{"id":"1bb75a4a-f195-4ff1-a5d0-ac7d7d89b565","arxiv_id":"2509.02296","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The authors derive the optimal heralded three-photon interference circuit for increasing photon indistinguishability, accounting for collective phases, and validate it on a programmable integrated photonic processor.","lead":"This paper derives and experimentally validates a three-photon protocol that increases photon indistinguishability by optimizing a linear-optical circuit that accounts for collective three-photon phases. It shows that the previously used fixed interferometer fails in most distinguishability scenarios, and that an optimized circuit recovers positive gains.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'optimal' claim is restricted to a two-photon-interference heralded architecture; the paper's argument that all-three-photon interference cannot be heralded is flawed, leaving global optimality unproven.","rationale":"The reader's conditional verdict is appropriate. The paper's internal results—Eq. (2), the construction in Eq. (B7), and the experimental demonstrations—are consistent and support optimality within the Fig. 1b family. However, the paper's central claim is phrased as 'optimal distillation' without restricting to that family. The only argument given for excluding all-three-photon interference circuits is that they cannot be heralded; that argument is logically insufficient, because postselection on a two-mode detection pattern does guarantee separate spatial modes. This is a load-bearing gap because if a three-photon-interference circuit achieves a higher conditioned visibility, the maximization over S and φu is not globally optimal. The concrete test above would settle the question. I do not see an internal inconsistency in the derivation of Eq. (2) or the tensor permanent formula; those are standard. Therefore the verdict should remain CONDITIONAL: the paper's claims are well supported within its stated architecture, but the global 'optimal' assertion needs a proof or an explicit scope limitation.","tokens_in":16839,"tokens_out":23835,"duration_ms":254134,"concrete_test":"For a fixed valid Gram matrix (e.g., V12=V13=V23=0.5, φ=0, and V12=V13=V23=0.25, φ=π), numerically optimize over all 3×3 (and 4×4) unitary matrices U acting on all three input photons, with the success event being exactly one photon in each of two output modes (third mode as herald). Compute the conditioned two-photon HOM visibility using the tensor permanent formula (Eq. A5), and compare to the maximum of Eq. (2) over S∈[0,∞), φu∈[0,2π) and over input permutations. If any U yields a larger visibility, the global optimality claim fails; if dense sampling (e.g., 10^5 Haar-random unitaries plus local optimization) finds no improvement, the restriction is empirically supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In 'Optimal distillation protocol', the optimization is over S and φu from Eq. (2), which describes only the circuit of Fig. 1b, where two photons enter U_D and the third is compared afterwards. The paper dismisses schemes with all three photons interfering by saying they 'cannot be considered heralded' because 'it is unclear how to guarantee that the two distilled photons would occupy different spatial modes, since they would always have a probability of bunching together.' This is not a valid argument: a heralded protocol is defined by postselecting on a detection pattern. If the success pattern is one photon in each of two designated output modes, the two distilled photons are guaranteed to be in different modes; bunching outcomes are simply discarded. Thus the paper has not shown that the maximum of Eq. (2) is the global optimum among all linear-optical heralded distillation protocols for three photons. The title's 'optimal' is therefore only proven for the specific architecture of Fig. 1b.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a three-photon protocol for distilling photonic indistinguishability. In the considered architecture (Fig. 1b), two photons pass through a three-mode interferometer U_D and are postselected to emerge one in each of two output modes; the photon in the first output mode is then compared by Hong-Ou-Mandel interference with a third, reference photon that did not enter U_D. The authors derive closed expressions for the postselected output visibility V_f and success probability P in terms of two parameters S and phi_u of U_D (Eqs. (2)-(3)), construct an explicit unitary that maximizes V_f and then P (Appendix B), and experimentally validate the protocol on an eight-mode integrated photonic processor fed by a demultiplexed quantum-dot source. They test several Gram matrices, including cases with non-zero three-photon Bargmann phase phi, and show that a previously used fixed interferometer U_0 often produces negative gain, whereas the optimized circuit produces positive gain. The paper also compares the optimized unitary against randomly sampled unitaries.","tokens_in":17093,"tokens_out":15368,"duration_ms":184245,"significance":"If the architecture is accepted as the intended class, this is a useful and well-executed contribution. The analytic parameterization of the distilled visibility and success probability is clean, the dependence on the collective triad phase is convincingly demonstrated experimentally, and the random-unitary comparison in Fig. 5 is a nice falsifiable test. The experimental platform is appropriate and the data are consistent with simulations that use measured input visibilities rather than fitting the output. However, the title and abstract claim an unqualified 'optimal distillation'. That global claim is not proven: the optimization over S and phi_u applies only to a two-photon-interference, one-reference-photon architecture, and the manuscript's argument for excluding all-three-photon interference schemes is not valid. The work is therefore stronger as an optimized and demonstrated protocol within a well-defined circuit family than as a proof of global optimality.","major_comments":[{"comment":"The sentence excluding all-three-photon interference protocols is not justified. The text says such schemes 'cannot be considered heralded' because the two distilled photons would have a probability of bunching together. But a heralded protocol is defined by postselecting on a detection pattern; if the success pattern is one photon in each of two designated output modes, the two distilled photons are guaranteed to occupy different spatial modes, and bunching events are simply discarded. This is exactly how the paper's own pattern [1,1,0] works. Since Eqs. (2)-(3), the numerical maximization over (S, phi_u), and the unitary construction in Appendix B are all derived for the Fig. 1b architecture, the global 'optimality' asserted in the title, abstract, and Discussion is not established. The authors should either prove an upper bound that includes three-photon-interference heralded protocol","section":"Optimal distillation protocol (after Eq. (3))"},{"comment":"The statement that the protocol is 'always able to provide a gain G >= 0' is trivially satisfied in the considered architecture: taking S = 0 in Eq. (2) gives V_f = V_13, so after relabeling inputs so that V_13 is the maximum, G = 0. This does not substantiate a claim of positive distillation gain. If the intended claim is that for every physical Gram matrix there is a circuit with strictly positive gain, that needs a separate condition and proof; otherwise the sentence should be reworded to avoid implying a substantive performance guarantee.","section":"Discussion (paragraph 2) and Appendix C"}],"minor_comments":[{"comment":"The text says 'We prove that ... the protocol is suboptimal', but the supporting evidence is numerical (10000 random Gram matrices plotted in Fig. 6). Please replace 'prove' with 'show numerical evidence' unless an analytical theorem is added.","section":"Appendix C"},{"comment":"The explicit unitary in Eq. (B7) has expressions with sqrt{1+cos(phi_u)} in denominators and is singular at phi_u = pi, which is one of the key optimal cases in the main text (the phi = pi example). Although the authors mention a row-switch variant for that case, a rigorous presentation should give the limiting form or a separate explicit matrix for phi_u = pi.","section":"Appendix B, Eq. (B7)"},{"comment":"The availability statement says data are available from the corresponding author on reasonable request. Given the quantitative character of the claims (G_max, comparisons with random unitaries, simulated curves), depositing raw visibility counts and processing scripts would strengthen reproducibility.","section":"Data availability"},{"comment":"The parameterization with G_12 = sqrt(V_12), G_23 = sqrt(V_23) e^{i phi}, G_13 = sqrt(V_13) is fine, but the text should state explicitly that the phase ambiguity of the individual internal states has been fixed, since the same Gram matrix can be represented by different-looking matrices related by diagonal unitary transformations.","section":"Notation around Eq. (1)"}],"recommendation":"major_revision","confidential_remarks":"The central concern is whether the 'optimal' claim can be defended. The restriction to the two-photon-interference architecture is an assumption, not a proven limitation, and the argument used to dismiss all-three-photon interference is clearly flawed. This is fixable by either proving a rigorous upper bound covering more general linear-optical heralded protocols or by systematically qualifying the title, abstract, and discussion. The experimental and analytical contributions inside the considered architecture are solid and worth publishing after the claims are made precise. I would not recommend rejection, but the revision needs to address the scope of optimality explicitly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing you should know: this is a solid step beyond the fixed-U0 protocols. The paper correctly identifies that the distilled visibility depends on S, phi_u, and the three-photon Bargmann phase, and it gives an explicit 3-mode unitary for any optimal pair. The experiment is real: several Gram matrices, real and complex triad phases, and a random-unitary comparison. Data agree with simulations, and the old U0 fails on most random Gram matrices, so the optimization matters.\n\nSoft spots: the global optimality claim outruns the proof. The text dismisses schemes where all three photons interfere by saying they cannot be heralded because of bunching. That is wrong. Postselecting on one photon in each of two output modes is a valid heralding condition; bunching outcomes are simply discarded. So the paper proves optimality within the class of circuits shown in Fig. 1b, not among all three-photon linear-optical distillation protocols. This should be fixed, either with a proof that all-three-interference schemes reduce to Fig. 1b or by narrowing the title and claims.\n\nAlso, the statement that a gain G >= 0 can always be found is asserted, not proven, and looks false for boundary cases. If one pair already has visibility 1 (say V23 = 1), no protocol can exceed max = 1, and Eq. (2) for the natural ordering gives Vf = V12 < 1. Maybe they intend to exclude already-perfect pairs or to allow some other success criterion, but as written it is an overstatement. The Appendix B construction is a unitary dilation for given S and phi_u, plus a success-probability maximization; it does not prove the global non-negative gain claim.\n\nWhat is genuinely good: the use of the Bargmann invariant to capture the three-photon phase is correct, the unitary dilation construction is neat, and the experiments are properly done on a programmable chip. The random-unitary comparison is a nice stress test, and the paper honestly shows that U0 fails without triad-phase knowledge. Minor: data are only available on request; code would help.\n\nWho is this for: people working on photon indistinguishability, linear-optical quantum computing, and resource-efficient error mitigation. It deserves a serious referee. I would send it to review, with the expectation that the authors clarify the scope of 'optimal' and either prove or retract the G >= 0 claim. My verdict: conditionally acceptable after those revisions. I would cite it for the optimized construction, not for the unconditional optimality claim.","headline":"The optimized three-photon distillation protocol that accounts for the Bargmann triad phase is real and experimentally supported, but the 'optimal' claim is only proven for the specific heralded architecture of Fig. 1b, not for all heralded three-photon linear-optical schemes.","tokens_in":17583,"tokens_out":4218,"would_cite":true,"duration_ms":54290,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["03.67.-a","42.50.Ex"],"model":"deepseek-v4-flash","headline":"For any three-photon partially distinguishable input, the paper constructs the optimal three-mode heralded circuit that maximizes the distilled Hong-Ou-Mandel visibility, and demonstrates it on a quantum dot source.","keywords":["photonic indistinguishability","distillation","Hong-Ou-Mandel visibility","Gram matrix","Bargmann invariant","triad phase","linear optical circuits","quantum dot source"],"falsifier":"Fix an input Gram matrix (e.g., V12=V13=V23=0.5, φ=π/4), compute the distilled visibility of the claimed optimal unitary Eq. (B7) by the tensor-permanent formula, and compare it with a brute-force numerical maximization over all 3×3 unitaries (and all input permutations). If any unitary yields a visibility exceeding the maximum of Eq. (2) over S and φu, the optimality claim is falsified. Experimentally, the same comparison can be run by programming a set of random unitaries in a universal interferometer and checking that none beats the optimal one.","tokens_in":16760,"feed_emoji":"⚛️","tokens_out":9337,"duration_ms":95931,"temperature":0.7,"pith_summary":"The paper establishes that photon indistinguishability can be distilled to its theoretically maximal value by a three-mode linear-optical circuit whose design is computed from the full three-photon Gram matrix, including the collective triad phase. Earlier distillation circuits used a fixed unitary and ignored multiphoton phases, which makes them yield negative gain for the vast majority of input scenarios; the new recipe always achieves non-negative gain after an input permutation. The central formula expresses the final Hong-Ou-Mandel visibility as a ratio of two simple functions of the interferometer parameters, so the optimum is found by a two-parameter maximization, and the paper supplies the explicit unitary attaining it while also maximizing the success probability. Experiments with a demultiplexed quantum dot source and a programmable eight-mode photonic processor confirm the predicted distilled visibilities for real and complex Gram matrices, including cases where the old circuit fails. A correct protocol of this kind matters because it offers a resource-efficient, integrable error-mitigation step for photonic quantum computing and communication.","feed_headline":"Three-mode circuit achieves maximum photon distillation","feed_subtitle":"Old fixed circuits fail on ~90% of inputs; the new recipe includes the collective phase and always gains.","key_machinery":"The carrying mechanism is a closed-form expression for the distilled Hong-Ou-Mandel visibility, Vf(S,φu), derived from the tensor-permanent formalism for partially distinguishable three-photon interference. It depends on the interferometer only through two parameters—S (the relative amplitude weight of the two paths) and φu (a phase of the unitary)—and on the input through the three pairwise visibilities V12,V13,V23 and the collective triad phase φ. Maximizing this ratio over S and φu, followed by a unitary-dilation construction of the optimal three-mode interferometer (Eq. B7) that also maximizes success probability, yields the claimed optimal protocol. The triad phase φ enters as the term","core_discovery":"For any three-photon partially distinguishable input state, characterized by the three pairwise Hong-Ou-Mandel visibilities V12, V13, V23 and the Bargmann-invariant triad phase φ, the paper claims that the maximum distilled visibility attainable in the heralded architecture—in which two photons interfere in a three-mode interferometer and the emerging photon is compared with the third by a HOM measurement—is obtained by maximizing the explicit ratio Vf(S,φu) = [V12S² + V13 + 2S√(V12V13V23) cos(φ+φu)] / [1 + S² + 2S V23 cos(φu)] over the two real parameters S and φu that encode the interferometer. The paper constructs the unitary that attains this maximum (Eq. B7), using the unitary dilation","pith_inferences":["For N>3 photons the same tensor-permanent formalism should yield optimal heralded distillation circuits as well, but the closed-form visibility formula and the explicit unitary would need to be re-derived; a natural extension is to treat the set of all Bargmann invariants as the input.","A recursive application—distilling the output pair against a fresh photon—should push visibility toward unity while dropping success probability; working out the tradeoff curve would tell whether such recursion is practically useful.","Since the optimal circuit depends sensitively on the triad phase, deployed systems will need to monitor φ in situ (e.g., via a Fourier-interferometer measurement) and re-program the interferometer accordingly.","The optimality of Eq. (B7) is tested only against 50 random unitaries; a stronger check would be a numerical global search over all 3×3 unitaries for each input Gram matrix."],"forward_implications":["Given the three pairwise visibilities and the triad phase, the recipe in Appendix B returns an explicit three-mode unitary that achieves the maximum possible distilled visibility for that input.","The protocol is guaranteed to give G ≥ 0 (after choosing the best input permutation), whereas the previously known fixed circuit U0 gives negative gain for about 91.6% of real and 87.4% of complex random Gram matrices.","For the balanced-visibility case with φ=π (visibilities ≤ 0.25), U0 gives no gain while the optimized circuit (S=1, φu=π) gives positive gain; for φ=±π/4 at V=0.5, the optimal circuit is the three-mode Fourier tritter with φu=∓2π/3.","The success probability can be maximized analytically among gain-optimal unitaries, so the protocol does not sacrifice visibility to gain heralding rate.","Because the protocol uses only three spatial modes and three photons, it can be embedded as a resource-efficient error-mitigation block in larger photonic circuits."],"supporting_citations":[{"why":"Supplies the Gram matrix and Bargmann invariant formalism used to characterize the three-photon distinguishability scenarios.","marker":"[46]"},{"why":"Provides the tensor permanent formula (Eq. A5) from which the distilled visibility expression is derived.","marker":"[49]"},{"why":"Supplies the theory of partially distinguishable photon interference used in the derivation of the outcome probabilities.","marker":"[48]"},{"why":"Proposes the earlier fixed distillation circuit U0 that the paper generalizes and shows to be suboptimal.","marker":"[38]"},{"why":"Reports the recent experimental demonstration of a distillation protocol based on U0, the baseline this work improves upon.","marker":"[42]"},{"why":"Gives the unitary dilation theorem used to embed the optimal 2x2 submatrix into the 3-mode unitary of Eq. (B7).","marker":"[50]"},{"why":"Experimental evidence for collective triad-phase effects in three-photon interference, motivating the inclusion of φ in the optimization.","marker":"[21]"},{"why":"Establishes realizability conditions for Gram matrices with given visibilities and triad phase, used to design the φ=π and complex-phase input states.","marker":"[51]"}],"fun_headline_variants":["Photon distillation hits theoretical max in three modes","Optimized three-photon circuit tops indistinguishability gain","Three-mode interferometer yields maximal photon visibility","Maximal distillation of photon indistinguishability via minimal circuit"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The optimality claim covers only the heralded architecture in which two photons enter the interferometer and the third is compared afterwards; the paper assumes without proof that protocols where all three photons interfere cannot be heralded, so a global optimum outside that class is not excluded.","fun_headline_variants_meta":{"raw":{"variants":["Photon distillation hits theoretical max in three modes","Optimized three-photon circuit tops indistinguishability gain","Three-mode interferometer yields maximal photon visibility","Maximal distillation of photon indistinguishability via minimal circuit"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000533,"raw_usage":{"total_tokens":2378,"prompt_tokens":695,"completion_tokens":1683,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":439,"completion_tokens_details":{"reasoning_tokens":1633}},"tokens_in":439,"tokens_out":1683,"duration_ms":16077,"temperature":1.0,"reasoning_tokens":1633,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T11:43:23.148631+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Fix an input Gram matrix (e.g., V12=V13=V23=0.5, φ=π/4), compute the distilled visibility of the claimed optimal unitary Eq. (B7) by the tensor-permanent formula, and compare it with a brute-force numerical maximization over all 3×3 unitaries (and all input permutations). If any unitary yields a visibility exceeding the maximum of Eq. (2) over S and φu, the optimality claim is falsified. Experimentally, the same comparison can be run by programming a set of random unitaries in a universal interferometer and checking that none beats the optimal one.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Gram matrix and Bargmann invariant formalism used to characterize the three-photon distinguishability scenarios."},{"cited_title":"Oszmaniec, D","cited_arxiv_id":null,"evidence_quote":"Supplies the theory of partially distinguishable photon interference used in the derivation of the outcome probabilities."},{"cited_title":"Somhorst, B","cited_arxiv_id":null,"evidence_quote":"Reports the recent experimental demonstration of a distillation protocol based on U0, the baseline this work improves upon."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the unitary dilation theorem used to embed the optimal 2x2 submatrix into the 3-mode unitary of Eq. (B7)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Experimental evidence for collective triad-phase effects in three-photon interference, motivating the inclusion of φ in the optimization."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes realizability conditions for Gram matrices with given visibilities and triad phase, used to design the φ=π and complex-phase input states."}],"review_version":1}