Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

Mode-mixing threshold for quadratic error suppression in entanglement swapping

Annan Chen

Published Oct 29, 2026• 0 citations
Publisher

Abstract

Partial distinguishability limits linear-optical entanglement swapping. For pure data packets with overlap $0 < c < 1$, static, internal-state-blind passive optics and spatial photon-number-resolving detection leave an unassisted conditional Bell error of at least $\epsilon = (1-c^2)/2$. We prove that a single independent auxiliary photon (the helper) cannot eliminate this first-order error using three or fewer two-mode mixings: the error remains at least $\epsilon/(1+c)$, even with an optimized helper, finite vacuum extensions and deterministic single-copy local operations and classical communication on the retained qubits. A fixed five-mode circuit with four mixings beats this bound for every $c$. Preparing the helper in the normalized sum of the phase-aligned data packets yields a selected record with probability $(2-c+2c^2)/72$ per supplied trial and error $(1-c)^2/(2-c+2c^2) = \epsilon^2/3 + O(\epsilon^3)$. At $c = 0.99$ (two-photon visibility 0.9801), the error is $3.4 \times 10^{-5}$, compared with $1.0 \times 10^{-2}$ without a helper and $5.0 \times 10^{-3}$ for the best three-mixing circuit. A non-polarizing beam splitter, a partially polarizing beam splitter and a half-wave plate implement the four mixings in polarization encoding. The target Bell weight is proportional to the fully symmetric three-photon internal-state weight. For symmetric depolarized-qubit sources, we derive the exact helper-optimized three-mixing ceiling and a unique impurity threshold for each $0 < c < 1$; stochastic impurity contributes at first order. We also give the fixed-instrument acceptance frontier, the calibration floor, and comparisons with two helpers and internal-mode filtering, which achieves higher heralding rates in the Gaussian scenarios studied.

DOI
10.5281/zenodo.23840815

Related papers

Fidelity limits and purity requirements for entanglement swapping with one auxiliary photon

Annan Chen

Internal-state mismatch limits heralded entanglement swapping. We determine the limits of using one independent auxiliary photon (a helper) with passive optics acting identically on every internal state, complete photon counting and deterministic local recovery. For pure data states of overlap $0<c<1$, put $t=1-c$. Optimizing over the helper and all allowed finite networks reduces the minimum Bell error from $(1-c^2)/2\sim t$ without a helper to $t^2/(2-c+2c^2)\sim t^2/3$. Attaining this minimum requires a pure midpoint helper and three equally spaced logical Bloch directions (a trine); the largest single-record probability at equality is $(2-c+2c^2)/72$. For a mixed helper supported on the data-state span, let its eigenvalues be $(1\pm r)/2$ and $\eta=1-r$. Under a complete-record probability floor $p\ge p_0$, with fixed $0<p_0<1/64$, the optimal error has order $\min\{t^2+\eta,t\sqrt{t+\eta}\}$ as $t,\eta\to0$. Quadratic suppression is possible precisely when $\eta=O(t^2)$, requiring an ideal Hong--Ou--Mandel helper self-visibility deficit of quadratic or smaller order in the data-visibility deficit. When $\eta$ is proportional to $t$, exchange-invariant helpers leave linear error, whereas a tilted helper and a real eight-mode network attain the optimal order $t^{3/2}$. The converse follows from compatibility of the optical amplitudes shared by all photon assignments. The same order holds for local unitaries, deterministic local operations and classical communication, and deterministic separable recovery, and for the average over a bounded number of accepted records at nonvanishing total acceptance probability.

•Nov 5, 2026•0 citations

Optimal entanglement swapping with one auxiliary photon: exact fidelity limit and purity requirements

Annan Chen

Abstract Wave-packet mismatch in entanglement swapping leaves which-source information and lowers the fidelity of the heralded remote pair. We determine the exact correction achievable with one independent auxiliary photon. For pure data photons with amplitude overlap c, every finite static passive network acting identically on internal states, followed by photon counting and deterministic local recovery, yields conditional Bell fidelity at most (1 + c + c2)/(2 − c + 2c2). A five-mode circuit with four two-mode mixings attains this limit and converts first-order mismatch error to second order, whereas circuits with at most three mixings in the logical-mode basis cannot exceed (1 + c)/2. At two-photon visibility 0.98, a pure auxiliary reduces the error from 1.0 × 10−2 to 3.4 × 10−5 at heralding probability 0.041. For mixed auxiliary states supported on the span of the data wave packets, quadratic suppression is possible if and only if the purity deficit vanishes at least quadratically in the mismatch. We establish a matching error law at fixed positive heralding probability. With data visibility 0.90 and auxiliary-copy visibility 0.99, the aligned circuit reduces Bell error 6.6-fold. A centered Gaussian auxiliary of optimized width preserves the optimal leading error.

•Nov 13, 2026•0 citations

Multipair-resilient entanglement swapping with complementary linear-optical measurements

Siavash Mirzaei Ghormish, Ryan M. Camacho

Multipair emission is a principal source of false heralds in entanglement swapping with spontaneous parametric down-conversion sources. A conventional Bell-state measurement (BSM) cannot distinguish the desired arrival of one photon from each neighboring source from a mixed-polarization double emission by one source accompanied by vacuum from the other. We show that a cascaded network can resolve this ambiguity by allowing its successive swapping stations to perform different measurements. A direct-basis BSM rejects same-polarization double emissions, whereas a balanced equatorial analyzer, implemented by a four-mode Green Machine (GM), uses two-photon interference to reject mixed-polarization double emissions. At the same time, it recovers same-polarization inter-source events rejected by the BSM as resolved $φ$-type Bell heralds. Their rejection sets jointly cover both single-source two-photon classes while retaining contributions from both useful inter-source classes. We classify the passive four-mode, number-resolving analyzers satisfying this condition and identify the BSM and GM as canonical balanced representatives. Using a Gaussian-state analysis exact to all orders of multipair emission, we evaluate three-source BSM--GM and four-source BSM--GM--BSM Pure Bell Pair sources under coupling, detector, and channel loss. The three-source network suppresses the leading infidelity, while the four-source network heralds an exact Bell state in the lossless limit, a property that holds for every alternating BSM--GM chain with four or more sources. The fidelity advantage over source-count-matched all-BSM networks persists at every loss level studied, a single-channel rate advantage appears without multiplexing, and spectral multiplexing raises the Bell-pair delivery probability toward the asymptotically deterministic limit.

quant-pharXiv:2608.25211•Aug 27, 2026•1 citation

Arbitrary-order suppression of source mismatch in photonic entanglement swapping

Annan Chen

Abstract We establish the minimum number of same-source photons needed for arbitrary-order suppression of source mismatch in photonic entanglement swapping and construct a static passive circuit family that attains it. The result applies to unencoded photons, optics that act identically on all internal states, complete photon-number-resolving detection, and local-unitary recovery of the retained qubits. For two pure source wave packets with squared overlap $u$, suppressing the bare mismatch error $\epsilon=(1-u)/2$ to order $\epsilon^r$ at nonzero ideal success probability requires at least $r$ photons from each source. A cyclic interferometer attains the minimum total of $2r$ photons at every order, with ideal acceptance $2^{2-3r}$. An exact transfer expression extends the calculation to distinct wave packets and correlated classical source noise. Near a common wave packet, the leading error is proportional to the summed delayed Hong--Ou--Mandel losses along a closed tour through each source's emissions. For equally spaced emissions, losses that increase concavely with delay select chronological wiring, whereas convex growth selects a folded order; independent fluctuations add the same leading penalty to every wiring. In a Gaussian drift model, folding lowers the smallest reachable error in the compared catalogs by about $23\%$ without adding photons. Including detection, storage, and routing losses, the photon-cost advantage depends on the target and is largest in a narrow target window; at target error $4.35\times10^{-4}$, the pooled cost is lower by a factor of about $5.6$ with equal acceptance and randomization options in both catalogs. A common spectral filter is cheaper in this finite-noise example; a separate pure-source analysis gives a regime with a better cyclic cost exponent.

•Nov 12, 2026•0 citations

Auxiliary Schmidt Rank as a Resource for Photonic Bell Measurements

Pradip Laha, Peter van Loock

In quantum communication and fusion-based quantum computation, photonic Bell measurements are fundamentally limited when only passive linear optics is employed. While for qubits, some Bell states can be unambiguously identified with static beam splitters and no extra photons or entanglement, additional auxiliary photons or at least additional auxiliary degrees of freedom with a certain level of additional entanglement are needed to approach or attain a complete, deterministic Bell measurement. Here, we prove an exact resource threshold when the same two photons carry system qudits of dimension $d$ and a fixed auxiliary entangled state $Φ$, possibly distributed over several additional degrees of freedom, with total Schmidt rank $r_Φ$. We show that a single conclusive Bell-label functional can occur for $r_Φ\geqslant\lceil d/2\rceil$, but deterministic discrimination of all $d^2$ Bell-state labels requires $r_Φ\geqslant d$. A maximally entangled rank-$d$ auxiliary state achieves the bound by local Bell-basis sorting between each photon's system and auxiliary degrees of freedom. Thus, the auxiliary Schmidt rank is a certified resource for ancilla-photon-free, embedded photonic Bell measurements.

quant-pharXiv:2606.24591•Jun 24, 2026•2 citations

Classical communication with an exact photon budget: finite-block limits and optimal input ensembles

Zixuan He

What can be transmitted with a fixed photon supply? An optical source that supplies exactly K photons across n modes poses a concrete communication problem: how many messages can it transmit at a chosen error probability, and which input states make the best use of those photons? This paper gives directly evaluable coding bounds for pure loss and identifies the exact optimal input ensembles for identical noisy Gaussian channels. The results allow researchers to assess finite optical blocks, quantify the information cost of fixing the photon supply, and determine when entanglement between separately constrained blocks can be dispensed with. Every message is supported on the same exact K-photon sector. Within that sector, encoding may use arbitrary mixed states and entanglement among modes, and the receiver may perform a joint measurement. Communication is unassisted. All logarithms below are natural; information is measured in nats. Finite-message bounds and the optimal third-order rate For n ≥ 2 identical pure-loss channels of transmissivity 0 < η < 1 and K ≥ 1, the number L of surviving photons has the same binomial distribution for every encoded state. Define \[p_\ell=\binom K\ell\eta^\ell(1-\eta)^{K-\ell},\qquad d_{n,\ell}=\binom{n+\ell-1}{\ell}.\] If ps*(M;n,K) is the best average success probability for M equiprobable messages, then \[\sum_{\ell=0}^{K}p_\ell\left[1-\frac{M-1}{d_{n,\ell}}\right]_+\le p_{\mathrm s}^{*}(M;n,K)\le\sum_{\ell=0}^{K}p_\ell\min\!\left\{1,\frac{d_{n,\ell}}{M}\right\}.\] Here [x]+ = max{x,0}. The lower bound provides an achievable guarantee; the upper bound limits every admissible code. Evaluating both sums brackets the performance at a specified photon supply and block length. Their proof has a simple physical interpretation: surviving photons determine the available output dimension, while one collective-mode codebook works across all possible survivor counts. For fixed E > 0, fixed error tolerance 0 < ε < 1, Kn = ⌊nE⌋ and en = Kn/n, the bounds determine the largest message count through third order: \[\log M^*(n,K_n,\varepsilon)=ng(\eta e_n)+\sqrt{n\eta(1-\eta)e_n}\,h(\eta e_n)\Phi_{\mathrm N}^{-1}(\varepsilon)-\tfrac12\log n+O(1).\] Here g(x) = (x+1)log(x+1) − x log x, h(x) = log(1+1/x), and ΦN is the standard normal distribution function. This fixes both the fluctuation scale and the optimal −½ log n correction. The finite bounds also yield the exact decay exponent of success above capacity and a matching error exponent on a specified interval below capacity. Which inputs are optimal under Gaussian noise? For n identical nonidentity gauge-covariant Gaussian channels, with thermal photon means transformed as r ↦ ar+b, a > 0 and b ≥ max{0,a−1}, the paper determines the exact fixed-photon Holevo optimum and every ensemble attaining it. Each optimal letter places all K photons in one collective mode: \[|K,u\rangle=\frac{\left(\sum_{j=1}^{n}u_j a_j^\dagger\right)^K}{\sqrt{K!}}|0\rangle,\qquad \sum_j|u_j|^2=1.\] An ensemble is optimal precisely when almost all its letters have this form and its average is uniform on the K-photon sector. A finite weighted optimal ensemble exists. These collective modes may themselves extend across, and be entangled among, physical optical modes. Thus the characterization identifies the structure to optimize without imposing a Gaussian-input restriction. The central proof step combines different loss histories that end in the same output photon sector. A common ordering of the comparison eigenstates allows the separate majorization bounds to be combined. For pure loss, a quantitative rigidity estimate further converts a small Holevo deficit into closeness to collective-mode number states. Resource cost and communication across blocks At fixed positive photon density and fixed nonidentity Gaussian noise, requiring an exact photon supply incurs a Holevo penalty of ½ log n plus an explicitly determined constant, relative to the usual mean-energy value. The paper separates this cost into photon-count uncertainty and the mismatch from a thermal output distribution. For separately constrained blocks of quantum-limited loss, quantum-limited amplification or entanglement-breaking Gaussian noise, the one-block Holevo values add exactly. Every optimal letter is a product of blockwise collective-mode number states, and the repeated-block unassisted classical capacity equals the one-block Holevo value. The low-rate reliability interval outside the matching range and repeated-block additivity for intermediate Gaussian noise remain open.

•Oct 13, 2026•0 citations
Partial distinguishability limits linear-optical entanglement swapping. For pure data packets with overlap 0<c<10 < c < 10<c<1, static, internal-state-blind passive optics and spatial photon-number-resolving detection leave an unassisted conditional Bell error of at least ϵ=(1−c2)/2\epsilon = (1-c^2)/2ϵ=(1−c2)/2. We prove that a single independent auxiliary photon (the helper) cannot eliminate this first-order error using three or fewer two-mode mixings: the error remains at least ϵ/(1+c)\epsilon/(1+c)ϵ/(1+c), even with an optimized helper, finite vacuum extensions and deterministic single-copy local operations and classical communication on the retained qubits. A fixed five-mode circuit with four mixings beats this bound for every ccc. Preparing the helper in the normalized sum of the phase-aligned data packets yields a selected record with probability (2−c+2c2)/72(2-c+2c^2)/72(2−c+2c2)/72 per supplied trial and error (1−c)2/(2−c+2c2)=ϵ2/3+O(ϵ3)(1-c)^2/(2-c+2c^2) = \epsilon^2/3 + O(\epsilon^3)(1−c)2/(2−c+2c2)=ϵ2/3+O(ϵ3). At c=0.99c = 0.99c=0.99 (two-photon visibility 0.9801), the error is 3.4×10−53.4 \times 10^{-5}3.4×10−5, compared with 1.0×10−21.0 \times 10^{-2}1.0×10−2 without a helper and 5.0×10−35.0 \times 10^{-3}5.0×10−3 for the best three-mixing circuit. A non-polarizing beam splitter, a partially polarizing beam splitter and a half-wave plate implement the four mixings in polarization encoding. The target Bell weight is proportional to the fully symmetric three-photon internal-state weight. For symmetric depolarized-qubit sources, we derive the exact helper-optimized three-mixing ceiling and a unique impurity threshold for each 0<c<10 < c < 10<c<1; stochastic impurity contributes at first order. We also give the fixed-instrument acceptance frontier, the calibration floor, and comparisons with two helpers and internal-mode filtering, which achieves higher heralding rates in the Gaussian scenarios studied.
Internal-state mismatch limits heralded entanglement swapping. We determine the limits of using one independent auxiliary photon (a helper) with passive optics acting identically on every internal state, complete photon counting and deterministic local recovery. For pure data states of overlap 0<c<10<c<10<c<1, put t=1−ct=1-ct=1−c. Optimizing over the helper and all allowed finite networks reduces the minimum Bell error from (1−c2)/2∼t(1-c^2)/2\sim t(1−c2)/2∼t without a helper to t2/(2−c+2c2)∼t2/3t^2/(2-c+2c^2)\sim t^2/3t2/(2−c+2c2)∼t2/3. Attaining this minimum requires a pure midpoint helper and three equally spaced logical Bloch directions (a trine); the largest single-record probability at equality is (2−c+2c2)/72(2-c+2c^2)/72(2−c+2c2)/72. For a mixed helper supported on the data-state span, let its eigenvalues be (1±r)/2(1\pm r)/2(1±r)/2 and η=1−r\eta=1-rη=1−r. Under a complete-record probability floor p≥p0p\ge p_0p≥p0​, with fixed 0<p0<1/640<p_0<1/640<p0​<1/64, the optimal error has order min⁡{t2+η,tt+η}\min\{t^2+\eta,t\sqrt{t+\eta}\}min{t2+η,tt+η​} as t,η→0t,\eta\to0t,η→0. Quadratic suppression is possible precisely when η=O(t2)\eta=O(t^2)η=O(t2), requiring an ideal Hong--Ou--Mandel helper self-visibility deficit of quadratic or smaller order in the data-visibility deficit. When η\etaη is proportional to ttt, exchange-invariant helpers leave linear error, whereas a tilted helper and a real eight-mode network attain the optimal order t3/2t^{3/2}t3/2. The converse follows from compatibility of the optical amplitudes shared by all photon assignments. The same order holds for local unitaries, deterministic local operations and classical communication, and deterministic separable recovery, and for the average over a bounded number of accepted records at nonvanishing total acceptance probability.
Abstract Wave-packet mismatch in entanglement swapping leaves which-source information and lowers the fidelity of the heralded remote pair. We determine the exact correction achievable with one independent auxiliary photon. For pure data photons with amplitude overlap c, every finite static passive network acting identically on internal states, followed by photon counting and deterministic local recovery, yields conditional Bell fidelity at most (1 + c + c2)/(2 − c + 2c2). A five-mode circuit with four two-mode mixings attains this limit and converts first-order mismatch error to second order, whereas circuits with at most three mixings in the logical-mode basis cannot exceed (1 + c)/2. At two-photon visibility 0.98, a pure auxiliary reduces the error from 1.0 × 10−2 to 3.4 × 10−5 at heralding probability 0.041. For mixed auxiliary states supported on the span of the data wave packets, quadratic suppression is possible if and only if the purity deficit vanishes at least quadratically in the mismatch. We establish a matching error law at fixed positive heralding probability. With data visibility 0.90 and auxiliary-copy visibility 0.99, the aligned circuit reduces Bell error 6.6-fold. A centered Gaussian auxiliary of optimized width preserves the optimal leading error.
Multipair emission is a principal source of false heralds in entanglement swapping with spontaneous parametric down-conversion sources. A conventional Bell-state measurement (BSM) cannot distinguish the desired arrival of one photon from each neighboring source from a mixed-polarization double emission by one source accompanied by vacuum from the other. We show that a cascaded network can resolve this ambiguity by allowing its successive swapping stations to perform different measurements. A direct-basis BSM rejects same-polarization double emissions, whereas a balanced equatorial analyzer, implemented by a four-mode Green Machine (GM), uses two-photon interference to reject mixed-polarization double emissions. At the same time, it recovers same-polarization inter-source events rejected by the BSM as resolved φφφ-type Bell heralds. Their rejection sets jointly cover both single-source two-photon classes while retaining contributions from both useful inter-source classes. We classify the passive four-mode, number-resolving analyzers satisfying this condition and identify the BSM and GM as canonical balanced representatives. Using a Gaussian-state analysis exact to all orders of multipair emission, we evaluate three-source BSM--GM and four-source BSM--GM--BSM Pure Bell Pair sources under coupling, detector, and channel loss. The three-source network suppresses the leading infidelity, while the four-source network heralds an exact Bell state in the lossless limit, a property that holds for every alternating BSM--GM chain with four or more sources. The fidelity advantage over source-count-matched all-BSM networks persists at every loss level studied, a single-channel rate advantage appears without multiplexing, and spectral multiplexing raises the Bell-pair delivery probability toward the asymptotically deterministic limit.
Abstract We establish the minimum number of same-source photons needed for arbitrary-order suppression of source mismatch in photonic entanglement swapping and construct a static passive circuit family that attains it. The result applies to unencoded photons, optics that act identically on all internal states, complete photon-number-resolving detection, and local-unitary recovery of the retained qubits. For two pure source wave packets with squared overlap uuu, suppressing the bare mismatch error ϵ=(1−u)/2\epsilon=(1-u)/2ϵ=(1−u)/2 to order ϵr\epsilon^rϵr at nonzero ideal success probability requires at least rrr photons from each source. A cyclic interferometer attains the minimum total of 2r2r2r photons at every order, with ideal acceptance 22−3r2^{2-3r}22−3r. An exact transfer expression extends the calculation to distinct wave packets and correlated classical source noise. Near a common wave packet, the leading error is proportional to the summed delayed Hong--Ou--Mandel losses along a closed tour through each source's emissions. For equally spaced emissions, losses that increase concavely with delay select chronological wiring, whereas convex growth selects a folded order; independent fluctuations add the same leading penalty to every wiring. In a Gaussian drift model, folding lowers the smallest reachable error in the compared catalogs by about 23%23\%23% without adding photons. Including detection, storage, and routing losses, the photon-cost advantage depends on the target and is largest in a narrow target window; at target error 4.35×10−44.35\times10^{-4}4.35×10−4, the pooled cost is lower by a factor of about 5.65.65.6 with equal acceptance and randomization options in both catalogs. A common spectral filter is cheaper in this finite-noise example; a separate pure-source analysis gives a regime with a better cyclic cost exponent.
In quantum communication and fusion-based quantum computation, photonic Bell measurements are fundamentally limited when only passive linear optics is employed. While for qubits, some Bell states can be unambiguously identified with static beam splitters and no extra photons or entanglement, additional auxiliary photons or at least additional auxiliary degrees of freedom with a certain level of additional entanglement are needed to approach or attain a complete, deterministic Bell measurement. Here, we prove an exact resource threshold when the same two photons carry system qudits of dimension ddd and a fixed auxiliary entangled state ΦΦΦ, possibly distributed over several additional degrees of freedom, with total Schmidt rank rΦr_ΦrΦ​. We show that a single conclusive Bell-label functional can occur for rΦ⩾⌈d/2⌉r_Φ\geqslant\lceil d/2\rceilrΦ​⩾⌈d/2⌉, but deterministic discrimination of all d2d^2d2 Bell-state labels requires rΦ⩾dr_Φ\geqslant drΦ​⩾d. A maximally entangled rank-ddd auxiliary state achieves the bound by local Bell-basis sorting between each photon's system and auxiliary degrees of freedom. Thus, the auxiliary Schmidt rank is a certified resource for ancilla-photon-free, embedded photonic Bell measurements.
What can be transmitted with a fixed photon supply? An optical source that supplies exactly K photons across n modes poses a concrete communication problem: how many messages can it transmit at a chosen error probability, and which input states make the best use of those photons? This paper gives directly evaluable coding bounds for pure loss and identifies the exact optimal input ensembles for identical noisy Gaussian channels. The results allow researchers to assess finite optical blocks, quantify the information cost of fixing the photon supply, and determine when entanglement between separately constrained blocks can be dispensed with. Every message is supported on the same exact K-photon sector. Within that sector, encoding may use arbitrary mixed states and entanglement among modes, and the receiver may perform a joint measurement. Communication is unassisted. All logarithms below are natural; information is measured in nats. Finite-message bounds and the optimal third-order rate For n ≥ 2 identical pure-loss channels of transmissivity 0 < η < 1 and K ≥ 1, the number L of surviving photons has the same binomial distribution for every encoded state. Define pℓ=(Kℓ)ηℓ(1−η)K−ℓ,dn,ℓ=(n+ℓ−1ℓ).p_\ell=\binom K\ell\eta^\ell(1-\eta)^{K-\ell},\qquad d_{n,\ell}=\binom{n+\ell-1}{\ell}.pℓ​=(ℓK​)ηℓ(1−η)K−ℓ,dn,ℓ​=(ℓn+ℓ−1​). If ps*(M;n,K) is the best average success probability for M equiprobable messages, then ∑ℓ=0Kpℓ[1−M−1dn,ℓ]+≤ps∗(M;n,K)≤∑ℓ=0Kpℓmin⁡ ⁣{1,dn,ℓM}.\sum_{\ell=0}^{K}p_\ell\left[1-\frac{M-1}{d_{n,\ell}}\right]_+\le p_{\mathrm s}^{*}(M;n,K)\le\sum_{\ell=0}^{K}p_\ell\min\!\left\{1,\frac{d_{n,\ell}}{M}\right\}.∑ℓ=0K​pℓ​[1−dn,ℓ​M−1​]+​≤ps∗​(M;n,K)≤∑ℓ=0K​pℓ​min{1,Mdn,ℓ​​}. Here [x]+ = max{x,0}. The lower bound provides an achievable guarantee; the upper bound limits every admissible code. Evaluating both sums brackets the performance at a specified photon supply and block length. Their proof has a simple physical interpretation: surviving photons determine the available output dimension, while one collective-mode codebook works across all possible survivor counts. For fixed E > 0, fixed error tolerance 0 < ε < 1, Kn = ⌊nE⌋ and en = Kn/n, the bounds determine the largest message count through third order: log⁡M∗(n,Kn,ε)=ng(ηen)+nη(1−η)en h(ηen)ΦN−1(ε)−12log⁡n+O(1).\log M^*(n,K_n,\varepsilon)=ng(\eta e_n)+\sqrt{n\eta(1-\eta)e_n}\,h(\eta e_n)\Phi_{\mathrm N}^{-1}(\varepsilon)-\tfrac12\log n+O(1).logM∗(n,Kn​,ε)=ng(ηen​)+nη(1−η)en​​h(ηen​)ΦN−1​(ε)−21​logn+O(1). Here g(x) = (x+1)log(x+1) − x log x, h(x) = log(1+1/x), and ΦN is the standard normal distribution function. This fixes both the fluctuation scale and the optimal −½ log n correction. The finite bounds also yield the exact decay exponent of success above capacity and a matching error exponent on a specified interval below capacity. Which inputs are optimal under Gaussian noise? For n identical nonidentity gauge-covariant Gaussian channels, with thermal photon means transformed as r ↦ ar+b, a > 0 and b ≥ max{0,a−1}, the paper determines the exact fixed-photon Holevo optimum and every ensemble attaining it. Each optimal letter places all K photons in one collective mode: ∣K,u⟩=(∑j=1nujaj†)KK!∣0⟩,∑j∣uj∣2=1.|K,u\rangle=\frac{\left(\sum_{j=1}^{n}u_j a_j^\dagger\right)^K}{\sqrt{K!}}|0\rangle,\qquad \sum_j|u_j|^2=1.∣K,u⟩=K!​(∑j=1n​uj​aj†​)K​∣0⟩,∑j​∣uj​∣2=1. An ensemble is optimal precisely when almost all its letters have this form and its average is uniform on the K-photon sector. A finite weighted optimal ensemble exists. These collective modes may themselves extend across, and be entangled among, physical optical modes. Thus the characterization identifies the structure to optimize without imposing a Gaussian-input restriction. The central proof step combines different loss histories that end in the same output photon sector. A common ordering of the comparison eigenstates allows the separate majorization bounds to be combined. For pure loss, a quantitative rigidity estimate further converts a small Holevo deficit into closeness to collective-mode number states. Resource cost and communication across blocks At fixed positive photon density and fixed nonidentity Gaussian noise, requiring an exact photon supply incurs a Holevo penalty of ½ log n plus an explicitly determined constant, relative to the usual mean-energy value. The paper separates this cost into photon-count uncertainty and the mismatch from a thermal output distribution. For separately constrained blocks of quantum-limited loss, quantum-limited amplification or entanglement-breaking Gaussian noise, the one-block Holevo values add exactly. Every optimal letter is a product of blockwise collective-mode number states, and the repeated-block unassisted classical capacity equals the one-block Holevo value. The low-rate reliability interval outside the matching range and repeated-block additivity for intermediate Gaussian noise remain open.