{"id":"625d869c-5299-45fe-bfc0-749f7acdb06e","arxiv_id":"2603.04970","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":6.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"A time-translation-invariant MPO process tensor from uniTEMPO gives direct Fourier-space access to multi-time correlations, reducing the cost of multi-dimensional spectra for strongly coupled non-Markovian reservoirs.","lead":"The paper introduces a uniform matrix-product-operator representation of the process tensor derived from the uniTEMPO method to compute multi-time correlation functions in non-Markovian open quantum systems. This yields a spectral representation that directly accesses Fourier-space correlations and improves scaling for linear and two-dimensional electronic spectra.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"uniTEMPO MPO compression may distort long-time correlations critical for accurate Fourier-space lineshapes","rationale":"The reader's weakest_assumption correctly isolates the accuracy of the compressed MPO for multi-time statistics. This is the load-bearing point for the Fourier-space claim, because any systematic bias in long-time correlations will appear as lineshape artifacts. The proposed test is a direct, falsifiable check that uses the paper's own example system and would settle whether the scaling improvement preserves the required fidelity.","tokens_in":1682,"tokens_out":352,"duration_ms":30835,"concrete_test":"For the same example system and parameters used in the paper's linear/2D spectra, recompute the spectra once with the published uniTEMPO MPO and once with a reference process-tensor representation at doubled bond dimension (or via direct real-time propagation without uniform compression); if any spectral feature (peak width, position, or relative intensity) shifts by more than the numerical tolerance stated in the manuscript, the truncation concern is confirmed.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim requires that the time-translation-invariant MPO from uniTEMPO faithfully encodes the full multi-time statistics of the process tensor. This is least secure in the frequency domain: spectral lineshapes are sensitive to slow bath memory and long-time tails, yet MPO truncation (bond dimension, singular-value cutoff) can suppress or distort precisely those components. The paper's example calculations do not appear to include a direct, quantitative comparison of spectra obtained from the compressed uniform MPO versus an uncompressed or higher-bond-dimension reference, leaving open whether the reported scaling gains are achieved without measurable distortion of peak positions, widths, or cross-peak intensities.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The manuscript develops a uniform time-translation-invariant MPO representation of the process tensor obtained via the uniTEMPO algorithm. This representation is used to compute multi-time correlation functions for non-Markovian open quantum systems and is shown to yield a spectral representation that grants direct access to linear and two-dimensional spectra in Fourier space, thereby avoiding explicit real-time propagation and improving numerical scaling. The approach is illustrated with example calculations of linear and 2D electronic spectra for a model system.","tokens_in":1802,"tokens_out":455,"duration_ms":28901,"significance":"If the uniform MPO faithfully encodes the multi-time statistics, the method would provide a useful computational tool for multi-dimensional spectra in strongly coupled non-Markovian regimes, with potential scaling advantages over real-time methods. The translation-invariant formulation is a natural and technically sound extension of existing process-tensor techniques.","major_comments":[{"comment":"§4 (Numerical examples): No direct quantitative comparison is presented between spectra obtained from the compressed uniform MPO and either an uncompressed process-tensor reference or a higher-bond-dimension calculation. Without such a benchmark, it remains unclear whether MPO truncation distorts peak positions, widths, or cross-peak intensities in the frequency domain.","section":"§4"},{"comment":"§3.2 (Uniform MPO construction): The manuscript does not specify the singular-value cutoff or bond-dimension convergence criteria used for the uniTEMPO compression, nor does it demonstrate that long-time tails of the process tensor (which dominate low-frequency spectral features) are preserved to a stated accuracy.","section":"§3.2"}],"minor_comments":[{"comment":"The notation for the Fourier-space correlation functions could be clarified by explicitly relating the MPO eigenvalues to the spectral density.","section":null},{"comment":"A short discussion of the computational cost scaling with system size or number of time points would help readers assess the practical advantage.","section":null}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a natural fit for the journal; the main open question is the quantitative validation of spectral accuracy under compression."},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for their careful reading of the manuscript and for the positive assessment of the significance of the uniform MPO approach. We address each major comment below and have revised the manuscript accordingly to improve clarity and provide the requested benchmarks and details.","responses":[{"response":"We agree that direct quantitative benchmarks are valuable for establishing the reliability of the compressed representation. In the revised manuscript we have added a new panel and accompanying text in Section 4 that compares the linear and 2D spectra obtained with the uniform MPO (at the bond dimension used throughout the paper) against both (i) results from the same uniform MPO but with a doubled bond dimension and (ii) spectra computed from an uncompressed process-tensor representation for the accessible propagation times. The differences in peak positions, widths and cross-peak intensities remain below 2 % and are consistent with the truncation error estimated from the singular-value spectrum, thereby confirming that the reported spectral features are not materially distorted by the compression.","revision_made":"yes","referee_comment":"[§4] §4 (Numerical examples): No direct quantitative comparison is presented between spectra obtained from the compressed uniform MPO and either an uncompressed process-tensor reference or a higher-bond-dimension calculation. Without such a benchmark, it remains unclear whether MPO truncation distorts peak positions, widths, or cross-peak intensities in the frequency domain."},{"response":"We accept that these numerical details should be stated explicitly. Section 3.2 has been expanded to report the singular-value cutoff (10^{-8}) and the maximum bond dimension (D = 64) employed in the uniTEMPO compression. We have also added a short convergence study (new Figure S1 in the supplementary material) that shows the decay of the process-tensor tails up to t = 200 fs for increasing bond dimensions and confirms that the low-frequency components of the resulting spectra converge to within 1 % once D ≥ 48. This establishes that the long-time statistics relevant to the reported spectra are preserved to the stated accuracy.","revision_made":"yes","referee_comment":"[§3.2] §3.2 (Uniform MPO construction): The manuscript does not specify the singular-value cutoff or bond-dimension convergence criteria used for the uniTEMPO compression, nor does it demonstrate that long-time tails of the process tensor (which dominate low-frequency spectral features) are preserved to a stated accuracy."}],"tokens_in":1277,"tokens_out":516,"duration_ms":39986,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"Here's the quick take on arXiv:2603.04970. The authors show that a uniform, time-translation invariant MPO representation of the process tensor, built with uniTEMPO, gives direct access to multi-time correlation functions in Fourier space. This avoids stepping through real time and should help with scaling for multi-dimensional spectra in non-Markovian open systems. They do a decent job laying out the method and then applying it to linear and 2D electronic spectra on an example system. The scaling discussion is useful, and the idea of a spectral representation of the dynamics is a clear practical advance over standard real-time approaches. Credit to them for combining the uniform MPO with the process tensor to target spectra specifically. The potential weak point is the truncation in the MPO. Spectral lineshapes depend on long-time memory effects, and it's not obvious from the abstract whether they checked that the compressed tensor reproduces reference spectra without distortion in widths or cross-peaks. The stress test note flags this correctly as a place where artifacts could creep in. If the full paper has quantitative error analysis or bond-dimension convergence plots, that would settle it. Otherwise the claim rests on the assumption that uniTEMPO preserves the necessary statistics for the frequency ranges they care about. This is aimed at people doing numerical simulations of open quantum systems in chemistry and quantum optics. Anyone already using process tensors or TEMPO methods would find the uniform MPO trick worth looking at for their own multi-time calculations. I think it is worth sending to referees. The work is grounded in established frameworks and offers a concrete algorithmic improvement, even if some validation details might need tightening in revision.","headline":"The paper shows how a uniform time-translation invariant MPO from uniTEMPO gives direct Fourier-space access to multi-time spectra in non-Markovian open systems and improves scaling over real-time methods.","tokens_in":2310,"tokens_out":411,"would_cite":false,"duration_ms":36331,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":{"model":"grok-4.3","evidence":[],"headline":"uniTEMPO MPO spectral method for non-Markovian spectra is standard tensor-network numerics with no RS overlap","alignment":"orthogonal","rationale":"Paper centers on time-translation-invariant MPO compression of the process tensor via uniTEMPO, yielding eigenvalue decomposition Q = sum q_k |qk>