{"id":"bd8c5631-2945-4732-966e-647b4e8fa7ba","arxiv_id":"2411.10067","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Entanglement shared between transmitters enlarges the achievable rate region of interference channels, and a magic-square-game channel shows a strict sum-rate advantage over classical strategies.","lead":"This paper derives the first general inner and outer bounds on the capacity of a two-sender, two-receiver interference channel in which the senders share quantum entanglement. It then exhibits a magic-square-game channel where entanglement strictly beats every classical strategy, raising the achievable sum rate from about 3.02 to 3.17 bits per channel use.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Outer-bound proof of Theorem 6 replaces the conditioning in (64) by Y2^{i-1} without justification; the replacement can shrink the mutual information, so the claimed upper bound is not established.","rationale":"The reader's weakest assumption is the imported classical bound from [7] used in the magic-square example. That is a reasonable concern, but the bound is a published external result and the transfer to the IC via a merged-output MAC is legitimate, so it is not the most load-bearing issue. The more serious problem is internal: the proof of Theorem 6, one of the paper's two main theorems, contains an unjustified change of conditioning in the individual-rate converse. The step from (64) to (70) is not an identity and can fail in a simple DMC with deterministic encoding, as sketched in the concrete test. If this step cannot be repaired, the claimed outer bound RET-o is unproven, and the paper's central claim that the capacity region is bracketed by Theorems 3 and 6 loses one of its two halves. The inner bound and the specific magic-square separation example may still be correct, so the appropriate disposition is to accept only conditionally on the outer-bound proof being fixed or the theorem statement being weakened. I therefore disagree with the reader's identification of the weakest assumption and recommend a CONDITIONAL verdict rather than an unconditional ACCEPT.","tokens_in":16872,"tokens_out":26013,"duration_ms":249549,"concrete_test":"Independently re-derive the step from (64) to (70) using the chain-rule identity I(X;Y|Z) = I(X;Y|Z,W) + I(X;W|Z) - I(X;W|Y,Z). Then test the claimed replacement on a concrete DMC: X1 = {0,1}, X2 = {0,1}, Y1 = X1 xor Bernoulli(eps), Y2 = X1, with a deterministic capacity-approaching code for some R1 < C(BSC) and R2 = 0. Compute the sum in (70) for the code's induced V0[i], V1[i], V2[i]; if it is less than I(M1; Y1^n | M2) around nC, the replacement is invalid and the converse proof of Theorem 6 fails as written. An analytic version: verify whether I(M1; Y2^{i-1} | Y1^{i-1}, M2) = I(M1; Y2^{i-1} | Y1^{i-1}, Y1[i], M2) for this example; it should not hold.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Section VII derives the individual-rate outer bounds by rewriting (64), n(R1 - eps) <= sum_i I(M1; Y1[i] | Y1^{i-1}, M2), as (70), whose summands are I(V1[i]; Y1[i] | V0[i], V2[i]) = I(M1; Y1[i] | Y1^{i-1}, Y2^{i-1}, M2). These two quantities are not equal in general. Conditioning on the extra Y2^{i-1} can either increase or decrease I(M1; Y1[i] | Y1^{i-1}, M2); the identity I(X;Y|Z) = I(X;Y|Z,W) + I(X;W|Z) - I(X;W|Y,Z) shows equality would require I(M1; Y2^{i-1} | Y1^{i-1}, M2) = I(M1; Y2^{i-1} | Y1^{i-1}, Y1[i], M2), which is not argued and is false for natural channels. For a deterministic encoder with channel Y1 = X1 xor Z, Y2 = X1, conditioning on Y2^{i-1} = X1^{i-1} reveals more about M1 than the noisy Y1^{i-1} alone, so the sum in (70) can be strictly smaller than the quantity it is meant to upper-bound. The same problem affects the R2 bound. Thus the proof of Theorem 6 does not establish C_ET subset RET-o; the central bracketing result is missing one of its two halves. The inner bound and the magic-square example are not affected by this specific gap, but the paper's headline claim of a bracketing pair is unsupported as written.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the two-sender, two-receiver classical interference channel with entanglement shared between the transmitters. It proposes an inner bound RET-HK (Theorem 3) obtained by a Han-Kobayashi-type superposition coding scheme with quantum measurements, and an outer bound RET-o (Theorem 6) obtained by single-letterizing the Fano-type inequalities. It also constructs a concrete example based on the magic-square game (Theorem 8) in which entanglement is claimed to strictly enlarge the capacity region: the classical sum-rate is at most 3.02 bits per use, while the entanglement-assisted sum-rate reaches 2 log2(3) ≈ 3.17 bits per use. The inner-bound proof follows the standard typicality analysis with the measurement layer folded into the input distribution, and the example uses a published classical upper bound for the corresponding MAC. The central claim is the bracketing relationship RET-HK ⊆ CET ⊆ RET-o.","tokens_in":17202,"tokens_out":16670,"duration_ms":166849,"significance":"If the bounds are correct, the paper extends entanglement-assisted communication from the multiple-access channel to the interference channel, a central network model, and provides a clean example of a quantum advantage. The inner-bound construction is a natural adaptation of Han-Kobayashi coding to entangled transmitters, and the magic-square example is well matched to known non-local-game techniques. The paper is also transparent about its limitations, explicitly noting the absence of cardinality bounds for the auxiliary variables U1 and U2. However, the outer bound proof contains a load-bearing gap in its single-letterization step, so the bracketing result is not established as written. The inner bound and the example are not affected by this specific gap, but the headline claim of a capacity-region bracketing relies on the validity of the outer bound.","major_comments":[{"comment":"The proof of Theorem 6 does not establish the claimed single-letterization for the individual-rate bounds. Starting from n(R1 - ε) ≤ Σ_i I(M1; Y1[i] | Y1^{i-1}, M2), the paper defines V0[i] = (Y1^{i-1}, Y2^{i-1}) and Vk[i] = (Mk, Y1^{i-1}, Y2^{i-1}) and then rewrites the sum as (1/n) Σ_i I(V1[i]; Y1[i] | V0[i], V2[i]). With these definitions, the summand equals I(M1; Y1[i] | Y1^{i-1}, Y2^{i-1}, M2), which is not equal to I(M1; Y1[i] | Y1^{i-1}, M2) in general. Conditioning on the additional Y2^{i-1} can strictly decrease the mutual information; for example, in a deterministic code for a channel with Y1 = X1 ⊕ Z and Y2 = X1 with X1 = f(M1) injective, Y2^{i-1} reveals M1 and makes the new sum strictly smaller than the quantity it is supposed to upper-bound. Consequently the inequality nR1 ≤ (1/n) Σ_i I(V1[i]; Y1[i] | V0[i], V2[i]) is not justified, and the proof of the individual outer bounds (27) and (28) fails. The sum-rate bound (29) is not affected, since it follows directly from the chain rule with (Y1^{i-1}, Y2^{i-1}) as the natural past. Because Theorem 6 is the outer-bound half of the central bracketing result, this gap is load-bearing and requires a corrected proof.","section":"Section VII, Eqs. (64) and (70)"},{"comment":"The single-letterization also relies on the assertion that the channel inputs Xk[i] can be represented by a product of measurements of the form Lk(xk|i, V0[i], Vk[i]) for a shared state φ. For a general block code whose encoding POVMs act jointly on n copies of the shared state, the marginal measurement at position i is a valid measurement on the full n-copy system, but it is not automatically a product of single-copy measurements on a fixed single-copy state. The paper refers to [21] for this technique; if the analogy is exact, the authors should state and prove the relevant lemma in their own notation, since the outer bound's region P′ requires the specific factorization in (26). This is part of the same proof gap and should be addressed together with the step from (64) to (70).","section":"Section VII, after Eq. (69)"}],"minor_comments":[{"comment":"There is a typo: \"advanatges\" should be \"advantages\".","section":"Section IV-A, paragraph 2"},{"comment":"The phrase \"As as result\" should be \"As a result\"; also \"winning the the game\" should be \"winning the game\".","section":"Section V-B, paragraph 4"},{"comment":"In the tensor product on the left-hand side, the second factor should be L̃(i,m2)_{E2→X2}, not L̃(i,m1)_{E1→X1}.","section":"Section VII, Eq. (68)"},{"comment":"The text says \"the bounds in (17) – (17)\" but should refer to (17)–(23).","section":"Section VIII, final paragraph"},{"comment":"The absence of cardinality bounds for U1 and U2 is acknowledged, but the paper could beneficially add a short discussion of whether finite bounds follow from the same arguments as in [21] or whether the union over unbounded alphabets is essential.","section":"Remark 2, Section IV-A"}],"recommendation":"major_revision","confidential_remarks":"The outer-bound proof gap is the decisive issue. The inner bound and the example may be correct, but the central bracketing claim requires a valid converse. I would recommend that the authors consult the MAC outer-bound technique in [21] and either repair the individual-rate bounds or modify the statement of Theorem 6, and resubmit. The reliance on [21], which has an overlapping author, is acceptable if the adaptation is made explicit and self-contained."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. This is the first general inner bound for the interference channel with entangled transmitters, and the magic-square example is a clean demonstration of quantum advantage. But the outer bound proof in Section VII has a load-bearing gap, exactly where the stress-test note says. The claimed bracketing theorem (Theorem 6) is not established as written.\n\nThe inner bound (Theorem 3) is a straightforward adaptation of Han-Kobayashi to entanglement-assisted POVM encoding, and the typicality error analysis is standard. The purification lemma and the |V0| ≤ 7 cardinality argument are fine. The magic-square IC example is neat: the quantum strategy achieves sum rate 2 log2(3) ≈ 3.17, and the classical upper bound 3.02 is imported from Seshadri et al. via a MAC argument. That import is legitimate, because coalescing the two receiver outputs into one MAC output can only help the decoder, so the separation example does not depend on the outer bound.\n\nThe soft spot is the outer bound proof. Equation (64) has summands I(M1;Y1[i]|Y1^{i-1},M2). The rewrite to (70) defines V0[i]=(Y1^{i-1},Y2^{i-1}) and V2[i]=(M2,Y1^{i-1},Y2^{i-1}), turning each summand into I(M1;Y1[i]|Y1^{i-1},Y2^{i-1},M2). These are not equal, and conditioning on the extra Y2^{i-1} can reduce the mutual information. The proof asserts equality without any bound in the right direction. The sum-rate bound in (72) survives because the chain rule there already conditions on the full past, but the individual R1 and R2 bounds are unproven. Minor issues: no finite alphabet bounds for U1,U2, and the imported classical bound is not re-derived.\n\nThis paper is for researchers in entanglement-assisted network information theory; the inner bound and the example will be useful to them. It deserves a serious referee, not a desk rejection, but it needs major revision before acceptance. I would send it out with a request to fix the outer bound proof or explicitly reduce the claim to the inner bound plus the example. As it stands, I would not cite the outer bound.","headline":"Useful inner bound and a nice example, but the outer bound proof has an unjustified conditioning swap, so the bracketing theorem is not proven.","tokens_in":17734,"tokens_out":9730,"would_cite":true,"duration_ms":89073,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["94A15","94A40","81P45"],"pacs":["03.67.-a","03.67.Hk"],"model":"deepseek-v4-flash","headline":"The paper establishes inner and outer bounds on the capacity region of the interference channel with entangled transmitters and exhibits a magic-square channel where entanglement strictly beats every classical strategy.","keywords":["interference channel","entanglement-assisted communication","capacity region","Han-Kobayashi inner bound","outer bound","magic square game","quantum advantage","multiple access channel"],"falsifier":"Compute the true classical sum capacity of the magic-square interference channel without merging the outputs; if any classical strategy reaches $R_1+R_2 \\ge 2\\log_2(3)\\approx 3.17$, the claimed strict quantum advantage is false. A second check is to search for an entanglement-assisted code exceeding the outer bound $R_{\\mathrm{ET-o}}$, which would refute the outer bound itself.","tokens_in":16667,"feed_emoji":"⚛️","tokens_out":10906,"duration_ms":101498,"temperature":0.7,"pith_summary":"This paper asks whether shared quantum entanglement between the two senders can raise communication rates over a classical two-sender, two-receiver interference channel. It supplies general capacity-region bounds for this setting: an achievable inner region RET-HK built from rate-splitting and superposition coding, and an outer region RET-o obtained from Fano-type arguments. The bounds are single-letter expressions that mirror the classical Han-Kobayashi and standard outer bounds, with the entangled measurement encoders recorded through a union over shared states and POVMs. The paper then exhibits a concrete channel, derived from the magic square game, where entanglement-assisted coding achieves sum rate $2\\log_2(3)\\approx 3.17$ bits per use while every classical strategy is limited to $3.02$, so the answer is yes: entanglement between transmitters can strictly enlarge the capacity region.","feed_headline":"Entangled transmitters beat classical limit on a magic-square channel","feed_subtitle":"Classical sum rate caps at 3.02 bits per use, while entangled transmitters reach 3.17.","key_machinery":"The machinery is a Han-Kobayashi-style rate-splitting/superposition code in which the encoders are POVMs $L_1,L_2$ acting on a shared bipartite entangled state $\\varphi_{E_1E_2}$, so the channel inputs $X_1,X_2$ inherit correlations from the entanglement and the auxiliary variables $V_k$ cannot be collapsed into $X_k$. The outer bound uses the same POVM structure with a time-sharing variable $V_0$ and Fano-based single-letterization. For the concrete advantage, the magic square game supplies a perfect quantum strategy with winning probability 1 and a classical optimum of 8/9, and the channel outputs the question pair only on winning inputs; this makes the entanglement-assisted sum rate $2\\log_2(3)$ achievable while the merged-output MAC argument caps classical strategies at 3.02.","core_discovery":"The central claim is that for every discrete memoryless interference channel with entangled transmitters the capacity region $C_{\\mathrm{ET}}$ satisfies $R_{\\mathrm{ET-HK}} \\subseteq C_{\\mathrm{ET}} \\subseteq R_{\\mathrm{ET-o}}$, where $R_{\\mathrm{ET-HK}}$ is the union over rate-splitting/superposition distributions, shared entangled states, and POVMs of the seven inequalities (17)-(23), and $R_{\\mathrm{ET-o}}$ is the union of the three inequalities (27)-(29). The inner bound is achieved by Han-Kobayashi-style random coding in which each message is split into a common and a private part and the channel input is produced by measuring a shared entangled state; the outer bound follows from Fano's inequality after single-letterizing with a time-sharing variable. The paper's example is an interference channel built from the magic square game, for which the entanglement-assisted region contains the point $R_1+R_2 = 2\\log_2(3)$, while every classical strategy has $R_1+R_2 \\le 3.02$. This establishes a strict gap between classical and entanglement-assisted capacity for a concrete two-sender, two-receiver channel.","pith_inferences":["The 3.17-versus-3.02 gap is computed through the MAC ceiling obtained by merging the two receiver outputs; a direct classical IC bound would likely give a smaller ceiling, so the true quantum advantage on this channel may be larger than the stated gap.","The inner and outer bounds do not coincide, so the paper leaves open whether entanglement helps on channels where the classical Han-Kobayashi region is already tight; finding strong-interference-type conditions for entangled encoders would close this gap.","Because any non-local game with a quantum-versus-classical winning gap yields a candidate IC by the same construction, the magic-square example is evidence that entanglement-assisted transmitter coordination is a general resource in interference networks, not an isolated example.","A concrete finite bound on the auxiliary alphabets and the entangled-state dimension is still missing; without it, the regions are defined by infinite unions and are not directly computable."],"forward_implications":["The inner bound $R_{\\mathrm{ET-HK}}$ is achievable for every discrete memoryless interference channel, so any rate pair inside it is a lower bound on the entanglement-assisted capacity region.","The outer bound $R_{\\mathrm{ET-o}}$ is a single-letter ceiling for every entanglement-assisted code, giving a concrete target that any claimed quantum advantage must beat.","On the magic-square-game channel, entanglement-assisted coding reaches $R_1+R_2 = 2\\log_2(3)$, and the paper's merged-output argument shows this advantage persists for any non-local-game channel whose MAC version has a quantum-versus-classical gap.","For channels built from non-local games, receiver cooperation in the IC is equivalent to a MAC, so the same classical sum-rate bound applies and the quantum strategy consistently wins."],"supporting_citations":[{"why":"Supplies the classical upper bound $R_1+R_2 \\le 3.02$ for the magic-square MAC, which the IC example must beat.","marker":"[7]"},{"why":"Defines MACs from non-local games and shows entanglement among transmitters yields a capacity advantage, the template for the IC example.","marker":"[5]"},{"why":"Provides the entangled-transmitter MAC capacity bounds and coding techniques that the IC inner and outer bounds adapt.","marker":"[8]"},{"why":"Standard reference for the classical IC inner/outer bounds, MAC capacity region, and Fourier-Motzkin elimination used in the proofs.","marker":"[16]"},{"why":"Original Han-Kobayashi inner bound that the entanglement-assisted inner bound mirrors.","marker":"[17]"},{"why":"Simplified form of the Han-Kobayashi region whose inequalities are the classical counterparts of the new inner-bound inequalities.","marker":"[18]"},{"why":"Observation that the IC capacity region is contained in the MAC capacity region when outputs are merged, used to transfer the classical sum-rate bound.","marker":"[20]"},{"why":"Establishes the perfect quantum strategy and classical optimal winning probability for the magic square game, which underlie the channel construction.","marker":"[15]"},{"why":"Supplies the proof that pure states suffice in the rate-region union and the single-letterization technique for the outer bound.","marker":"[21]"}],"fun_headline_variants":["Entanglement boosts interference channel capacity","Magic-square channel shows quantum edge in sum rate","Entangled transmitters outpace classical on interference","Interference channel with entanglement: new capacity bounds","Quantum entanglement gives strict advantage in interference channel"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The strict quantum-advantage example rests on an external result that every classical strategy on the magic-square MAC has sum rate at most 3.02 bits per use, together with the transfer of that bound to the interference channel by merging the receiver outputs into one MAC; if either the external bound or the transfer is invalid, the claimed 3.17-versus-3.02 separation collapses.","fun_headline_variants_meta":{"raw":{"variants":["Entanglement boosts interference channel capacity","Magic-square channel shows quantum edge in sum rate","Entangled transmitters outpace classical on interference","Interference channel with entanglement: new capacity bounds","Quantum entanglement gives strict advantage in interference channel"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000428,"raw_usage":{"total_tokens":2150,"prompt_tokens":870,"completion_tokens":1280,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":486,"completion_tokens_details":{"reasoning_tokens":1213}},"tokens_in":486,"tokens_out":1280,"duration_ms":10630,"temperature":1.0,"reasoning_tokens":1213,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T20:03:24.859876+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the true classical sum capacity of the magic-square interference channel without merging the outputs; if any classical strategy reaches $R_1+R_2 \\ge 2\\log_2(3)\\approx 3.17$, the claimed strict quantum advantage is false. A second check is to search for an entanglement-assisted code exceeding the outer bound $R_{\\mathrm{ET-o}}$, which would refute the outer bound itself.","supporting_citations":[{"cited_title":"On the separation of correlation-assisted sum capacities of multiple access channels,","cited_arxiv_id":null,"evidence_quote":"Supplies the classical upper bound $R_1+R_2 \\le 3.02$ for the magic-square MAC, which the IC example must beat."},{"cited_title":"Playing games with multiple access channels,","cited_arxiv_id":null,"evidence_quote":"Defines MACs from non-local games and shows entanglement among transmitters yields a capacity advantage, the template for the IC example."},{"cited_title":"The multiple-access channel with entangled transmitters,","cited_arxiv_id":null,"evidence_quote":"Provides the entangled-transmitter MAC capacity bounds and coding techniques that the IC inner and outer bounds adapt."},{"cited_title":"El Gamal and Y .-H","cited_arxiv_id":null,"evidence_quote":"Standard reference for the classical IC inner/outer bounds, MAC capacity region, and Fourier-Motzkin elimination used in the proofs."},{"cited_title":"A new achievable rate region for the interference channel,","cited_arxiv_id":null,"evidence_quote":"Original Han-Kobayashi inner bound that the entanglement-assisted inner bound mirrors."},{"cited_title":"On the Han–Kobayashi region for the interference channel,","cited_arxiv_id":null,"evidence_quote":"Simplified form of the Han-Kobayashi region whose inequalities are the classical counterparts of the new inner-bound inequalities."},{"cited_title":"Two-user communication channels,","cited_arxiv_id":null,"evidence_quote":"Observation that the IC capacity region is contained in the MAC capacity region when outputs are merged, used to transfer the classical sum-rate bound."},{"cited_title":"Quantum pseudo-telepathy,","cited_arxiv_id":null,"evidence_quote":"Establishes the perfect quantum strategy and classical optimal winning probability for the magic square game, which underlie the channel construction."},{"cited_title":"The Multiple-Access Channel with Entangled Transmitters","cited_arxiv_id":"2303.10456","evidence_quote":"Supplies the proof that pure states suffice in the rate-region union and the single-letterization technique for the outer bound."}],"review_version":1}