amp_damping_shift_sharp_spectral_bound
Fix $d \ge 2$, $V = |0\rangle\langle 1|$, $P = |1\rangle\langle 1|$, $G(X) = VXV^\dagger - \tfrac12(PX + XP)$ and $L_c = G + c\,\mathrm{id}$ ($c \in \mathbb{R}$). Then $e^{tL_c}$ is completely positive for all $t \ge 0$, trace-preserving for some $t > 0$ only if $c = 0$, $\operatorname{Tr} L_c = cd^2 - d$, $m(L_c) = c - 1$, $s(L_c) = c$, and hence $\operatorname{Tr} L_c = d\,m(L_c) + (d^2 - d)\,s(L_c)$.
Statement
theorem amp_damping_shift_sharp_spectral_bound (d : ℕ) (h : 2 ≤ d) (c : ℝ) :
(∀ t : ℝ, 0 ≤ t → has_kraus_representation d (superop_exp d (amp_damping_generator d h + (c : ℂ) • LinearMap.id) t)) ∧
(∀ t : ℝ, 0 < t → is_trace_preserving_linear_map d (superop_exp d (amp_damping_generator d h + (c : ℂ) • LinearMap.id) t) → c = 0) ∧
LinearMap.trace ℂ (Matrix (Fin d) (Fin d) ℂ) (amp_damping_generator d h + (c : ℂ) • LinearMap.id) = (c : ℂ) * (d : ℂ) ^ 2 - d ∧
spectral_min_re d (amp_damping_generator d h + (c : ℂ) • LinearMap.id) = c - 1 ∧ spectral_max_re d (amp_damping_generator d h + (c : ℂ) • LinearMap.id) = c ∧
LinearMap.trace ℂ (Matrix (Fin d) (Fin d) ℂ) (amp_damping_generator d h + (c : ℂ) • LinearMap.id) =
(d : ℂ) * (spectral_min_re d (amp_damping_generator d h + (c : ℂ) • LinearMap.id) : ℂ) + ((d : ℂ) ^ 2 - d) * (spectral_max_re d (amp_damping_generator d h + (c : ℂ) • LinearMap.id) : ℂ)- Formalizes
- arXiv:2610.37187 Proposition 9 (eq. 7.1)
- Verified
- 08 Nov 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 11
- amp_damping_shift_completely_positiveFor every integer $d \ge 2$ and every real number $c$, let $G$ be the amplitude-damping generator…
- amp_damping_shift_trace_preserving_only_ifIf $e^{tL_c}$ is trace-preserving for some $t > 0$, then $c = 0$.
- amp_damping_shift_traceFor every integer $d \ge 2$ and every real number $c$, let $G$ be the amplitude-damping generator…
- amp_damping_shift_spectral_min_reThe smallest real part of an eigenvalue of $L_c = G + c\,\mathrm{id}$ is $c - 1$.
- amp_damping_shift_spectral_max_reThe largest real part of an eigenvalue of $L_c = G + c\,\mathrm{id}$ is $c$.
- has_kraus_representationA linear map on d × d complex matrices has a Kraus representation if it can be written as X ↦ ∑ᵢ Kᵢ X Kᵢ† for finitely many Kraus…
- superop_expFor a $\mathbb{C}$-linear endomorphism $L$ of $M_d(\mathbb{C})$ and $t \in \mathbb{R}$, the linear endomorphism $e^{tL}$: the map whose…
- is_trace_preserving_linear_mapA linear map Φ on d × d complex matrices is trace-preserving if the trace of Φ(X) equals the trace of X for all d × d complex matrices X.
- amp_damping_generatorThe $\mathbb{C}$-linear map $G : M_d(\mathbb{C}) \to M_d(\mathbb{C})$ (bundled as a linear endomorphism, for $d \ge 2$),…
- spectral_min_reThe minimal real part m(Φ) of an eigenvalue of a linear map Φ on d × d complex matrices is the infimum of the real parts of the elements…
- spectral_max_reThe maximal real part s(Φ) of an eigenvalue of a linear map Φ on d × d complex matrices is the supremum of the real parts of the elements…
Read back from the Lean
For any natural number $d$ such that $2 \le d$ and any real number $c$, let $L$ be the $\mathbb{C}$-linear map on $M_d(\mathbb{C})$ given by $L = \mathcal{L} + c \cdot \mathrm{id}$, where $\mathcal{L}(X) = E_{0,1} X E_{0,1}^* - \frac{1}{2}(E_{1,1} X + X E_{1,1})$ is the amplitude damping generator. Then: 1. For all real numbers $t \ge 0$, $\exp(t L)$ has a Kraus representation (i.e., is completely positive). 2. For all real numbers $t > 0$, if $\exp(t L)$ is trace-preserving, then $c = 0$. 3. The operator trace of $L$ (over $\mathbb{C}$ on $M_d(\mathbb{C})$) equals $c d^2 - d$. 4. The minimum of the real parts of the spectrum of $L$ is $c - 1$, and the maximum of the real parts of the spectrum of $L$ is $c$. 5. The operator trace of $L$ equals $d \cdot \operatorname{spectral\_min\_re}(L) + (d^2 - d) \cdot \operatorname{spectral\_max\_re}(L)$.
Written by a model that saw only the Lean, never the English above. If the two disagree, that is worth a challenge.
The proof, explained
The theorem packages six properties of the shifted amplitude damping generator $L_c = G + c\,\mathrm{id}$ into a conjunction. The first five conjuncts—complete positivity of $e^{tL_c}$ for $t \ge 0$, necessity of $c = 0$ for trace preservation, the trace formula $\operatorname{Tr} L_c = cd^2 - d$, and the minimal and maximal real eigenvalues $m(L_c) = c - 1$ and $s(L_c) = c$—are discharged by directly invoking their corresponding specialized lemmas. The final identity expressing the superoperator trace as $d\,m(L_c) + (d^2 - d)\,s(L_c)$ is verified by substituting these values and simplifying the resulting polynomial equation.
The first five properties of the shifted generator $L_c$ hold: complete positivity for $t \ge 0$, trace preservation only for $c = 0$, the trace formula $\operatorname{Tr} L_c = c d^2 - d$, and the spectral bounds $m(L_c) = c - 1$ and $s(L_c) = c$.
The goal is a conjunction of six statements. Using the tuple constructor (`refine ⟨...⟩`), the first five conjuncts are supplied directly by previously established theorems: `amp_damping_shift_completely_positive` establishes that $e^{t L_c}$ admits a Kraus representation for $t \ge 0$; `amp_damping_shift_trace_preserving_only_if` shows that trace-preservation at some $t > 0$ forces $c = 0$; `amp_damping_shift_trace` evaluates the superoperator trace as $c d^2 - d$; and `amp_damping_shift_spectral_min_re` and `amp_damping_shift_spectral_max_re` provide the minimal real eigenvalue $c - 1$ and maximal real eigenvalue $c$. The sixth conjunct, expressing the trace as a linear combination of these spectral extrema, is left as the remaining goal.
uses
amp_damping_shift_completely_positive,amp_damping_shift_trace_preserving_only_if,amp_damping_shift_trace,amp_damping_shift_spectral_min_re,amp_damping_shift_spectral_max_rerefine ⟨fun t ht => amp_damping_shift_completely_positive d h c t ht, fun t ht htp => amp_damping_shift_trace_preserving_only_if d h c t ht htp, amp_damping_shift_trace d h c, amp_damping_shift_spectral_min_re d h c, amp_damping_shift_spectral_max_re d h c, ?_⟩
The trace identity $\operatorname{Tr} L_c = d \cdot m(L_c) + (d^2 - d) \cdot s(L_c)$ holds.
We rewrite $\operatorname{Tr} L_c$, $m(L_c)$, and $s(L_c)$ using the formulas proven in the previous lemmas (`amp_damping_shift_trace`, `amp_damping_shift_spectral_min_re`, and `amp_damping_shift_spectral_max_re`), reducing the claim to $c d^2 - d = d (c - 1) + (d^2 - d) c$ in ℂ. Pushing the real-to-complex coercions through the subtraction in $c - 1$ (`push_cast`) leaves a polynomial identity, which is closed by ring arithmetic (`ring`).
uses
amp_damping_shift_trace,amp_damping_shift_spectral_min_re,amp_damping_shift_spectral_max_rerw [amp_damping_shift_trace, amp_damping_shift_spectral_min_re, amp_damping_shift_spectral_max_re] push_cast ring
amp_damping_shift_completely_positive- For any $t \ge 0$, the superoperator $e^{t L_c}$ has a Kraus representation (and is therefore completely positive).
amp_damping_shift_trace_preserving_only_if- If the superoperator $e^{t L_c}$ is trace-preserving for some $t > 0$, then $c = 0$.
amp_damping_shift_trace- The trace of the superoperator $L_c$ acting on $d \times d$ complex matrices is $c d^2 - d$.
amp_damping_shift_spectral_min_re- The minimum of the real parts of the eigenvalues of the superoperator $L_c$ is $c - 1$.
amp_damping_shift_spectral_max_re- The maximum of the real parts of the eigenvalues of the superoperator $L_c$ is $c$.
- Formally, `push_cast` is needed to distribute the coercion from ℝ to ℂ over the subtraction in $c - 1$ so that `ring` can recognise the identity as a polynomial equality over ℂ.
Written by a model (gemini-3.8-flash) from the Lean proof and the goal states Lean reported at each step. Lean verified the proof; this explanation is a reading aid and is not itself machine-checked.
Lean source view module on GitHub
Choose a tactic — a marked line — to see the goals Lean had before and after it, as in Lean’s infoview. Step with ← → or j k.
theorem amp_damping_shift_sharp_spectral_bound (d : ℕ) (h : 2 ≤ d) (c : ℝ) : (∀ t : ℝ, 0 ≤ t → has_kraus_representation d (superop_exp d (amp_damping_generator d h + (c : ℂ) • LinearMap.id) t)) ∧ (∀ t : ℝ, 0 < t → is_trace_preserving_linear_map d (superop_exp d (amp_damping_generator d h + (c : ℂ) • LinearMap.id) t) → c = 0) ∧ LinearMap.trace ℂ (Matrix (Fin d) (Fin d) ℂ) (amp_damping_generator d h + (c : ℂ) • LinearMap.id) = (c : ℂ) * (d : ℂ) ^ 2 - d ∧ spectral_min_re d (amp_damping_generator d h + (c : ℂ) • LinearMap.id) = c - 1 ∧ spectral_max_re d (amp_damping_generator d h + (c : ℂ) • LinearMap.id) = c ∧ LinearMap.trace ℂ (Matrix (Fin d) (Fin d) ℂ) (amp_damping_generator d h + (c : ℂ) • LinearMap.id) = (d : ℂ) * (spectral_min_re d (amp_damping_generator d h + (c : ℂ) • LinearMap.id) : ℂ) + ((d : ℂ) ^ 2 - d) * (spectral_max_re d (amp_damping_generator d h + (c : ℂ) • LinearMap.id) : ℂ) := by refine ⟨fun t ht => amp_damping_shift_completely_positive d h c t ht, fun t ht htp => amp_damping_shift_trace_preserving_only_if d h c t ht htp, amp_damping_shift_trace d h c, amp_damping_shift_spectral_min_re d h c, amp_damping_shift_spectral_max_re d h c, ?_⟩ rw [amp_damping_shift_trace, amp_damping_shift_spectral_min_re, amp_damping_shift_spectral_max_re] push_cast ring