{"id":"dc73c21c-b650-4479-a993-1d8a47bd9fc9","arxiv_id":"2509.02810","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A hybrid atomic memory using GEM for write-in and EIT for readout, and the reverse, maps a pulse's spectrum to its arrival time and its timing to output frequency.","lead":"The authors combined two established quantum memory techniques, gradient echo memory and electromagnetically induced transparency, in a single rubidium memory to convert light pulses between the time and frequency domains. The hybrid scheme could help quantum networks swap between time-bin and frequency encodings without a separate converter.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central quantum-memory claim is not established: all reported evidence is bright-field heterodyne data with no single-photon test, no retrieval efficiency, and no nonclassicality witness, so the demonstrated reversible time-frequency mapping is only classical.","rationale":"The central claim is that the device is a hybrid quantum memory and reversible time-frequency transducer. For that claim to hold, the demonstrated GEM-to-EIT and EIT-to-GEM mappings must work at the single-photon level and preserve the relevant quantum coherences. The paper's evidence consists entirely of coherently averaged heterodyne signals from bright pulses; no efficiency, no photon-counting, no nonclassicality witness, and no error bars are reported. That is a direct gap between the evidence and the strongest claim, not a disagreement with an external consensus. I therefore identify the missing single-photon/nonclassicality characterization as the load-bearing concern. The reader's weakest assumption points to exactly this gap, so I agree with the conditional verdict. The classical frequency-time conversion itself is supported by the data and simulations; hence no reject-level objection. A single-photon or weak-coherent test would settle the quantum-memory question; until then, the manuscript's title and abstract should either be softened or the supporting experiment added.","tokens_in":8274,"tokens_out":6948,"duration_ms":86271,"concrete_test":"Repeat the GEM-to-EIT and EIT-to-GEM sequences with input attenuated to mean photon number mu ~ 1 (or with a heralded single-photon source), detect the retrieved mode with photon counting or homodyne, and compare the mapped temporal/frequency profile and the retrieved state's second-order correlation g^(2) with the bright-field result. If the single-photon-level retrieval shows the same delay-frequency mapping and g^(2)(0) < 1 (or nonclassical cross-correlation with the input), the quantum-memory claim is supported; if the mapping smears out or g^(2) indicates classical/thermal statistics, the title and abstract overclaim and the result should be described as a classical frequency-time converter.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing assumption is that the linear time-frequency transformation demonstrated with bright coherent pulses (Secs. IV-V) is also a quantum memory transformation, i.e., it preserves coherence and reversibility at the single-photon level and can be used as a transducer for quantum states. Nothing in the reported data supports this: detection is heterodyne on classical fields, averages are coherent over 200 sequences, no storage/retrieval efficiency is quoted, no error bars are shown, and no g^(2) or other nonclassicality witness is reported. The Maxwell-Bloch simulations (Eqs. 5-6) are classical field equations and cannot certify quantum operation. Moreover, the protocol's usefulness as a quantum communication tool depends on the stored coherence retaining phase and amplitude for weak fields; this is precisely what is untested. If the GEM/EIT mapping is lossy or decoheres at low photon number, the 'quantum memory' claim fails even though the classical frequency-time converter may work. The paper's own conclusion only claims 'proper realization of the presented protocol' in the current classical parameter regime, and this is the fair reading.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a hybrid atomic memory combining gradient-echo memory (GEM) and electromagnetically induced transparency (EIT) in a cold 87Rb ensemble, with the goal of reversible time-frequency transformations. In the theory, Eqs. (2) and (4) express the GEM frequency-to-position and EIT time-to-position mappings. The authors demonstrate two experimental sequences: GEM write-in with EIT readout (frequency-to-time conversion) and EIT stop with GEM readout (time-to-frequency conversion). Heterodyne detection with coherent averaging over 200 sequences, together with Maxwell-Bloch simulations (Eqs. (5)-(6)), shows qualitative agreement for delays and fitted Gaussian widths for narrow and broad pulses. The paper claims this as a demonstration of a hybrid quantum memory and suggests applications in quantum communication and Rydberg-polariton studies. All reported measurements, however, use bright coherent pulses; no efficiency, fidelity, or photon-counting statistics are reported.","tokens_in":8521,"tokens_out":5220,"duration_ms":57067,"significance":"The dual GEM/EIT mapping concept is sound and, if demonstrated at the quantum level, would provide a useful frequency-time transducer for quantum networks. A notable strength is that the simulations are anchored to measured parameters (OD reaching 80, coupling Rabi frequency 2π×6.9 MHz, cloud super-Gaussian profile), and the data are deposited in a repository. The two-directional mapping is a useful classical demonstration. As submitted, however, the evidence does not establish the central quantum-memory claim: all data are classical heterodyne signals, no nonclassicality witness or few-photon test is provided, and the conversion is characterized only by peak positions and widths. The significance is therefore currently that of a classical time-frequency converter with a plausible path toward quantum operation, not a demonstrated quantum memory.","major_comments":[{"comment":"The central claim of a 'hybrid quantum memory' is not supported by the reported data. Every measurement is a heterodyne detection of bright coherent pulses coherently averaged over 200 sequences; no storage/retrieval efficiency, no fidelity, no single-photon- or few-photon-level test, and no nonclassicality witness (e.g., g^(2)) are reported. Equations (5)-(6) are classical Maxwell-Bloch equations and cannot certify quantum operation. The Conclusions themselves only state that the current parameters enable 'proper realization of the presented protocol,' which is a classical statement. Either quantum-level measurements must be added, or the claims should be revised to 'classical reversible time-frequency conversion based on GEM/EIT' with the quantum version framed as a perspective.","section":"Sections IV-V, Conclusions"},{"comment":"The quantitative evidence consists of fitted Gaussian delays and widths without error bars. For a memory or converter, the essential metrics are end-to-end efficiency, conversion bandwidth, delay range, signal-to-noise ratio, and mode fidelity (overlap between output and expected transformed input). None of these is reported. Without them, the qualitative delay-versus-frequency and width trends cannot be quantitatively compared with the simulations, and the device's usefulness for quantum communication cannot be assessed.","section":"Figs. 3 and 5"},{"comment":"The claimed frequency-to-time and time-to-frequency conversions are only demonstrated by peak positions in two-component experiments. A reversible Fourier relationship means that the complex amplitude distribution in one domain is mapped to the other. The paper does not show that relative amplitudes and phases across frequency/time bins are preserved: Fig. 5 shows two peaks but no quantitative extraction of their amplitudes, phases, or fidelities. Without such an analysis, the observed effect could simply be frequency-dependent delay rather than a full time-frequency conversion.","section":"Eqs. (2), (4) and Fig. 5"},{"comment":"Coherently averaging 200 sequences preserves the mean field but suppresses shot-to-shot technical noise and single-shot fluctuations. It does not demonstrate that each individual input mode is transformed coherently; a quantum memory must preserve the field operator, not only the average amplitude. The authors should report single-shot or intensity statistics, phase stability, and preferably a weak-coherent or single-photon test. This is directly load-bearing for the quantum-memory claim.","section":"Heterodyne averaging, Sec. V"}],"minor_comments":[{"comment":"Typos and labeling: 'Delyas' in Fig. 3; 'Di fferent' throughout; the axis in Fig. 5(b) labeled 'Frequency (µs)' should presumably be MHz or time in µs.","section":"Figs. 3 and 5"},{"comment":"The text says 'The sequences used in the experiment are presented in Fig. 1(a,b)', but the experimental sequences appear to be in Fig. 1(c,d). Panel references should be corrected throughout.","section":"Section IV"},{"comment":"Please define the normalization of n(z), the dimensions of OD, and the boundary/initial conditions used in the simulations. Also list the numerical values of the gradient β, the storage times T1 and T2, and the cloud length.","section":"Section III, Eq. (5)"},{"comment":"Reference [35] states data are deposited but gives no DOI or URL. Provide an identifier so the data are actually accessible.","section":"Data availability"},{"comment":"Reference [34] is mentioned as related work but never discussed; add a sentence making the relation and the distinction from the present protocol explicit.","section":"Section VI"}],"recommendation":"major_revision","confidential_remarks":"This is a borderline case. If the journal regards 'quantum memory' claims as requiring quantum-level evidence, the current manuscript should not be accepted as a demonstration of a quantum memory. However, the classical frequency-time conversion is plausible and the protocol concept is sound; a major revision that either adds weak-field/single-photon measurements and efficiency/fidelity metrics or substantially reframes the claims would be appropriate. The scope and title would need to change if only the classical demonstration is provided."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the short version: the paper shows a working classical version of something interesting—using GEM for the frequency-to-position mapping and EIT for position-to-time, and the reverse, to convert between temporal and spectral profiles of an optical pulse. The physics is sound, the simulations match the data qualitatively, and the figures tell a coherent story. What it doesn't do is demonstrate a quantum memory in any strict sense. Every reported measurement is a bright coherent pulse detected by heterodyne, coherently averaged over 200 shots. There are no error bars, no storage or retrieval efficiency, no fidelity, no nonclassicality witness. So the reversible mapping is established at the classical level only. That may be fine for a proof-of-principle of the protocol, but the title and abstract promise a quantum memory, and the conclusion even says 'proper realization of the presented protocol' in the current classical regime—so the mismatch is easy to fix with language.\n\nOn the plus side: the two-directional hybrid combination is the actual new piece, and they are transparent about the closely related work. They cite Papneja et al. [34] and their own previous transducer [16], which is good. What's missing is an explicit statement of what this adds beyond those. The data itself is plausible—narrow pulses get delayed more, broad pulses don't, and the reverse case shows frequency shifts from time-separated pulses. The simulations are standard Maxwell-Bloch and are not fitted to the delay data, which is reassuring.\n\nThe main soft spots are the quantum claim and the missing error bars. The quantum claim is load-bearing: if anyone wants to use this as a transducer for single photons, the coherence at low photon number is exactly what's untested. The linearity of the equations suggests it should work, but 'should' is not a measurement. The error bars are a smaller but real issue in an experimental paper.\n\nWho is this for? People working on quantum memories and time-frequency interfaces will want to see it. It's a reasonable combination of two known protocols with a clean demonstration. It deserves peer review, but the authors should be asked to revise the language, add error bars if the data still exists, and explicitly position their work relative to [34] and [16]. I wouldn't cite it as evidence of quantum operation, but I'd cite the hybrid protocol idea if I worked on these interfaces.","headline":"Classical hybrid GEM+EIT time-frequency converter with honest measurements; the 'quantum memory' label outstrips the data.","tokens_in":9040,"tokens_out":2341,"would_cite":true,"duration_ms":26105,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["03.67.-a","42.50.Gy","42.50.Ex"],"model":"deepseek-v4-flash","headline":"A hybrid GEM/EIT atomic memory performs reversible mapping between the time and frequency structure of light.","keywords":["quantum memory","gradient echo memory","electromagnetically induced transparency","frequency-time conversion","atomic coherence","slow light","cold rubidium ensemble","heterodyne detection"],"falsifier":"Send a heralded single-photon or weak-coherent field through the same GEM→EIT and EIT→GEM sequences and detect the retrieved field with time-resolved photon counting plus spectral filtering; if the frequency-delay relation disappears at single-excitation level, or if a nonclassicality witness on the output vanishes while the classical pulse mapping survives, the quantum-memory claim fails even though the frequency-time conversion may stand.","tokens_in":8148,"feed_emoji":"⚛️","tokens_out":8512,"duration_ms":96503,"temperature":0.7,"pith_summary":"This paper tries to establish that a single cold-atom memory can combine two standard storage protocols—gradient echo memory (GEM) and electromagnetically induced transparency (EIT)—in either order, so that atomic coherence left by one protocol is released by the other. GEM write-in followed by EIT readout turns spectral structure into arrival time, while EIT write-in followed by GEM readout turns arrival time into spectral structure. Both directions are demonstrated with bright coherent pulses read out by heterodyne detection, and the observed delays and spectral shifts match optical-Bloch simulations. If the result carries over to weak fields, the same memory becomes a bidirectional time-frequency converter that could link spectrally multiplexed and temporally multiplexed quantum channels.","feed_headline":"Atomic memory converts time to frequency and back","feed_subtitle":"One rubidium cloud runs GEM and EIT in sequence, swapping pulse timing for spectrum in both directions.","key_machinery":"The central mechanism is the Λ system in a cold 87Rb cloud—ground states |g⟩ and |h⟩, excited state |e⟩—with the atomic coherence ϱgh as the storage variable. GEM imposes a magnetic-field gradient δ(z)=βz, which maps spectral components to positions via e^{iβzT}Ã(βz); EIT creates a slow-light polariton whose temporal profile maps to position. Sequencing GEM→EIT converts frequency to time, and EIT→GEM converts time to frequency. Heterodyne detection with a local oscillator reads out the retrieved field in both time and frequency simultaneously.","core_discovery":"The central claim is that GEM and EIT, usually treated as separate storage protocols, are complementary halves of one reversible time-frequency interface. Written with GEM, a pulse's frequency components are distributed along the ensemble by the gradient-imposed phase; read back under EIT slow-light conditions, each spatial slice exits at a different delay, so frequency maps to time. Reversed, EIT stops the pulse and imprints its temporal profile along the ensemble; applying and reversing a GEM gradient then maps position to readout frequency, so time maps to frequency. Two-frequency and two-pulse inputs demonstrate both directions, and the measured delays and widths follow optical-Bloch sim","pith_inferences":["Verified at single-photon level, the two directions would let a quantum node change qubit encoding between time-bin and frequency-bin inside the memory itself, without an external interferometric transducer.","The GEM→EIT readout's delay-versus-frequency dependence suggests a tunable dispersive delay line; scanning many frequencies or using chirped pulses could map the full transfer function and bandwidth.","EIT→GEM readout of two stopped pulses effectively performs a time-to-frequency Fourier transform on their separation; testing with more pulses and varying storage times would probe linearity and multimode capacity.","The reported measurements are all classical; a clean nonclassical test would use photon pairs and check that one photon's retrieval preserves correlations with the other, upgrading the claim from classical conversion to true quantum memory operation."],"forward_implications":["A spectral-mode optical signal can be written into one atomic memory and retrieved as a time-domain signal, connecting network nodes that use different mode encodings.","The same memory can reverse the conversion, acting as a bidirectional time-frequency transducer rather than a one-way device.","The spatial position of atomic coherence along the ensemble is directly readable as either a delay or a frequency shift, giving a diagnostic for impurity- or Rydberg-modified coherence.","The demonstrated two-frequency and two-pulse cases indicate multimode operation in both directions, so the conversion should generalize to more modes within the memory bandwidth."],"supporting_citations":[{"why":"Defines the gradient echo memory protocol with its spectral-to-position mapping, used for GEM write-in and readout.","marker":"[19,20]"},{"why":"Establishes EIT-based storage and slow-light conditions used for EIT write-in and readout.","marker":"[21,22]"},{"why":"Gives the time-to-position mapping for EIT-stopped light that underlies the EIT→GEM direction.","marker":"[26]"},{"why":"Demonstrates stopping light in an atomic ensemble by switching the coupling field.","marker":"[30]"},{"why":"Supplies the dark-state polariton formalism describing slowed and stopped polaritons.","marker":"[31]"},{"why":"Numerical solver used to integrate the optical-Bloch equations that the experiments are compared against.","marker":"[32]"},{"why":"Prior camera-based spatial homodyne tomography that the proposed frequency/time readout could replace.","marker":"[33]"}],"fun_headline_variants":["Two-in-one memory swaps pulse timing for spectrum","Quantum memory flips time and frequency roles reversibly","Rubidium memory maps time to spectrum and back","GEM-EIT hybrid memory converts time to frequency and vice versa","Reversible time-frequency conversion in a single atomic memory"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The load-bearing premise is that the storage and retrieval dynamics shown with bright, coherently averaged pulses remain coherent and reversible at the single-photon level; the paper reports no single-photon test, storage efficiency, or nonclassicality witness.","fun_headline_variants_meta":{"raw":{"variants":["Two-in-one memory swaps pulse timing for spectrum","Quantum memory flips time and frequency roles reversibly","Rubidium memory maps time to spectrum and back","GEM-EIT hybrid memory converts time to frequency and vice versa","Reversible time-frequency conversion in a single atomic memory"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000369,"raw_usage":{"total_tokens":1747,"prompt_tokens":607,"completion_tokens":1140,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":351,"completion_tokens_details":{"reasoning_tokens":1063}},"tokens_in":351,"tokens_out":1140,"duration_ms":12336,"temperature":1.0,"reasoning_tokens":1063,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T11:22:59.085766+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Send a heralded single-photon or weak-coherent field through the same GEM→EIT and EIT→GEM sequences and detect the retrieved field with time-resolved photon counting plus spectral filtering; if the frequency-delay relation disappears at single-excitation level, or if a nonclassicality witness on the output vanishes while the classical pulse mapping survives, the quantum-memory claim fails even though the frequency-time conversion may stand.","supporting_citations":[{"cited_title":"Mikaeili, A","cited_arxiv_id":null,"evidence_quote":"Gives the time-to-position mapping for EIT-stopped light that underlies the EIT→GEM direction."},{"cited_title":"Günter, H","cited_arxiv_id":null,"evidence_quote":"Demonstrates stopping light in an atomic ensemble by switching the coupling field."},{"cited_title":"Fleischhauer and M","cited_arxiv_id":null,"evidence_quote":"Numerical solver used to integrate the optical-Bloch equations that the experiments are compared against."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior camera-based spatial homodyne tomography that the proposed frequency/time readout could replace."}],"review_version":1}