--- title: Amplitude Damping-Affected Quantum Network (AQN) url: https://www.emergentmind.com/topics/amplitude-damping-affected-quantum-network-aqn type: topic --- # Amplitude Damping-Affected Quantum Network (AQN) Searching arXiv for the core AQN paper and closely related amplitude-damping quantum-network work. arXiv search: "Amplitude damping-affected quantum network" Amplitude Damping-Affected Quantum Network (AQN) denotes a quantum network operating under amplitude damping noise, with the term used explicitly for a homogeneous, repeater-based linear quantum network under non-Pauli amplitude damping noise. In that setting, the network state is not fully Bell-diagonal under evolution; instead, the end-to-end description is block-diagonal in the Bell basis and requires four parameters, in contrast to the single-parameter Pauli-twirled approximation. More broadly, the same amplitude-damping framework appears in two-qubit teleportation links, buffered queue-channels, multipartite secret-sharing networks, and higher-dimensional network links, making AQN a common analytical setting for dissipative quantum communication [2509.18064]. ## 1. Noise model and channel-theoretic setting The elementary noise model in an AQN is the single-qubit amplitude-damping channel. With damping parameter \(\gamma\in[0,1]\), its Kraus operators are \[ K_0 = |0\rangle\langle 0| + \sqrt{1-\gamma}\,|1\rangle\langle 1|, \qquad K_1 = \sqrt{\gamma}\,|0\rangle\langle 1|, \] and the induced CPTP map is \[ \rho \mapsto \Lambda_\gamma(\rho)=K_0 \rho K_0^\dagger + K_1 \rho K_1^\dagger. \] Equivalent notation with \(p\in[0,1]\) is also standard: \[ E_{1} = \begin{pmatrix} 1 & 0 \\ 0 & \sqrt{1-p} \end{pmatrix}, \qquad E_{2} = \begin{pmatrix} 0 & \sqrt{p} \\ 0 & 0 \end{pmatrix}. \] For multi-qubit network links, local damping acts independently on each qubit via tensor-product Kraus operators, so a bipartite or repeater-link state evolves by summing over all local Kraus branches [1705.07064]. Several generalized models refine this basic picture. The generalized amplitude damping channel (GADC) \(\Phi_{p,n}\) introduces a mixing parameter \(n\in[0,1]\), with the symmetric case \(n=1/2\) being unital and treating \(|0\rangle\) and \(|1\rangle\) identically. In buffered quantum networks, this leads to a waiting-time dependent channel \(\Phi_{p_{\mathrm{eff}}(W_i),1/2}\), where \(p_{\mathrm{eff}}(w)\) is increasing in the queue waiting time \(w\) [2107.13486]. In discrete-time repeater memories, the same damping can be parameterized by a coherence time \(m^\star\), with \[ \gamma = 1-e^{-1/m^\star}, \qquad \mathcal N^{\circ m}(|1\rangle\langle 1|)=e^{-m/m^\star}|1\rangle\langle 1|, \] and off-diagonal terms decaying as \(e^{-m/(2m^\star)}\) [2509.18064]. A non-perturbative variant arises when each qubit couples to a bosonic bath. There the excitation amplitude \(p(t)\) satisfies the exact integro-differential equation \[ \dot p(t) = -\int_0^t f(t-\tau)\,p(\tau)\,d\tau,\qquad p(0)=1, \] and the reduced dynamics still admits a Kraus form. In the strong-coupling regime, the asymptotic behavior can satisfy \(\lim_{t\to\infty}|p(t)|=|Z|>0\), which is the basis of quenched decoherence under strong amplitude-damping noise [1310.6843]. ## 2. Repeater-chain formulation and the four-parameter link family In the repeater-based linear AQN formulation, an elementary Bell pair \(\sigma_{PQ}(0)=|\Phi^+\rangle\langle\Phi^+|\), with \(|\Phi^+\rangle=(|00\rangle+|11\rangle)/\sqrt2\), remains analytically tractable under local amplitude damping. After \(m\) time steps, setting \[ a=(1-\gamma)^m=e^{-m/m^\star}, \] the Bell-basis representation is \[ \sigma_{PQ}(m) = c_1\,\Phi_{00} + c_2\,\Phi_{01} + c_3\,(\Phi_{01}^{00}+\Phi_{00}^{01}) + \frac{1-c_1-c_2}{2}\,(\Phi_{10}+\Phi_{11}), \] with \[ c_1=\frac{1+a^2}{2}, \qquad c_2=\frac{(1-a)^2}{2}, \qquad c_3=\frac{1-a}{2}. \] This state is block-diagonal in the Bell basis with three independent parameters and is a special case of a four-parameter family [2509.18064]. The full AQN closure property is that arbitrary concatenations of local amplitude-damping noise and entanglement swapping preserve the four-parameter family \[ \rho(\{h_i\}_{i=1}^4) = h_1\,\Phi_{00} + h_2\,\Phi_{01} + h_3\,(\Phi_{01}^{00}+\Phi_{00}^{01}) + h_4\,(\Phi_{11}^{10}+\Phi_{10}^{11}) + \frac{1-h_1-h_2}{2}\,(\Phi_{10}+\Phi_{11}). \] Operationally, the simulation method keeps track of these four parameters for each entangled link, together with the number of times noise acts on it, i.e., its age, until it is consumed for swapping [2509.18064]. If such a link is stored for an additional \(n\) steps, letting \(t=(1-\gamma)^n\), the noise update is \[ \begin{aligned} \tilde h_{1,2} &= \frac14\Bigl[ h_1\bigl(1\pm 2t +(1-t)^2+t^2\bigr) +h_2\bigl(1\mp 2t +(1-t)^2+t^2\bigr) \\ &\qquad\quad +2h_3\bigl(1+t^2-(1-t)^2\bigr) +4h_4\,t(1-t) \Bigr],\\ \tilde h_3 &= \frac12(1-t)+t\,h_3,\qquad \tilde h_4=t\,h_4, \end{aligned} \] where the upper sign gives \(\tilde h_1\) and the lower sign gives \(\tilde h_2\). If two such links are swapped at an intermediate node, the resulting link remains in the same family with swap-update \[ \begin{aligned} h_{1,2} &= 4(s_1+s_2\pm 2s_3),\qquad h_{3,4}=4(s_4\pm s_5),\\ s_1 &= \frac1{16},\qquad s_2=\frac1{16}[1+2(f_1+f_2)][1+2(g_1+g_2)],\\ s_3 &= \frac1{16}(f_1-f_2)(g_1-g_2),\qquad s_4=\frac18(f_3+f_4),\\ s_5 &= \frac18(f_3-f_4)[1+2(g_1+g_2)]. \end{aligned} \] This closed algebraic structure is the main technical distinction between AQN and Pauli-twirled models [2509.18064]. The principal performance metrics are the Bell fidelity and concurrence: \[ F=\langle \Phi^+|\rho|\Phi^+\rangle = h_1, \] and \[ C(\rho) = \max\Bigl\{ 0,\; |h_1-h_2| -\sqrt{(1-h_1-h_2)^2-4h_4^2} \Bigr\}. \] In a linear chain, the final \(\{h_i\}\) are obtained by iterating elementary generation, noise updates, and swaps under a chosen policy, after which one reports average fidelity \(\bar F\) and average concurrence \(\overline C\) over Monte-Carlo realizations [2509.18064]. ## 3. Teleportation, entanglement distribution, and long-time behavior Amplitude damping directly constrains teleportation over network links. For two-qubit teleportation, one can use either a maximally entangled Bell channel or a nonmaximally entangled Bell-like channel \[ |Q\rangle = (\cos\theta_1|00\rangle+\sin\theta_1|11\rangle)_{A_1B_1} \otimes (\cos\theta_2|00\rangle+\sin\theta_2|11\rangle)_{A_2B_2}. \] For an arbitrary two-qubit pure input, the average teleportation fidelity is \[ \langle F\rangle = \frac15 + \frac15 \prod_{k=1}^{2} \Bigl[ 1 + (2p_Ap_B-p_A-p_B)\sin^2\theta_k +\sqrt{(1-p_A)(1-p_B)}\,\sin 2\theta_k \Bigr]. \] Maximizing over \(\theta_k\) yields \[ \theta_k^{\mathrm{opt}} = \frac12 \arctan \frac{2\sqrt{(1-p_A)(1-p_B)}}{p_A+p_B-2p_Ap_B}, \qquad k=1,2, \] and substitution gives the optimized closed-form fidelity. In particular, the optimized Bell-like channel outperforms both the add-noise strategy of X. Hu et al. and the classical limit \(2/5\) for all \(p\), while remaining deterministic with \(100\%\) success probability because the channel-selection method is trace-preserving and uses no postselection [1705.07064]. This result corrects a recurrent misconception in amplitude-damping teleportation: adding more amplitude damping to more qubits need not be the preferred strategy. The paper explicitly states that choosing an appropriate quantum channel “enhances the ability of teleportation better and negates the fact that more amplitude damping noise more quality” [1705.07064]. In a different dynamical regime, strong non-Markovian amplitude damping can preserve useful channel quality at long times. For two independent qubits initially in \(|\Phi^+\rangle\), the joint state remains X-shaped, with concurrence \[ C(t)=|p(t)|^4, \] and teleportation fidelity \[ F(t)=\frac13\Bigl[1+\max\{1+|p|^4,\; |p|^2 q^2\}\Bigr], \qquad q(t)=\sqrt{1-|p(t)|^2}. \] In weak coupling, \(p(t)\to 0\), so \(C\to 0\) and \(F\to 2/3\), the classical limit. In strong coupling, \(p(t)\to Z\), so \[ C(\infty)=|Z|^4>0,\qquad F(\infty)=\frac23+\frac{|Z|^4}{3}> \frac23. \] For \(\omega_0=0.1\omega_c\), super-Ohmic \(s=3\), and \(\eta_0=0.5\), the reported value \(Z\approx 0.84\) gives \(C(\infty)\approx 0.50\) and \(F(\infty)\approx 0.82\), demonstrating finite entanglement and better-than-classical teleportation fidelity at long times [1310.6843]. ## 4. Queueing, memory, and non-i.i.d. network noise AQNs need not be memoryless. In a buffered network, qubits experience waiting-time dependent generalized amplitude damping before service. The GAD queue-channel is defined on a \(G/G/1\) queue with FCFS discipline, where arrival times \(A_i\) are i.i.d. with mean \(1/\lambda\), service times \(S_i\) are i.i.d. with mean \(1/\mu\), \(\lambda<\mu\) for stability, and waiting times satisfy Lindley’s recursion \[ W_{i+1} = \max\{W_i-A_i,0\}+S_{i+1}. \] Each qubit then experiences \(\Phi_{p_{\mathrm{eff}}(W_i),1/2}\), so the resulting noise is non-i.i.d. because consecutive waiting times are correlated [2107.13486]. Conditioned on the entire waiting-time sequence, however, the channel acts independently across qubits, and because the symmetric GADC is additive, the exact classical capacity of the queue-channel is \[ C_{\mathrm{queue}} = \lambda\,\mathbb E_{W\sim \pi} \Bigl[ 1-h_2\Bigl(\frac{1-\sqrt{1-p_{\mathrm{eff}}(W)}}{2}\Bigr) \Bigr], \] where \(\pi\) is the stationary waiting-time distribution. In the common physical model \(p_{\mathrm{eff}}(w)=1-e^{-\kappa w}\), the induced binary symmetric crossover is \[ q(W)=\frac12(1-e^{-\kappa W/2}). \] The design trade-off is explicit: if \(\lambda\to\mu^{-}\), mean waiting diverges and capacity goes to zero; if \(\lambda\ll\mu\), waiting is negligible but throughput is small, so there is an optimal \(\lambda^\star\in(0,\mu)\) maximizing the rate. For fixed mean service or arrival rates, deterministic service in \(M/G/1\) and deterministic arrivals in \(G/M/1\) maximize \(C_{\mathrm{queue}}\) by minimizing waiting variability [2107.13486]. A different memory model arises when a train of qubits interacts sequentially with a damped harmonic oscillator through a Jaynes-Cummings coupling. This memory amplitude-damping channel is forgetful, so the standard coding theorems apply. With oscillator relaxation time \(\tau_d=1/\Gamma\), inter-use spacing \(\tau\), and memory parameter \[ \mu=\frac{\tau_d}{\tau+\tau_d}, \] the channel interpolates between the memoryless limit \(\mu\to 0\) and strong memory \(\mu\to 1\). Numerical analysis of two uses shows that memory effects improve both coherent-information and Holevo rates over the memoryless approximation, and dephasing the oscillator after the first use removes this advantage, showing that qubit-oscillator entanglement is responsible for the gain [1103.0747]. These results establish that amplitude-damping network analysis cannot, in general, be reduced to i.i.d. link noise. Waiting times, finite-memory oscillators, and structured environmental feedback materially alter transmission rates and link behavior [2107.13486]. ## 5. Protection, recovery, and purification protocols inside an AQN Several protocols aim to mitigate amplitude damping on network links without replacing the underlying AQN model. One class uses weak measurement followed by measurement reversal. For a single qubit, the weak-measurement and reversal operators are \[ M_w(p)=\mathrm{diag}(1,\sqrt{1-p}), \qquad M_r(p_r)=\mathrm{diag}(\sqrt{1-p_r},1), \] with reversal strength \[ p_r=(1-\gamma)p+\gamma. \] For two qubits on a symmetric link, one uses \(M_w^{(2)}=M_w(p)\otimes M_w(p)\) and \(M_r^{(2)}=M_r(p_r)\otimes M_r(p_r)\). Starting from \(|\Phi\rangle=\alpha|00\rangle+\beta|11\rangle\), the total success probability is \[ P_{\mathrm{succ}} = \bigl[|\alpha|^2+(1-p)^2|\beta|^2\bigr] \cdot \bigl[|\alpha|^2+(1-p_r)^2(1-\gamma)^2|\beta|^2\bigr]. \] For the maximally entangled case \(\alpha=\beta=1/\sqrt2\), the paper reports representative improvements in both entropic and geometric discord across \(\gamma\in[0,1]\), while noting the trade-off that larger \(p\) protects better but lowers \(P_{\mathrm{succ}}\). In an AQN deployment, each sender applies \(M_w(p)\) before transmission, each receiver applies \(M_r(p_r)\) immediately after reception, and a classical side-channel communicates the chosen \(p\) so that \(p_r\) can be set adaptively [1508.00972]. A second protocol uses only Hadamard and CNOT gates. For an input \[ |\psi_{\mathrm{in}}\rangle=\alpha|00\rangle+\beta|01\rangle+\gamma|10\rangle+\delta|11\rangle, \] independent amplitude damping yields concurrence \[ C_d(p)=\max\{0,\,2q[|\alpha\delta-\beta\gamma|-p|\delta|^2]\},\qquad q=1-p. \] After introducing two ancillas in \(|0\rangle_A\), applying \[ H_\theta= \begin{bmatrix} \cos\theta & -\sin\theta\\ \sin\theta & \cos\theta \end{bmatrix}, \qquad \theta=\arctan(1/\sqrt q), \] followed by two CNOTs and postselection on ancilla outcome \(|00\rangle_A\), the restored concurrence becomes \[ C_r(p)= \max\left\{ 0,\; \frac{2[|\alpha\delta-\beta\gamma|-p|\delta|^2]} {1+p(1+|\delta|^2-|\alpha|^2)+p^2|\delta|^2} \right\}. \] The “good” outcome probability is \(P_{\mathrm{succ}}=P_{00}(p)\), with \[ P_{00}(p)=\left[\frac{q}{N_2(1+q)}\right]^2, \qquad N_2=1+p(|\beta|^2+|\gamma|^2+2|\delta|^2)+p^2|\delta|^2. \] The scheme is explicitly probabilistic, but the paper contrasts it with weak-measurement reversal by emphasizing that it uses only coherent Hadamard and CNOT gates and no weak measurement in the reversal process [1208.4187]. A third protocol performs purification by postselecting the no-jump branch. With one ancilla initialized in \(|0\rangle\), an \(H\) on the ancilla, a CZ between ancilla and system, and ancilla measurement in the computational basis, the retained outcome \(|0\rangle_A\) collapses the system to \(E_0|\psi\rangle\). For \(|\psi\rangle=\alpha|0\rangle+\beta|1\rangle\), the success probability is \[ p_0 = |\alpha|^2+(1-\gamma)|\beta|^2, \] and the post-selected state is \[ |\psi'\rangle = \alpha|0\rangle+\sqrt{1-\gamma}\,\beta|1\rangle. \] The initial and final fidelities satisfy \[ F_{\mathrm{out}} = \frac{\bigl(|\alpha|^2+\sqrt{1-\gamma}\,|\beta|^2\bigr)^2} {|\alpha|^2+(1-\gamma)|\beta|^2}, \] and the paper states that \(F_{\mathrm{out}}>F_{\mathrm{in}}\) for all \(0<\gamma<1\). The same construction extends to channel purification through the Choi–Jamiołkowski isomorphism, with one or two ancillas and two Clifford gates per purified qubit or Choi pair [2509.05709]. ## 6. Multipartite, higher-dimensional, and capacity-oriented extensions AQNs are not limited to bipartite repeater chains. In a multipartite secret-sharing setting, a four-qubit GHZ state \[ |GHZ\rangle_{ABDD}=\frac{|0000\rangle+|1111\rangle}{\sqrt2} \] is distributed asymmetrically, with Alice and Bob each holding one qubit and Dennis holding two. In the ideal channel, the protocol decodes a secret two-bit message with unit probability in one execution, using a globally operated quantum teleportation operator. Under amplitude damping on the transmitted qubits, \[ \rho_i'=\sum_{u,v=0,1}(E_u\otimes E_v\otimes I\otimes I)\,\rho_i\,(E_u\otimes E_v\otimes I\otimes I)^\dagger, \] the task remains possible through an optimization algorithm based on parameterized POVMs. The paper also reports channel-quality measures, including branch fidelities \[ F_1=F_2=F_7=F_8=\sqrt{1-\gamma+\gamma^2/4}, \qquad F_3=F_4=F_5=F_6=\sqrt{1-\gamma}, \] the \(l_1\)-norm coherence \(C_{l_1}(\rho_i')=|1-\gamma|\), and closed-form relative-entropy coherence expressions [1712.10219]. Higher-dimensional amplitude damping leads to multi-level amplitude damping (MAD) channels on \(\mathbb C^d\), with Kraus operators \[ K_{ij}=\sqrt{\gamma_{ji}}\,|i\rangle\langle j|, \qquad K_0=|0\rangle\langle 0|+\sum_{j=1}^{d-1}\sqrt{1-\xi_j}\,|j\rangle\langle j|, \] where \(\xi_j=\sum_{0\le i1/2\) is also crossed at lower \(p_l\) in AQN, and heat maps reveal “absolute-advantage” regions in \((m^\star,p_l)\) where TAQN fails while AQN succeeds in distributing end-to-end entanglement. The advantage persists in nine-node chains, although the \(\mathcal I=100\%\) region shrinks [2509.18064]. Two additional misconceptions are addressed by the broader amplitude-damping literature. First, amplitude damping is not uniformly detrimental in every structured setting: strong coupling to a bath can quench decoherence and preserve finite entanglement and teleportation fidelity above the classical limit at long times [1310.6843]. Second, memory is not always harmful: both the GAD queue-channel and the damped-oscillator memory channel show that transmission rates depend on structured waiting-time statistics and inter-use correlations, and in the Jaynes-Cummings model memory can improve both classical and quantum transmission rates [2107.13486]. Taken together, these results define AQN as a non-Pauli network model in which local dissipation, storage age, queueing, swapping policy, and environment structure all affect the effective state space and performance metrics. The main technical lesson is that amplitude damping generally cannot be compressed into a Bell-diagonal or purely i.i.d. description without discarding physically relevant degrees of freedom, and the main practical lesson is that policy optimization, protection schemes, and capacity estimation should be carried out in the native amplitude-damping model whenever coherence time or elementary-link success probability is moderate or small [2509.18064]. Source: https://www.emergentmind.com/topics/amplitude-damping-affected-quantum-network-aqn