{"id":"a5b69234-f43e-470b-9059-aa365a1ec93b","arxiv_id":"2607.28964","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"First single-photon experiment constructs non-Hermitian parent Hamiltonians with prescribed matrix-product ground states and observes an intrinsic ground-state transition.","lead":"Using single photons, this experiment realizes non-Hermitian parent Hamiltonians whose left and right ground states are chosen matrix product states, the first such realization. It then uses a three-site model to show a ground-state level-crossing transition, a step toward designing custom non-Hermitian phases.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Fixed τ=10 projection lacks convergence check near N=3 level crossings; without a residual-population bound, reconstructed 'ground states' may mix excited eigenstates and the observed order-parameter jumps could be partly an artifact.","rationale":"The reader's weakest assumption is precisely the fixed τ=10 projection without a convergence check. This is the most load-bearing concern because the N=3 intrinsic phase transition is the headline application; if the projection is incomplete near level crossings, the central claim loses its experimental support. The paper has independent support in the N=2 direct MPS-preparation cross-check and the parameter-free agreement of four order parameters, which lowers circularity concerns. However, none of that validates the N=3 transition regions, where the gap closes. The proposed test (τ-scaling and residual-population computation) is concrete and would settle whether the concern lands. Since the reader already issued CONDITIONAL, and this concern does not change the verdict, I keep it UNCHANGED.","tokens_in":13386,"tokens_out":11163,"duration_ms":102260,"concrete_test":"Using the explicit N=3 H from the Supplemental Material, compute ρτ ∝ Ũτ (I/8) Ũτ† for τ=5,10,15,20 (with Ũτ the passive normalized evolution) at φ densely sampled around each crossing. Evaluate the fidelity to the exact right ground state, w_R(φ,τ)=⟨Ψ~R|ρτ|Ψ~R⟩/Tr(ρτ) (similarly w_L for the left ground state), and the residual weight on all excited eigenstates. If at τ=10 any w < 0.95, or if the order parameter ⟨Ψ~L|O|Ψ~R⟩/⟨Ψ~L|Ψ~R⟩ computed from ρτ shifts by more than the experimental error between τ=10 and τ=20, the fixed-τ projection is insufficient and the phase-transition claim is not established by the current data.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central phase-transition claim rests on the assumption that Uτ=e^{-Hτ} and Uτ† project ρ0 onto the right/left ground states of H for every φ scanned, including φ near π/6, π/3, 2π/3, 5π/6 where the real-part gap can vanish. The paper fixes τ=10 and gives no convergence test, no residual excited-state population estimate, and no τ-scaling check. This matters specifically because in non-Hermitian systems the suppression of excited states is only e^{-τΔ} in the diagonalizable case; at a level crossing the Hamiltonian can be defective, and e^{-Hτ} then contains polynomial-in-τ Jordan-block factors, so even large τ may not yield a pure ground state. If the reconstructed states near the crossing are mixtures of the designated zero-energy MPS and the newly emerged negative-energy eigenstates, the measured order parameter would be a weighted average and the observed abrupt jump could be an artifact of incomplete projection rather than the intrinsic transition. The N=2 data are less vulnerable (gap 1, direct MPS cross-check), but the N=3 result is the demonstrative application of the title.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports the first experimental realization of non-Hermitian parent Hamiltonians (NH-PHs) using single-photon interferometry. The authors start from a pair of matrix product states (MPSs) that serve as the designed left and right zero-energy ground states, construct the corresponding NH-PH following Ref. [48], and implement imaginary-time evolution through passive, gain-free non-unitary operations. For an N=2 asymmetric AKLT model, they reconstruct the left and right ground states by quantum state tomography and measure four order parameters (non-reciprocal, chiral, and antiferromagnetic) as a function of the asymmetry parameter μ, finding agreement with the analytic predictions. They also provide a direct MPS-preparation cross-check. For an N=3 model, they observe level crossings in the complex spectrum and abrupt jumps in a ferromagnetic order parameter, which they interpret as an intrinsic non-Hermitian phase transition. The paper claims to be the first experimental generation and characterization of NH-PHs with controllable and customizable properties.","tokens_in":13713,"tokens_out":12482,"duration_ms":102130,"significance":"If the results hold, the paper provides a useful proof-of-principle that the non-Hermitian parent Hamiltonian construction of Ref. [48] can be implemented in a photonic platform, enabling the reverse engineering of non-Hermitian Hamiltonians with prescribed biorthogonal ground states. The N=2 direct MPS-preparation cross-check is a particular strength, as it independently verifies that the target MPS states carry the designed correlations. The extension to N=3 and the observation of level-crossing-driven order-parameter jumps is conceptually interesting, but its interpretation as an 'intrinsic phase transition' rests on finite-system data and on the adequacy of the fixed imaginary-time evolution. The validation is largely a consistency check by construction, because the theoretical curves are computed from the same MPS states used to build the Hamiltonian; nevertheless, the experimental implementation of non-unitary evolution and the agreement with exact-diagonalization curves are nontrivial. With additional convergence and finite-size analysis, the paper would be a valuable contribution to non-Hermitian quantum simulation.","major_comments":[{"comment":"The imaginary-time evolution is performed at a single fixed value τ=10, and no convergence criterion is given. In the N=3 model the real-part gap closes at the boundaries of φ∈(π/6,π/3) and φ∈(2π/3,5π/6) (Fig. 3). At such level crossings the Hamiltonian may become defective, in which case e^{-Hτ} contains polynomial-in-τ Jordan-block factors and a fixed, large τ is not guaranteed to project onto the zero-energy MPS. Without a residual excited-state population bound or a τ-scaling test, the abrupt jumps in ⟨Õ⟩ could in principle be an artifact of incomplete projection rather than the claimed intrinsic transition. Please add: (i) the N=3 real-part gap as a function of φ; (ii) the residual population of the designated zero-energy MPS after e^{-Hτ} at τ=10; and (iii) a comparison at least for τ=5,20,50 or state fidelities to the exact ground states.","section":"Experimental implementation; Extension to a new N=3 model"},{"comment":"The claim that the transition is 'not a finite-size effect' is supported only by a pointer to the Supplemental Material for N=4,6,10. No gap scaling, discontinuity size, or level-crossing data are presented. Since the transition is defined by finite-system level crossings and the authors explicitly separate it from conventional thermodynamic transitions, the word 'intrinsic' and the robustness statement need quantitative support. Please include a figure or table showing, for N=3,4,6,10 (and possibly larger N), the persistence of the level crossing and the behavior of the order-parameter jump with system size.","section":"Extension to a new N=3 model"},{"comment":"The theoretical curves are computed from the same MPS pair used to construct the Hamiltonian. Agreement therefore validates the state-preparation and tomography chain, but is partly a consistency check rather than an independent confirmation of the NH-PH construction. The direct MPS-preparation cross-check for N=2 is a good control. To support the 'first experimental generation' claim, I recommend reporting state fidelities F(|Ψ_s^R⟩,|R⟩) and F(|Ψ_s^L⟩,|L⟩) for both N=2 and N=3, or an equivalent quantitative closeness measure, so the reader can assess how faithfully the imaginary-time evolution reproduces the designed zero modes.","section":"Experimental results; Fig. 2 and Fig. 3"}],"minor_comments":[{"comment":"The definitions σ± = σx ± iσy are inconsistent with the explicit /2 in O_left and O_right and with the quoted predictions (−μ/4, −1/(4μ)) unless σ± are normalized as (σx ± iσy)/2. Please fix the notation.","section":"Experimental results"},{"comment":"Specify the energy scale: 'τ=10 in natural units' is vague without the norm of H; the reader cannot assess whether 10 is large relative to the inverse gap. A statement of the smallest real-part gap encountered in the scanned parameter range would help.","section":"Experimental implementation"},{"comment":"The phrase 'first experimental generation/realization' appears twice. Given the consistency-check nature of the validation, a more hedged phrasing such as 'first photonic implementation' would be safer unless the absence of prior experimental work is explicitly established.","section":"Introduction; Conclusion"},{"comment":"Typo in caption: 'MPSs tensors' should be 'MPS tensors'. Also, the relation between the boundary conditions and the tensor elements in Eq. (7) is not fully spelled out; a brief derivation would improve reproducibility.","section":"Fig. 4"}],"recommendation":"major_revision","confidential_remarks":"The paper is in scope and addresses a timely topic. The main technical risk is the fixed-τ projection protocol in the N=3 experiment: without a residual-population estimate or τ-scaling check, the phase-transition interpretation is not fully supported. The finite-size robustness claim is also under-supported in the main text. These issues are fixable within the manuscript's scope, so I recommend major revision rather than rejection. The novelty claim (first experimental realization of NH-PHs) should be verified against the literature by the editor, as I am not aware of prior experimental work but have not exhaustively checked all platforms."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know this paper reports the first experimental implementation of non-Hermitian parent Hamiltonians (NH-PHs) using single photons. The N=2 example is the same as in Shen, Guo, Yang (PRL 2023), so the theoretical construction isn't new; the experimental realization and the N=3 model are. The paper does a clean job validating N=2: it measures four order parameters across μ, extracts entanglement spectra, and — importantly — prepares the target MPS states directly via quantum circuits and shows matching expectation values. That cross-check is real evidence that the imaginary-time evolution is doing what's claimed.\n\nWhere it gets softer is the N=3 phase-transition claim. The theoretical curves in Fig. 3 come from exact diagonalization of the constructed Hamiltonian, so agreement with experiment is a consistency check of the state-preparation apparatus rather than an independent test of the physics. The bigger issue is that the imaginary-time evolution uses a fixed τ=10 with no convergence analysis. Near the level crossings at φ=π/6, π/3, etc., the real-part gap can be small, and the suppression of excited states goes as e^{-2τΔ}. If τ is not long enough, the reconstructed state is a mixture and the observed order-parameter jump could be softened or shifted. The stress-test note raises the possibility of polynomial factors if the Hamiltonian becomes defective at the crossing; that's a real theoretical concern but not obviously applicable here — the paper neither reports exceptional points nor gives a τ-scan. A simple convergence plot or residual-population estimate would settle it. Without that, the N=3 result is suggestive but not fully convincing.\n\nAlso minor: they call it a phase transition, but they carefully say it's a finite-system level crossing distinct from thermodynamic transitions. That's honest. The N=4,6,10 numerical checks are only in the supplement, so the 'not a finite-size effect' claim isn't in the main text.\n\nAll that said, the paper deserves a serious referee. The experimental apparatus is nontrivial and the N=2 cross-check is genuinely good. The N=3 data are consistent with theory, and the authors flag caveats. A referee should ask for the τ-convergence analysis and perhaps state fidelities, but the work is a legitimate first demonstration of NH-PH engineering.","headline":"First photonic realization of non-Hermitian parent Hamiltonians; the N=2 validation is solid and the N=3 level-crossing claim is interesting but needs a convergence check for the fixed imaginary-time evolution.","tokens_in":14157,"tokens_out":3891,"would_cite":true,"duration_ms":33388,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Researchers demonstrate the first experimental generation of non-Hermitian parent Hamiltonians, using single-photon interferometry to realize Hamiltonians whose left and right ground states are a prescribed matrix product pair.","keywords":["non-Hermitian parent Hamiltonian","matrix product states","imaginary-time evolution","asymmetric AKLT state","single-photon interferometry","non-Hermitian phase transition","biorthogonal quantum mechanics","chiral correlations"],"falsifier":"Compute the excited-state population in the evolved state ρ_τ at the N=3 transition points for τ=10; if it is not exponentially small relative to the ground-state population, the observed order-parameter jumps are consistent with incomplete projection. Alternatively, repeat the experiment with τ=20 and check that the jump locations in φ remain unchanged.","tokens_in":13306,"feed_emoji":"⚛️","tokens_out":7226,"duration_ms":58535,"temperature":0.7,"pith_summary":"The paper reports the first experimental construction of non-Hermitian parent Hamiltonians (NH-PHs) from a prescribed pair of matrix product states (MPSs). Using single-photon interferometry and imaginary-time evolution, the authors realize a two-site asymmetric AKLT model and measure order parameters that match the designed chiral, non-reciprocal, and antiferromagnetic correlations. Extending to a three-site model, they observe an intrinsic non-Hermitian phase transition signaled by abrupt jumps in an order parameter when the designed zero-energy state ceases to be the global ground state. The work demonstrates that non-Hermitian Hamiltonians can be reverse-engineered to have user-specified ground states, opening a route to engineered non-Hermitian phases and phase transitions.","feed_headline":"Researchers build first tailored non-Hermitian Hamiltonians","feed_subtitle":"Photonic setup builds Hamiltonians from preset matrix product states and detects a non-Hermitian phase transition.","key_machinery":"The load-bearing object is the non-Hermitian parent Hamiltonian (NH-PH), analytically constructed from a pair of MPSs that serve as the desired left and right zero-energy ground states. The experimental machinery is single-photon interferometry implementing imaginary-time evolution U_τ = e^{-Hτ} via a passive, gain-free mapping obtained by singular value decomposition: the unitary factors are realized with beam displacers and half-wave plates, while a diagonal loss matrix implements mode-selective attenuation. For the N=2 asymmetric AKLT state, the asymmetry parameter μ in the virtual bond controls non-Hermiticity; for the N=3 model, the phase φ in the local tensor drives the spectral crossi","core_discovery":"The central claim is that the non-Hermitian parent Hamiltonian method—which analytically constructs a local Hamiltonian whose left and right zero-energy ground states are a given pair of MPSs—can be implemented experimentally with single photons, and that the resulting states faithfully reproduce the designed properties. For the N=2 asymmetric AKLT chain, the measured order parameters ⟨O_AF⟩, ⟨O_left⟩, ⟨O_right⟩, and ⟨O_chiral⟩ agree with theoretical predictions across the asymmetry parameter μ. For the N=3 model, the ground-state energy and a ferromagnetic order parameter exhibit abrupt jumps at spectral crossings, providing experimental evidence of an intrinsic non-Hermitian phase transiti","pith_inferences":["A testable extension is to repeat the N=3 scan with different imaginary-time evolution lengths (e.g., τ=5 and τ=20) to verify that the order-parameter jumps stay put; if they shift, the reported transition could be an artifact of incomplete projection rather than an intrinsic property.","The asymmetric-AKLT design suggests that longer chains built from the same tensors should exhibit a non-zero chiral order parameter in the bulk; measuring this on a photonic platform with more qubits is a natural next step.","Because the parent-Hamiltonian construction fixes left and right ground states independently, the scheme could be adapted to engineer transport or state-transfer channels that exploit a chosen biorthogonal pair—a direction the paper leaves implicit.","The demonstration that a passive loss-only operation can implement e^{-Hτ} for an arbitrary local H suggests the same optics toolbox can simulate other non-unitary processes, such as open-system dynamics, without gain."],"forward_implications":["The same photonic platform can construct parent Hamiltonians for arbitrary MPS pairs, enabling systematic studies of designed non-Hermitian phases beyond the AKLT and N=3 examples.","The order-parameter jumps observed at spectral crossings provide a direct signature of the breakdown of the non-Hermitian variational principle, which the authors connect to enhanced sensing near criticality.","The direct MPS-to-circuit mapping demonstrated in the cross-check offers a general recipe for preparing matrix product states on small quantum circuits.","The method is not limited to non-Hermitian systems: the authors note it can prepare ground states of Hermitian Hamiltonians by the same imaginary-time evolution, broadening its use as a quantum simulator.","The persistence of the phase transition for N=4, 6, and 10 (verified numerically) indicates the effect is not a finite-size artifact and should survive in larger implementations."],"fun_headline_variants":["Photons build first non-Hermitian parent Hamiltonians","First experimental non-Hermitian Hamiltonians from photons","Non-Hermitian phase transition observed in photonic system","Single photons realize tailored non-Hermitian Hamiltonians","Photonic setup demonstrates non-Hermitian parent Hamiltonian"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The experiment assumes that a fixed imaginary-time evolution of τ=10 fully projects the initial mixed state onto the targeted ground state for every scanned parameter, including near spectral crossings where the energy gap shrinks.","fun_headline_variants_meta":{"raw":{"variants":["Photons build first non-Hermitian parent Hamiltonians","First experimental non-Hermitian Hamiltonians from photons","Non-Hermitian phase transition observed in photonic system","Single photons realize tailored non-Hermitian Hamiltonians","Photonic setup demonstrates non-Hermitian parent Hamiltonian"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000437,"raw_usage":{"total_tokens":2059,"prompt_tokens":747,"completion_tokens":1312,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":491,"completion_tokens_details":{"reasoning_tokens":1234}},"tokens_in":491,"tokens_out":1312,"duration_ms":8690,"temperature":1.0,"reasoning_tokens":1234,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T16:22:02.966159+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the excited-state population in the evolved state ρ_τ at the N=3 transition points for τ=10; if it is not exponentially small relative to the ground-state population, the observed order-parameter jumps are consistent with incomplete projection. Alternatively, repeat the experiment with τ=20 and check that the jump locations in φ remain unchanged.","supporting_citations":[],"review_version":1}