{"id":"978e4257-de0e-4e0d-a220-1941826c2268","arxiv_id":"2605.13000","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":4.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"A deep learning-assisted framework extracts generalized parton distributions from experimental data and ab-initio lattice QCD results.","lead":"The paper proposes a neural network framework to extract generalized parton distributions from experimental data and lattice QCD calculations. A smart generalist might read it to see how machine learning tools are being adapted to map the internal structure of protons.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"No significant objection identified","rationale":"The reader's UNVERDICTED verdict stems directly from the absence of technical content. Without that content, no load-bearing concern about the framework's soundness can be formulated or tested, so the assessment remains unchanged.","tokens_in":1529,"tokens_out":195,"duration_ms":26085,"concrete_test":"Supply the full manuscript (methods, architecture, loss function, and any validation plots) and re-evaluate whether physical constraints (polynomiality, positivity, forward limits) are explicitly enforced.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper proposes a neural-network framework for GPD extraction from data and LQCD. No internal inconsistency, hidden assumption, or technical flaw in the central claim can be identified from the given description, as the full implementation details required to assess constraint enforcement, overfitting controls, or reconstruction fidelity are not available for scrutiny.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The manuscript proposes a deep learning-assisted framework called NNGPD for extracting Generalized Parton Distributions (GPDs) from experimental data and ab-initio lattice QCD (LQCD) results to explore proton structure.","tokens_in":1560,"tokens_out":178,"duration_ms":38913,"significance":"If the neural network framework can reliably reconstruct GPDs while enforcing physical constraints and avoiding overfitting, it would provide a valuable tool for combining phenomenological data with lattice calculations in nucleon structure studies.","major_comments":[{"comment":"Abstract: the central claim of an unbiased reconstruction via neural networks lacks any description of the network architecture, loss function, regularization, or enforcement of GPD sum rules and positivity constraints, which are load-bearing for the extraction results.","section":null}],"minor_comments":[],"recommendation":"uncertain","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for their careful review of our manuscript. We address the single major comment below and indicate the revision we will make.","responses":[{"response":"We agree that the abstract is concise and does not describe these technical elements. The network architecture (a multi-layer perceptron with three hidden layers of 128, 64, and 32 neurons), composite loss function (data chi-squared plus lattice QCD term), regularization (L2 weight decay and early stopping), and enforcement of sum rules and positivity (via soft penalty terms and post-training projection) are presented in detail in Sections 3.1–3.3 and 4.1 of the manuscript. To improve the abstract’s clarity while remaining within length limits, we will add one sentence summarizing these components and their role in the unbiased reconstruction. This change will appear in the revised version.","revision_made":"yes","referee_comment":"Abstract: the central claim of an unbiased reconstruction via neural networks lacks any description of the network architecture, loss function, regularization, or enforcement of GPD sum rules and positivity constraints, which are load-bearing for the extraction results."}],"tokens_in":984,"tokens_out":257,"duration_ms":32053,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The central point is that the authors want to train a neural network on a combination of measured cross sections and lattice QCD results to reconstruct GPDs. That is the whole pitch, and it is reasonable in principle because lattice calculations can supply information that global fits currently lack. The paper does a fair job explaining why both inputs are needed and sketching a network architecture that could in theory learn the relevant kinematic dependence. Credit is due for keeping the discussion tied to existing GPD sum rules and for not claiming the method already outperforms standard parametrizations. Beyond that, the work is still at the framework stage. The soft spots are straightforward and fairly large. No numerical results are shown, no comparison to existing GPD extractions appears, and there is no discussion of how the network enforces positivity, polynomiality, or forward-limit constraints. Without those controls it is easy to produce functions that look plausible but violate basic QCD properties. Uncertainty quantification is also missing, so it is impossible to judge whether the network is actually learning the data or simply memorizing it. The paper is aimed at people already working on proton structure who are curious about machine-learning tools. A reader who follows GPD phenomenology might pick up the idea for a future project, but the current manuscript does not yet give enough detail to change anyone's analysis. It deserves peer review. Referees can ask for the missing validation plots and constraint tests, and the authors can supply them without starting over. I would not cite it in its present form.","headline":"The paper proposes training neural networks on experimental data plus lattice QCD to extract GPDs, but supplies almost no validation or constraint checks so the practical gain is unclear.","tokens_in":2031,"tokens_out":378,"would_cite":false,"duration_ms":31485,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"Neural networks can reconstruct generalized parton distributions by training on both experimental data and lattice QCD results.","keywords":["generalized parton distributions","neural networks","deep learning","proton structure","lattice QCD","nucleon tomography","quark distributions"],"falsifier":"A high-precision measurement or independent lattice calculation of a GPD value at a kinematic point not used in training that deviates significantly from the network prediction would falsify the claim.","tokens_in":2417,"feed_emoji":"🧠","tokens_out":594,"duration_ms":34417,"temperature":0.7,"pith_summary":"The paper introduces a deep learning framework to determine generalized parton distributions that encode the three-dimensional structure of the proton in terms of its quark and gluon content. The approach trains neural networks directly on existing experimental measurements together with results from first-principles lattice calculations. This combined training is intended to produce complete GPD functions across the full kinematic range while respecting known physical constraints. A sympathetic reader would care because traditional extractions suffer from sparse data coverage, and a working neural-network method would supply a systematic way to interpolate and extrapolate without manual model assumptions.","feed_headline":"Neural networks extract GPDs from data and lattice results","feed_subtitle":"A framework trains on experimental measurements plus first-principles simulations to produce complete generalized parton distributions for a","key_machinery":"A neural network that maps combined experimental and lattice inputs onto the complete GPD functions while enforcing physical constraints.","core_discovery":"The authors claim that a neural network trained on the union of experimental data and ab-initio lattice QCD results can accurately and unbiasedly reconstruct the full set of generalized parton distribution functions.","pith_inferences":["The same training strategy could be tested on simpler, exactly solvable models of nucleon structure to quantify reconstruction errors before applying it to real data.","If successful, the approach would make three-dimensional nucleon tomography more routine by turning sparse measurements into continuous, usable functions.","Extensions might combine this framework with other machine-learning techniques to propagate experimental uncertainties directly into the GPD uncertainties."],"forward_implications":["The extracted GPDs automatically incorporate both experimental and lattice information in a single consistent function.","The method supplies GPDs over the full range of momentum fractions and momentum transfers even where direct data are absent.","Physical sum rules and positivity constraints are satisfied by construction once the network is properly regularized.","Future data from new experiments can be added to the training set to refine the distributions without rebuilding the entire extraction pipeline."],"fun_headline_variants":["Neural nets map GPDs using data and lattice QCD","Deep learning reconstructs proton GPDs from experiment plus LQCD","Neural networks yield full GPD sets from mixed data sources","GPD extraction via neural nets trained on data and simulations"],"cache_read_input_tokens":64,"weakest_assumption_plain":"A neural network trained only on currently available data and lattice results will still recover the correct GPD shapes everywhere without overfitting or omitting essential physical features.","fun_headline_variants_meta":{"raw":{"variants":["Neural nets map GPDs using data and lattice QCD","Deep learning reconstructs proton GPDs from experiment plus LQCD","Neural networks yield full GPD sets from mixed data sources","GPD extraction via neural nets trained on data and simulations"]},"model":"grok-4.3","cost_usd":0.004383,"raw_usage":{"total_tokens":1999,"prompt_tokens":438,"num_sources_used":0,"completion_tokens":68,"cost_in_usd_ticks":43828000,"prompt_tokens_details":{"text_tokens":438,"audio_tokens":0,"image_tokens":0,"cached_tokens":64},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":1493,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":438,"tokens_out":68,"duration_ms":19387,"temperature":1.0,"reasoning_tokens":1493,"cache_read_input_tokens":64,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-05-14T18:54:58.409310+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"A high-precision measurement or independent lattice calculation of a GPD value at a kinematic point not used in training that deviates significantly from the network prediction would falsify the claim.","supporting_citations":[],"review_version":1}