.43), as the text says
- encoding: the 2J+1 dimensional representation of su(2) → the standard matrices spinRep (2J) (S_z diagonal, real nonnegative ladder coefficients); reason: all such irreps are unitarily equivalent and W absorbs the change of basis; the normalization is that of (8.50)
- encoding: J ∈ {n/2, n/2 − 1, …, 0 or 1/2}; n_J = (2J+1) n!/((n/2+J+1)!(n/2−J)!) → r ∈ Fin (⌊n/2⌋+1) with J = n/2 − r; (m r : ℚ) = (n−2r+1)·n!/((n−r+1)!·r!); reason: the same numbers without half-integers
- encoding: the decomposition of the Hilbert space → a matrix W indexed by BlockIdx n m × (Fin n → Fin 2) with W†W = I and WW† = I; reason: a unitary between two finite index sets
- generalization: n qubits, n ≥ 1 (the book starts at n = 1, (8.44)) → every n ∈ ℕ, including n = 0; reason: at n = 0 the space is one-dimensional and every collective operator is 0, where the statement also holds; for n ≥ 1 it is the book's statement
import Mathlib import Definitions.Def_BBPSSW_Distill_pauli import Definitions.Def_BaconThesis_Collective_onQubit import Definitions.Def_BaconThesis_Collective_collS import Definitions.Def_BaconThesis_Collective_spinRep import Definitions.Def_BaconThesis_Collective_BlockIdx import Definitions.Def_BaconThesis_Collective_blockOp import Definitions.Def_BaconThesis_Collective_degeneracy open scoped Matrix
namespace BaconThesis.Collective
/-- **(8.41)–(8.43): the decomposition of `n` qubits under strong collective decoherence.**
Bacon, *Decoherence, Control, and Symmetry in Quantum Computers*, arXiv:quant-ph/0305025v1, §8.6, (8.41)–(8.43), p.118 (PDF p.135): "The algebra `A` generated from `S_x^[n]`,
`S_y^[n]`, and `S_z^[n]` plus identity `I` can be decomposed as
`A ≅ ⊕_{J=0(1/2)}^{n/2} I_(n_J) ⊗ M_(2J+1)` (8.41) … The degeneracy of the `J`th irrep is given
by `n_J = (2J+1) n! / ((n/2 + J + 1)! (n/2 − J)!)` (8.42). Corresponding to the decomposition
Eq. (8.41) the action of the `S_α`'s act as `S_α = ⊕_{J=0(1/2)}^{n/2} I_(n_J) ⊗ S_α(2J+1)`
(8.43), where `S_α(2J+1)` is the `2J+1` dimensional representation of `su(2)`."
For every number of qubits `n` there are multiplicities `n_J` and a unitary
`W : ℂ^(2^n) → ⊕_J ℂ^(n_J) ⊗ ℂ^(2J+1)` (`J = n/2, n/2 − 1, …, 0 or 1/2`) such that
1. (8.43) `W S_α^[n] W† = ⊕_J I_(n_J) ⊗ S_α(2J+1)` for `α = x, y, z`;
2. (8.42) `n_J = (2J+1) n! / ((n/2 + J + 1)! (n/2 − J)!)` for every `J`;
3. (8.41) an operator `A` on the `n` qubits lies in the complex algebra generated by `I`, `S_x`,
`S_y`, `S_z` if and only if `W A W† = ⊕_J I_(n_J) ⊗ X_J` for some operators `X_J` on
`ℂ^(2J+1)`: the algebra is all of `⊕_J I_(n_J) ⊗ M_(2J+1)`.
**Formalization Note.** `J = n/2 − r` with `r ∈ Fin (⌊n/2⌋ + 1)`; the multiplicities are `m r`,
the irrep factor is `Fin (n − 2r + 1)` (`= 2J + 1`), and (8.42) is `degeneracy n r` (factorial
arguments `n − r + 1 = n/2 + J + 1`, `r = n/2 − J`). The unitary is a matrix `W` with
`W† W = I` and `W W† = I`. `Algebra.adjoin ℂ` contains the identity. The book takes `n ≥ 1`; the
statement is also true (and included) for `n = 0`. -/
theorem eq_8_41_8_43 (n : ℕ) :
∃ (m : Fin (n / 2 + 1) → ℕ) (W : Matrix (BlockIdx n m) (Fin n → Fin 2) ℂ),
Wᴴ * W = 1 ∧ W * Wᴴ = 1 ∧
(∀ α : BBPSSW.Distill.Axis, W * collS n α * Wᴴ = blockOp n m (fun r => spinRep (n - 2 * r.val) α)) ∧
(∀ r : Fin (n / 2 + 1), (m r : ℚ) = degeneracy n r.val) ∧
(∀ A : Matrix (Fin n → Fin 2) (Fin n → Fin 2) ℂ,
A ∈ Algebra.adjoin ℂ (Set.range (collS n)) ↔
∃ X : (r : Fin (n / 2 + 1)) →
Matrix (Fin (n - 2 * r.val + 1)) (Fin (n - 2 * r.val + 1)) ℂ,
W * A * Wᴴ = blockOp n m X) := by sorry
end BaconThesis.Collective
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.