{"id":"132ec25f-0309-41f8-b91f-0a5e56def1c1","arxiv_id":"2508.16136","paper_version":2,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":4.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"Noisy qubit preparation and measurement can be distilled to near-zero error by repeating the noisy SPAM round with extra ancilla qubits, achieving 10^-3 with two ancillas and 10^-6 with four.","lead":"This paper proposes a protocol that repeatedly runs noisy qubit preparation and measurement steps with extra ancilla qubits, distilling the errors away. The authors claim 5% preparation and measurement errors can be pushed to roughly 0.1% with two ancillas and to 10^-6 with four.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unverified suppression claims: gate-error budget is unstated, and the supplied full text is corrupted/mismatched (header cites arXiv:2508.16135 cs.LG), so the 10^-3/10^-6 figures cannot be checked.","rationale":"The reader's UNVERDICTED verdict is appropriate. The abstract gives a specific, checkable claim, but the full text supplied is corrupted and its embedded header mismatches the paper's identifier (2508.16136 quant-ph vs. 2508.16135 cs.LG), so no proof or protocol details can be verified. Even taking the abstract at face value, the most fragile step is the implicit gate-error threshold: any purification by post-selection relies on checks that themselves have errors; without bounding those errors, the claimed suppression factors are not derivable. This is a standard technical condition for error-suppression protocols, not a dispute with consensus. The concrete test—obtaining the actual manuscript and simulating with finite gate error—would settle whether the 10^-3 and 10^-6 numbers hold. If the clean text is not available or if the gate-error analysis is absent, the paper remains unverified. I am not alleging misconduct; the likely explanation is a rendering/copy error, but that does not resolve the missing verification. The verdict should remain UNCHANGED from the reader's UNVERDICTED.","tokens_in":25917,"tokens_out":5978,"duration_ms":67829,"concrete_test":"Retrieve a clean machine-readable copy of arXiv:2508.16136; extract the protocol definition and any stated error bound. Then simulate the accepted-state error as a function of SPAM error p and entangling-gate error q for p=0.05 with 2 and 4 ancillas, sweeping q = 0, 5e-3, 5e-2. If the error at q=5e-2 is not below 1e-3 (respectively 1e-6 for 4 ancillas), the abstract's 'realistic scenario' omits a necessary gate-error condition. If the paper's own equations assume q=0, re-derive the numbers for realistic superconducting gate fidelities (q ~ 1e-2 to 1e-3) and report whether the suppression factors survive.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The quantitative claim—0.05 SPAM error reduced to about 10^-3 with two ancillas and 10^-6 with four—requires a protocol that repeats noisy preparations/measurements and post-selects. For such post-selection to work, the entangling gates used to correlate copies must have errors well below the target error, and the SPAM noise must be independent and stochastic in a known basis. The abstract states neither condition. If a verification CNOT or ancilla readout has error q comparable to 0.05, accepted events can be corrupted at rate O(q), so the 10^-3 figure cannot be reached; for 10^-6, q must be far below 10^-6 unless the protocol includes further fault-tolerant encoding. The provided full text is unreadable due to character-encoding corruption, and its embedded header—'arXiv:2508.16135v1 [cs.LG] 22 Aug 2025'—does not match this paper's arXiv identifier and category (2508.16136, quant-ph). This unusual inserted passage suggests the body text may belong to a different document, so no equation, circuit diagram, or proof can be inspected. Thus we cannot confirm whether the error analysis includes ancilla gate noise, whether ideal gates are assumed, or whether the claimed numbers carry caveats. This is the most load-bearing unresolved condition for the central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims a protocol for purifying noisy qubit state preparation and measurement (SPAM) by repeating noisy SPAM operations and post-selecting on auxiliary qubits. The abstract gives concrete quantitative claims: with per-round preparation and measurement error around 0.05, the protocol suppresses the error to about 10^-3 with two ancillas and 10^-6 with four ancillas, and the method is claimed to be feasible with superconducting qubits and applicable to variational quantum algorithms, quantum error correction, and quantum repeaters. The supplied full text, however, is largely unreadable character-encoding corruption and contains an embedded header 'arXiv:2508.16135v1 [cs.LG] 22 Aug 2025', which does not match the declared arXiv identifier 2508.16136 or its quant-ph category. As a result, no circuit, equation, derivation, or numerical analysis can be inspected, and the central quantitative claims are unsupported in the submitted text.","tokens_in":26068,"tokens_out":6658,"duration_ms":75572,"significance":"If the claimed protocol is correct and its assumptions are made explicit, the result would be significant: SPAM errors are a recognized bottleneck in near-term quantum computing and communication. The idea of distilling near-ideal SPAM from repeated noisy SPAM is concrete, falsifiable, and potentially impactful. The abstract provides specific target numbers (0.05 -> 10^-3 with two ancillas, 10^-6 with four), which are the kind of quantitative predictions that can be checked. However, because the submission as provided contains no readable derivation, noise model, or gate-error budget, I cannot currently verify those predictions. The claimed benefits are therefore conditional on details that are absent.","major_comments":[{"comment":"The supplied full text is not readable: it is a character-encoding corruption, and the embedded header reads 'arXiv:2508.16135v1 [cs.LG] 22 Aug 2025', which does not match this paper's arXiv identifier (2508.16136) or quant-ph category. No equation, circuit, or proof can be inspected. Because the abstract's quantitative claims (0.05 to 10^-3 with two ancillas and 10^-6 with four ancillas) depend on an explicit protocol and derivation, the central claim is unsupported in the submitted text. The authors must supply a legible and correctly matched full text before a technical review is possible.","section":"Full text (header)"},{"comment":"The abstract does not state the noise model. The claimed suppression requires assumptions: the SPAM noise is stochastic and in a known basis (e.g., bit-flip or depolarizing with known rate), copies are independent, and post-selection is exact. If preparation or measurement errors are coherent rotations or have unknown basis, repeated-copy post-selection need not purify them. The authors should specify the noise model and provide a derivation of the output error probability p_out(n,p) as a function of the SPAM error rate p and the number of ancillas n, including a check that p_out(2,0.05) is approximately 10^-3.","section":"Abstract"},{"comment":"The protocol requires entangling gates and ancilla readout. If each verification CNOT has error q, accepted events can be corrupted at rate O(q); for the four-ancilla claim of 10^-6, q must be far below 10^-6 unless additional encoding is used. The text does not state any gate-error budget or ancilla-readout error requirement. The authors should include gate noise q in the error analysis and give the threshold condition q < f(p,n) required to reach the claimed suppression levels.","section":"Abstract / assumed gate budget"},{"comment":"No success probability or repetition overhead is given. Distilling by post-selection typically has acceptance probability p_success(n,p) that decreases with the number of ancillas, and the practical overhead is 1/p_success. Without this quantity, the claim that the protocol is 'feasible with superconducting qubits' cannot be evaluated. Please provide the acceptance probability and compare the effective per-logical-operation cost with direct noisy SPAM.","section":"Full text / success probability"}],"minor_comments":[{"comment":"The phrase 'suppress error rates up to 10^-3' should read 'suppress error rates to 10^-3' or 'down to 10^-3'.","section":"Abstract"},{"comment":"SPAM should be expanded at first use, and 'purify noisy SPAM' should specify that both the preparation channel and the measurement channel are being distilled.","section":"Title / abstract"},{"comment":"The embedded arXiv header mismatch should be corrected; the supplied body appears to belong to a different document.","section":"Full text"},{"comment":"Please add references to existing SPAM characterization and mitigation techniques (e.g., randomized benchmarking, readout calibration, measurement error mitigation) to contextualize the novelty.","section":"Introduction / related work"}],"recommendation":"major_revision","confidential_remarks":"The submitted file appears to be mis-uploaded or corrupted: the body is unreadable and its embedded arXiv header (2508.16135, cs.LG) does not match the declared paper (2508.16136, quant-ph). I recommend that the editor ask the authors to provide the correct, legible manuscript before any technical evaluation. If the supplied text is in fact the intended submission, then the paper is not reviewable in its current form and should be rejected administratively; if a correct version exists, the substantive concerns in the major comments should be addressed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the abstract promises a concrete SPAM-purification protocol with specific numbers—0.05 to 10^-3 with two ancillas, 10^-6 with four—and that is exactly the kind of result people in near-term quantum computing would care about. The problem is I cannot check any of it. The supplied full text is unreadable character garbage, and the one clean line in it is an arXiv header for a different paper (2508.16135, cs.LG), not this one. So this review is abstract-only.\n\nWhat the paper does well: it identifies a real pain point, gives a concrete protocol mechanism (repeat noisy SPAM, post-select to distill cleaner SPAM), and states quantitative targets rather than vague hopes. The applications listed (variational algorithms, QEC, repeaters) are sensible. No circularity or fitting visible in the abstract; these look like derived claims under an unstated noise model.\n\nThe soft spots are proportionate to what I can see. First, the abstract gives no noise model. Any post-selection scheme has a threshold: if the entangling gates used to correlate copies have error q comparable to the 0.05 SPAM error, the best you can post-select to is O(q). To reach 10^-6 you need gate errors far below 10^-6 unless the protocol includes additional encoding. The abstract does not state this budget, and it is load-bearing. Second, no success probability is given. If the acceptance rate is exponential in the number of ancillas, the protocol becomes impractical at four ancillas; if it is polynomial, that is worth saying. Third, the claim of suppressing to \"arbitrarily low level\" is too strong as stated; with finite ancillas and non-zero gate noise there is a floor. Presumably the full text says this, but the abstract alone invites a skeptical reader.\n\nNone of these are necessarily fatal. The ideas are not new at the principle level—multi-copy purification is a known family—but applying it to SPAM with explicit ancilla counts is a useful contribution if the derivation holds. I have not been able to verify the derivation, the circuit, the references, or the comparison with prior work because the body text is not usable.\n\nBottom line: if the actual arXiv version is intact, this is a candidate for peer review. As presented, I cannot recommend citing it or trusting the numbers. The editor should first confirm the submission is not corrupted. Then a referee can check the gate-error budget, the success probability, and the comparison with randomized benchmarking or other SPAM-mitigation work.","headline":"Interesting SPAM-purification claim with concrete numbers, but the supplied full text is unreadable and its header cites a different arXiv paper, so I couldn't verify anything beyond the abstract.","tokens_in":26722,"tokens_out":2934,"would_cite":false,"duration_ms":32237,"reading_group":"maybe","serious_thinker":"unclear","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Repeated noisy SPAMs can be post-selected into near-perfect qubit preparation and measurement, suppressing realistic 5% errors to 10^-3 with two ancillas and 10^-6 with four.","keywords":["SPAM purification","quantum state preparation","quantum measurement noise","post-selection","ancilla-assisted error mitigation","quantum error mitigation","superconducting qubits","entanglement distribution"],"falsifier":"Run the protocol on a superconducting qubit whose preparation and measurement errors are independently measured near 5%, then measure the SPAM error of the post-selected output with two and four ancillas. If the residual error does not drop to about 10^-3 and 10^-6, or if the post-selection success probability is too low to be usable, the central claim is falsified. A second check: inject a known coherent preparation error; if the protocol cannot suppress it, that confirms the protocol's stochastic-noise scope.","tokens_in":25664,"feed_emoji":"⚛️","tokens_out":6136,"duration_ms":64781,"temperature":0.7,"pith_summary":"The paper claims that state preparation and measurement (SPAM) noise, a major error source in quantum information processing, can be purified away by repeating noisy SPAM operations and post-selecting on the outcomes. In a realistic scenario with 0.05 error in both preparation and measurement, it reports suppression to about 10^-3 with two ancilla qubits and 10^-6 with four. The idea is to distill error-free SPAM from noisy SPAM rather than to correct each error as it happens, so the protocol needs no pre-existing clean SPAM source. If the claim holds, SPAM errors become a negligible budget item for variational quantum algorithms, quantum error correction, and entanglement distribution through quantum repeaters. The paper also argues the protocol is feasible with superconducting qubits.","feed_headline":"Four ancillas cut qubit prep-and-measurement error from 5% to 10^-6","feed_subtitle":"Repeated noisy rounds are post-selected to distill near-ideal preparation and measurement, all but removing SPAM from the error budget.","key_machinery":"The mechanism is a distillation circuit built from repeated noisy SPAM rounds plus ancilla qubits: the noisy SPAM is run multiple times, the ancillas are coupled to those runs, and a round is kept only when the ancilla outcomes indicate an error-free run. The load-bearing step is post-selection, which converts a stochastic per-round preparation or measurement error into a much lower probability that an error survives all checks. The number of ancillas sets the order of purification: two ancillas give the stated 10^-3 suppression from 0.05 noise, and four give 10^-6.","core_discovery":"The paper's central claim is that noisy state preparation and measurement can be purified to an arbitrary level by running many noisy SPAMs and using ancilla qubits to flag or reject the noisy runs. With preparation and measurement error rates around 5%, two ancillas are said to reduce the residual error to roughly 10^-3 and four ancillas to roughly 10^-6; adding more ancillas pushes it lower. The protocol is presented as a way to distill error-free SPAM out of noisy SPAM, which would let experiments keep the convenience of imperfect preparation and measurement while effectively eliminating their contribution to the error budget.","pith_inferences":["I infer that the main hidden cost is post-selection overhead: suppressing error from 5% to 10^-6 must reject most rounds, so the protocol is most useful when SPAM errors are the dominant bottleneck and total shot count is not the limiting resource.","A natural extension would be to apply the same repeated-copy post-selection idea to detector calibration or measurement tomography, where one noisy device is compared with itself.","The protocol's scope likely assumes stochastic SPAM noise with a known structure; if preparation errors are dominated by coherent rotations, additional randomization or twirling would probably be needed for the post-selection checks to isolate errors.","On platforms where two-qubit gate error is comparable to the 5% SPAM error, the ancilla checks themselves would inject errors, so the claimed suppression hinges on the gates being substantially cleaner than the SPAM being removed."],"forward_implications":["Variational quantum algorithms can stop treating SPAM as a dominant error source, since imperfect initialization and readout can be purified into negligible overhead.","Quantum error correction can assume near-ideal state injection and measurement, simplifying the error budget for fault-tolerant operation.","Entanglement distribution through quantum repeaters becomes viable with imperfect local SPAM, because purified SPAM can be produced on demand via post-selection.","Error suppression scales with ancilla count, offering a tunable trade between extra qubits and the achievable SPAM error rate.","Because the protocol needs no clean SPAM source, it can be implemented with the existing noisy operations plus relatively clean entangling gates for the ancillas."],"supporting_citations":[],"fun_headline_variants":["Four ancillas distill 10^-6 SPAM error from 5% noise","Repeat noisy SPAM rounds to purify qubit prep and measurement","Four ancillas shrink prep-and-measurement error 50,000-fold","Distill clean qubit SPAM from repeated noisy rounds with ancillas","Purify noisy qubit SPAM arbitrarily using ancilla-assisted distillation"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The load-bearing premise is that SPAM noise is stochastic and structured enough that repeated noisy rounds plus ancilla post-selection can separate clean runs from errored runs, and that the entangling gates used in the checks are much less noisy than the 5% preparation and measurement error being removed.","fun_headline_variants_meta":{"raw":{"variants":["Four ancillas distill 10^-6 SPAM error from 5% noise","Repeat noisy SPAM rounds to purify qubit prep and measurement","Four ancillas shrink prep-and-measurement error 50,000-fold","Distill clean qubit SPAM from repeated noisy rounds with ancillas","Purify noisy qubit SPAM arbitrarily using ancilla-assisted distillation"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00045,"raw_usage":{"total_tokens":2067,"prompt_tokens":666,"completion_tokens":1401,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":410,"completion_tokens_details":{"reasoning_tokens":1315}},"tokens_in":410,"tokens_out":1401,"duration_ms":14653,"temperature":1.0,"reasoning_tokens":1315,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T17:30:30.006944+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the protocol on a superconducting qubit whose preparation and measurement errors are independently measured near 5%, then measure the SPAM error of the post-selected output with two and four ancillas. If the residual error does not drop to about 10^-3 and 10^-6, or if the post-selection success probability is too low to be usable, the central claim is falsified. A second check: inject a known coherent preparation error; if the protocol cannot suppress it, that confirms the protocol's stochastic-noise scope.","supporting_citations":[],"review_version":1}