{"id":"3e12ffcf-168e-4ab1-88e7-4669614d1f2e","arxiv_id":"2611.16430","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Half-filled antiphase stripes in the 2D Hubbard model produce local d-wave pairing but no resolved long-range superconducting order up to 32x32 sites.","lead":"Using a new computational method, this paper finds that in a model of high-temperature superconductors, electron stripes generate local pairs but do not lead to long-range superconductivity. The result helps clarify a decades-old debate about whether stripes help or hinder superconductivity in cuprates.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek/deepseek-v4-flash","headline":"Transferred-parameter and shallow-projection limitations on 32×32 leave the central claim about long-range pairing unresolved; the states studied may not yet be representative of the ground state's pairing correlations.","rationale":"The reader identified the weakest assumption as the expressiveness of the wavefunction ansatz for capturing long-range pairing. I agree that this is a concern, but the paper partially addresses it by demonstrating competitive variational energies and by validating that local pairing is robust across different K and symmetry projections. My more specific, load-bearing concern is about the adequacy of the variational optimization specifically for the long-range pairing channel on large lattices, given that the evidence on 24×24 and 32×32 relies on transferred parameters and shallower projection depths. The paper itself flags this limitation (sentences b75, b79, b107) and calls for independent per-size re-optimization. Because the central claim about unresolved long-range coherence is crucial for the paper's conclusion about a clear separation between local pair formation and phase coherence, and because the existing evidence for that part of the claim is explicitly tentative, I recommend a CONDITIONAL verdict: accept the local pairing findings as robust, but condition acceptance of the broader claim about inter-stripe coherence on the outcome of the proposed concrete test. This adjustment reflects that the paper's most novel result—that half-filled walls intrinsically generate local d-wave pairs—is well supported, while the assertion about the absence of long-range order awaits stronger computational confirmation.","tokens_in":20922,"tokens_out":2809,"duration_ms":24018,"concrete_test":"Perform independent variational re-optimization of the MR-nVAFQMC wavefunction on the 32×32 lattice (t′/t = −0.2, δ = 1/8) at the same deep projection depth τ ≈ 1.26 used for the 32×32 energy benchmark in Table I, with K = 4 or K = 8 references, and symmetry group extended to include stripe translations (|G| ≥ 64). Evaluate the four-bond site-centered distant-window pairing Δ²_SC,4b(L) with sufficient statistics (∼100 independent Markov chains or equivalent) to achieve a 1σ uncertainty below 0.03×10⁻³. If the result is statistically consistent with zero at that precision, the claim of no resolved long-range coherence is strengthened; if a positive signal emerges (Δ²_SC,4b > 0.05×10⁻³ at >2σ), the central claim would need revision.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The paper's central claim that no resolved long-range pairing coherence exists across stripes rests critically on the exploratory fixed-depth transferred-parameter series (Fig. 5B,C) and the unified same-checkpoint evaluations. The authors themselves state that the 24×24 and 32×32 wavefunctions in the transferred series are constructed by tiling and transferring parameters optimized on 16×16 without independent variational re-optimization (steps = 0 on 32×32) and use shallower projection depths (τ = 1.15–1.16) than the energy-benchmark checkpoints (τ = 1.26–1.45). The unified same-checkpoint evaluations on the Table I SOTA wavefunctions cover only L = 16 and L = 32, not intermediate sizes, and the statistical uncertainties on the four-bond Δ²_SC,4b are large enough that the predicted weak superconductivity scale from NN-fPEPS (Δ² ∼ 0.05×10⁻³) lies within the noise. The paper openly acknowledges that establishing fully converged ground-state scaling requires independent per-size variational re-optimization at matched deep τ. Without that re-optimization, the failure to resolve long-range pairing may reflect insufficient optimization or projector depth rather than a physical absence of order. This concern does not affect the robust local pairing result, but it weakens the evidence for the second half of the central claim, i.e., that no long-range coherence exists in the true ground state.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek/deepseek-v4-flash","summary":"The paper develops multi-reference neural variational auxiliary-field quantum Monte Carlo (MR-nVAFQMC) to study the 2D t–t′–U Hubbard model at U/t=8 on periodic lattices up to 32×32. By superposing unpaired Slater determinants across stripe periods and domain-wall positions, the method reaches variational energies competitive with leading neural- and tensor-network benchmarks without explicit pairing fields. The central results are: (i) negative t′/t induces a switch from filled to half-filled antiphase stripes at 1/8 hole doping, stabilized by kinetic delocalization; (ii) each half-filled domain wall intrinsically hosts short-range d-wave pairs, independent of whether the stripe is pinned or coherently delocalized; (iii) across antiferromagnetic domains, pairing correlations decay within one stripe period and long-range coherence is not resolved in the states studied, including exploratory transferred-parameter series up to 32×32. The authors openly discuss caveats about transferred parameters, projector depth, and statistical noise, and outline scenarios that could reconcile their single-layer findings with experimental observations.","tokens_in":21368,"tokens_out":1385,"duration_ms":13387,"significance":"If the results hold, the paper provides a clear microscopic separation between local pair formation (intrinsic to half-filled antiphase walls) and inter-stripe phase coherence in the Hubbard model, directly addressing the long-standing debate about whether stripes compete with or promote superconductivity. The demonstration that a multi-reference ansatz without explicit pairing fields can match or improve upon the best neural and tensor-network benchmarks on 256- and 1024-site tori is a methodological advance. The candid discussion of limitations—transferred parameters, projector depth, noise floor for weak order parameters—strengthens the work by delineating what is robust (local pairing) from what remains unresolved (long-range pairing).","major_comments":[],"minor_comments":[{"comment":"The unit 'mt' (millit = 10^-3 t) is used frequently but defined only implicitly in the text (e.g., 'meV' comparison is absent). Explicitly state at first use that 1 mt = 10^{-3} t to avoid ambiguity for readers less familiar with the convention.","section":"Abstract and throughout","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The caption for Fig. 3 uses 'χvtxSC(R≤4)' without defining the symbol in the main text; 'χvtxSC(R≤4)' is defined later in §III but a forward reference or brief parenthetical explanation would improve readability.","section":"Fig. 3 caption","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The definition of M_L for the distant-window average could be given explicitly for L=16,24,32 to aid reproducibility; the text states M_L = 159, 355, and 623 but the corresponding integer sequence is not derived (probably the number of sites with |r| ≥ L/(2√2)). A brief note on how it is computed would help.","section":"§III B, Eq. (7)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The caption describes |G| for some states but the notation '|G| = 8' versus '|G| = 64,128' could be clarified: list which symmetry generators are included for each group size.","section":"Table I caption","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The sentence referencing 'the Sorella–Tocchio result' might be simplified: the reader's report notes the citation chain is clear, but the text could be slightly shortened without loss of precision.","section":"§II C, 'Sorella–Tocchio result'","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"accept"},"author_rebuttal":null,"desk_editor":{"model":"deepseek/deepseek-v4-flash","letter":"This paper does something genuinely useful: it builds a multi-reference extension of VAFQMC and uses it to settle that half-filled antiphase walls in the 2D Hubbard model at U/t=8, t'/t=-0.2, δ=1/8 host intrinsic short-range d-wave pairing. The unpinning experiment (Fig. 3) is clean and convincing—local pairing survives even when static charge modulation is suppressed by coherent delocalization. The energetic competition between filled and half-filled stripes is well explained via kinetic frustration. Credit where due: this is a real methodological advance applied to a physically sharp question.\n\nThe soft spot is the second half of the central claim—that no long-range pairing coherence exists across stripes. The pressure-test concern lands exactly here. The exploratory series (Fig. 5B,C) uses transferred parameters without per-size re-optimization at shallower projection depth than the energy benchmarks, and the unified SOTA checkpoints cover only L=16 and L=32. The authors state these limitations openly (they label the series “exploratory” and say independent re-optimization is needed), so my criticism is not that they hid something—it’s that the evidence for the absence of long-range order is not yet commensurate with the strength of the claim. The statistical noise on Δ²_SC,4b is large enough that the weak superconductivity scale from NN-fPEPS (Δ² ∼ 0.05×10⁻³) is within uncertainties. A careful reader should treat the long-range result as unresolved rather than definitively absent.\n\nThe local-pairing result is solid and will survive. The long-range question remains open for a deeper, re-optimized study. I would bring this to a reading group because the methodology and the unpinning analysis are worth discussing seriously. The paper deserves peer review: the method is novel, the local-pairing argument is well-supported, and the limitations are honestly stated. The authors are serious thinkers—their conclusions about local pairing are cautious and data-anchored.\n\nRecommendation: accept with revision. Require the authors to either upgrade the long-range scaling with properly re-optimized states (which they acknowledge is future work) or rephrase the central claim so that it unambiguously reports an unresolved result on long-range order, not a negative detection.","headline":"Careful numerical work that cleanly establishes local d-wave pairing in half-filled stripes, but the claim about absence of long-range coherence rests on under-converged exploratory data that the authors themselves flag as such.","tokens_in":21817,"tokens_out":863,"duration_ms":8577},"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek/deepseek-v4-flash","headline":"The paper uses multi-reference neural variational auxiliary-field quantum Monte Carlo to study stripe and pairing correlations in the Hubbard model at 1/8 doping.","keywords":["Hubbard model","stripe order","d-wave pairing","multi-reference variational","auxiliar-field quantum Monte Carlo","half-filled domain walls","cuprate superconductivity","neural quantum states"],"falsifier":"A direct calculatio on a 48x48 torus with independently optimized multi-reference states at deeper τ that recovers a non-zero uniform d-wave pair order parameter (Δ2_SC,4b > 0.001) with reasonable statistics would contradict the paper's conclusion that long-range pairing is unresolved. Alternatively, an observation of phase-coherent supercurrents across half-filled domain walls in a cold-atom emulation would falsify the claim that inter-stripe coherence is absent.","tokens_in":23936,"feed_emoji":"","tokens_out":1388,"duration_ms":48345,"temperature":0.7,"pith_summary":"The paper develops a multi-reference neural variational auxiliary-field quantum Monte Carlo (MR-nVAFQMC) method to study the two-dimensional t-t'-U Hubbard model at hole doping δ=1/8 with U/t=8. For cuprate-relevant hopping t'/t=-0.2, it finds that the ground state selects half-filled antiphase domain walls (charge period 4, spin period 8) rather than filled stripes. Each half-filled wall intrinsically generates strong short-range d-wave pairing, but the connected pair correlation decays rapidly and becomes unresolved beyond one stripe period. Coherently unpinning the walls suppresses static charge modulation without altering local pairing, showing the pairing is intrinsic to the half-filled geometry. The variational energies compete with leading benchmarks up to 32x32 sites, yet no long-range superconducting order is resolved.","feed_headline":"Half-filled walls host short-range d-wave pairing","feed_subtitle":"Multi-reference wavefunctions reveal intrinsic pairing without long-range order","key_machinery":"The key machinery is the multi-reference neural variational auxiliary-field quantum Monte Carlo (MR-nVAFQMC) wavefunction, which superposes K (up to 16) canonical particle-conserving Slater determinants spanning different stripe periods and domain-wall positions, combined with space-group symmetry projection and a shared neural-network-parametrized auxiliary-field projector of imaginary time depth τ. This architecture allows the variational state to spontaneously select the energetically optimal stripe configuration and to distinguish coherent quantum superpositions from classical mixtures of shifted walls.","core_discovery":"The central claim is that in the two-dimensional t-t'-U Hubbard model at δ=1/8 and t'/t=-0.2, the ground state consists of half-filled antiphase domain walls (λc=4, λs=8) that inherently generate local d-wave pairing, but the pairing correlation decays rapidly and no long-range phase coherence is established across the stripes. This conclusion is supported by controlled comparisons between pinned and delocalized wall configurations and by competitive variational energies from a wavefunction built without explicit pairing fields.","pith_inferences":["The paper's transferred-parameter scaling to 24x24 and 32x32 without per-size re-optimization leaves open the possibility that fully relaxed large-lattice states could resolve weak long-range pairing, as reported by some competing methods.","If future calculations with independent optimization on larger lattices confirm the absence of long-range order, the single-band Hubbard model would be an incomplete model for cuprate superconductivity, necessitating extended Hamiltonians.","The observation of local pairing without resolved long-range coherence aligns with experimental 2D superconducting fluctuations in stripe-ordered cuprates, but the paper does not directly address the Berezinskii-Kosterlitz-Thouless transition."],"forward_implications":["If the central claim is correct, half-filled antiphase walls are the generic source of local d-wave pairing in the hole-doped Hubbard model wherever they appear.","Suppressing static stripe order (coherent unpinning of walls) does not enhance short-range pairing or induce long-range coherence in this model.","Long-range superconducting order at δ=1/8, if it exists, requires additional physics beyond the single-band t-t'-U model, such as multi-orbital effects or longer-range interactions.","The multi-reference approach without explicit pairing fields achieves competitive variational energies, confirming that pairing correlations can emerge from charge and spin order alone without a geminal initialization.","The near-degeneracy of uniform and period-eight pair-matrix modes on 64-site tori suggests weak inter-wall Josephson coupling, consistent with a pair-density wave scenario but without established phase coherence."],"supporting_citations":[{"why":"Provides the primary compariso of NN-fPEPS and Pfaffian backflow energies and pairing, against which MR-nVAFQMC is benchmarked.","marker":"[10]"},{"why":"Supplies the Sorella VAFQMC method and twist-averaged TDL energy used as a key basis of compariso.","marker":"[19]"},{"why":"Reports coexisiting d-wave superconductivity with partially filled stripes via constraine-path AFQMC, serving as a contrasting scenario.","marker":"[31]"},{"why":"Provides symmetry-preserving BCS backflow results, another high-bench variational approach, for compariso on the same lattices.","marker":"[35]"},{"why":"Covers the V-score dataset and variational benchmarks, including the VAFQMC energy used as a reference.","marker":"[34]"}],"fun_headline_variants":["Half-filled stripes host short-range d-wave pairing","Local d-wave pairing emerges without long-range stripe coherence","Multireference wavefunctions show intrinsic d-wave pairs in stripe walls","Stripe walls generate local d-wave pairs but no phase coherence","Delocalized walls suppress charge order, preserve local d-wave pairing"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The variational state uses a finite number of Slater determinants (K≤16) and a finite projection depth τ, and the large-lattice results rely on transfered parameters without independent re-optimization; if these restrictions bias the state against long-range phase coherence, the absence of long-range pairing could be an artifact rather than a physical property of the model.","fun_headline_variants_meta":{"raw":{"variants":["Half-filled stripes host short-range d-wave pairing","Local d-wave pairing emerges without long-range stripe coherence","Multireference wavefunctions show intrinsic d-wave pairs in stripe walls","Stripe walls generate local d-wave pairs but no phase coherence","Delocalized walls suppress charge order, preserve local d-wave pairing"]},"model":"deepseek/deepseek-v4-flash","effort":"low","cost_usd":0.000298,"raw_usage":{"cost":0.000297808,"is_byok":false,"cost_details":{"upstream_inference_cost":0.000297808,"upstream_inference_prompt_cost":9.9848e-05,"upstream_inference_completions_cost":0.00019796},"total_tokens":1625,"prompt_tokens":918,"completion_tokens":707,"prompt_tokens_details":{"audio_tokens":0,"video_tokens":0,"cached_tokens":256,"cache_write_tokens":0},"completion_tokens_details":{"audio_tokens":0,"image_tokens":0,"reasoning_tokens":623}},"tokens_in":918,"tokens_out":707,"cost_source":"openrouter_usage","duration_ms":6358,"image_pages":[],"temperature":1.0,"billed_attempts":1,"reasoning_tokens":623,"transport_provider":"openrouter","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-11-13T15:13:47.957680+00:00","model_set":{"reader":"deepseek/deepseek-v4-flash"},"pipeline_version":"pith-pipeline@v1.4.1-daily-deepseek-or-medium","input_kind":null,"input_chars":null,"input_total_chars":null,"passes_missing":null,"published_standing":null,"published_confidence":null,"objection_audit":null,"run_record":null,"basis":{"kind":"not_recorded","source":null,"chars_given":null,"chars_total":null},"receipt":null,"falsifier":"A direct calculatio on a 48x48 torus with independently optimized multi-reference states at deeper τ that recovers a non-zero uniform d-wave pair order parameter (Δ2_SC,4b > 0.001) with reasonable statistics would contradict the paper's conclusion that long-range pairing is unresolved. Alternatively, an observation of phase-coherent supercurrents across half-filled domain walls in a cold-atom emulation would falsify the claim that inter-stripe coherence is absent.","supporting_citations":[{"cited_title":"Extreme competition between stripes and superconductivity in the hole-doped Hubbard model","cited_arxiv_id":"2610.30321","evidence_quote":"Provides the primary compariso of NN-fPEPS and Pfaffian backflow energies and pairing, against which MR-nVAFQMC is benchmarked."},{"cited_title":"not variational","cited_arxiv_id":null,"evidence_quote":"Supplies the Sorella VAFQMC method and twist-averaged TDL energy used as a key basis of compariso."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports coexisiting d-wave superconductivity with partially filled stripes via constraine-path AFQMC, serving as a contrasting scenario."},{"cited_title":"Wuet al., Variational benchmarks for quantum many-body problems.Science386, 296–301 (2024)","cited_arxiv_id":null,"evidence_quote":"Provides symmetry-preserving BCS backflow results, another high-bench variational approach, for compariso on the same lattices."},{"cited_title":"Backflow transformations via neural networks for quantum many-body wave functions.Phys. Rev. Lett.122, 226401","cited_arxiv_id":null,"evidence_quote":"Covers the V-score dataset and variational benchmarks, including the VAFQMC energy used as a reference."}],"lane_accuracy":{"status":"not yet measured","lane":{"model":"deepseek/deepseek-v4-flash","input_kind":null},"url":"https://pith.science/accuracy"},"review_version":1}