{"id":"4e566944-08d7-4699-a39f-0617e47bae05","arxiv_id":"2411.12399","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The authors develop a noncommutative random restriction technique to prove quantum versions of the Eldan-Gross, Talagrand isoperimetric, and KKL-type inequalities for the matrix algebra M_{2^n}.","lead":"This paper adapts the random restriction method to the quantum hypercube and derives several quantum analogues of classical Boolean analysis inequalities, including Eldan-Gross and Talagrand-type bounds. The results provide new tools for the study of quantum Boolean functions and partial progress toward the open quantum KKL conjecture.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The invalid negative-term relaxation at (4.5) is a real gap in the proof of Theorem 1.8, but it is repairable using W(H)<=W(T) together with the D_>= condition.","rationale":"The reader's weakest assumption correctly locates the invalid inequality at (4.5), and my pass confirms the sign-direction error: the negative term is relaxed as though sqrt(W(H)) were at its lower endpoint, while only an upper bound is available. I also notice that the final inequality in (4.5) uses the D_>= condition in a way that should be made explicit. This is a genuine proof gap for Theorem 1.8 and consequently for Theorem 1.9 as derived from it. The gap is nevertheless repairable: replacing the offending relaxation by W(H)<=W(T), bounding sqrt(Inf_p(H)) by 2^{p/2} sqrt(Inf_p(T)), and using the defining property of D_>= leaves enough room for a contradiction after Lemma 4.3, possibly after enlarging one universal constant. The other main results, especially the quantum Eldan-Gross inequality and the Talagrand-type isoperimetric inequality, appear to be supported by independent semigroup and random-restriction arguments, and the p=1 case of Theorem 1.8 has independent support from Blecher et al. Therefore the central claims are likely correct, but the submitted proof of Theorem 1.8 is not valid as written. This is exactly the kind of conditional accept that the reader recommended: fix or clarify (4.5), or explicitly downgrade the affected statements.","tokens_in":32510,"tokens_out":19399,"duration_ms":192717,"concrete_test":"Re-derive (4.5) using the valid bounds W(H)<=W(T) and sqrt(Inf_p(H)) <= 2^{p/2} sqrt(Inf_p(T)) instead of the unjustified substitution, and check whether the resulting lower bound, after summing over D_>= via Lemma 4.3, exceeds 2 Inf_p(T). If the constant 1040K_p in assumption (4.4) is too small, increase it, e.g. to 2048K_p, and verify that the log-lower-bound step in (4.5) still holds. Also test the edge case d=1 with T=(1+sigma_1)/2, where W(H)=W(T)/4, to confirm that the original displayed inequality (4.5) fails verbatim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The weakest load-bearing step is (4.5). Corollary 3.10 applied to H_d gives Inf(H)+var(H) >= (1/16) log(1/max ||d_j(H)||^p) W(H) - (K/16) sqrt(Inf_p(H)) sqrt(W(H)). The paper replaces the negative term by -(K/64) sqrt(Inf_p(H)) sqrt(W(T)). This requires sqrt(W(H)) <= sqrt(W(T))/4, i.e. W(H) <= W(T)/16. But Lemma 4.2(iii) only guarantees W(T)/16 <= W(H) <= W(T); for d=1, taking T=(1+sigma_1)/2 gives W(H)=W(T)/4 exactly. Thus the displayed inequality is false in general: the negative term has been relaxed as though W(H) sat at its lower endpoint. This is not fatal. Using the valid upper bound W(H)<=W(T) for the negative term gives -(K/16) sqrt(Inf_p(H)) sqrt(W(T)). Since sqrt(Inf_p(H)) <= 2^{p/2} sqrt(Inf_p(T)) and, for d in D_>=, sqrt(Inf_p(T) W(T)) <= 4 Inf_p(T)/var(T) W(T), the extra loss is at most 2^{p/2} K/4 Inf_p(T)/var(T) W(T). With 1040K_p/256 = 4.0625K_p, the remaining coefficient is about 3.56K_p, and after applying Lemma 4.3 the bound still exceeds 2 Inf_p(T). A modest enlargement of the constant 1040 repairs the proof. As written, however, the proof of Theorem 1.8 is incomplete and must be revised.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a noncommutative random restriction method on the matrix algebra M_{2^n} and applies it to prove several quantum analogues of Boolean analysis inequalities. The main results are a dimension-free quantum KKL inequality for L^p-influences with 1 ≤ p < 2 (Theorem 1.8), a related quantum KKL-type inequality (Theorem 1.9), a quantum Talagrand isoperimetric inequality (Theorem 1.10), and the quantum Eldan-Gross inequality (Theorem 1.11). The paper also derives applications to CAR-algebra KKL-type results and a stability result for the L^1-influence KKL theorem. The proofs of Theorems 1.10 and 1.11 are based on the quantum Ornstein-Uhlenbeck semigroup, a Buser-type inequality, a local reverse Poincaré inequality, and Fourier-spectrum estimates, while the proof of Theorem 1.8 uses a dyadic decomposition of the Fourier spectrum combined with Lemmas 4.2 and 4.3.","tokens_in":32789,"tokens_out":8260,"duration_ms":75435,"significance":"If the main results are correct, the paper provides a unified random-restriction framework for quantum Boolean analysis, recovers recent results of Rouzé-Wirth-Zhang and Jiao-Luo-Zhou, and gives a new proof of the quantum Eldan-Gross inequality. The paper is generally careful with constants and includes a useful counterexample (Remark 4.4) showing that the p=2 analogue of Theorem 1.8 fails even in the commutative hypercube. However, the proof of Theorem 1.8 contains an invalid inequality at (4.5), and Theorem 1.9 is derived from Theorem 1.8; therefore the central KKL-type claims need repair before the results can be accepted.","major_comments":[{"comment":"The step in the proof of Theorem 1.8 that replaces sqrt(W_{≈d}(H_d(T))) by sqrt(W_{≈d}(T)) in the negative term is unjustified. Corollary 3.10 applied to H_d(T) gives a negative contribution proportional to -Kp/16 sqrt(Inf_p(H_d(T))) sqrt(W_{≈d}(H_d(T))); the paper instead writes -Kp/64 sqrt(Inf_p(H_d(T))) sqrt(W_{≈d}(T)). This replacement is valid only if sqrt(W_{≈d}(H_d(T))) ≤ sqrt(W_{≈d}(T))/4, i.e. W_{≈d}(H_d(T)) ≤ W_{≈d}(T)/16. Lemma 4.2(iii) gives W_{≈d}(H_d(T)) ≥ W_{≈d}(T)/16, and Lemma 4.2(iv) gives W_{≈d}(H_d(T)) ≤ W_{≈d}(T); both inequalities are compatible with values strictly larger than W_{≈d}(T)/16. For example, for d=1 and T=(1+σ_1)/2, one has W_{≈1}(H_1(T)) = W_{≈1}(T)/4, so the required endpoint behavior fails. Since the negative term is being relaxed in the wrong direction, the subsequent lower bound '≥ 4Kp Inf_p(T)/var(T) W_{≈d}(T)' does not follow. Because Theorem 1.9 is proved by invoking Theorem 1.8, the proof of Theorem 1.8 must be revised; the gap appears repairable by keeping W_{≈d}(H_d(T)) in the negative term and using the valid upper bound W_{≈d}(H_d(T)) ≤ W_{≈d}(T), at the cost of adjusting the constants.","section":"Section 4, Eq. (4.5)"}],"minor_comments":[{"comment":"The name 'Motanaro' should be 'Montanaro'.","section":"Abstract"},{"comment":"Items (i) and (iii) of Proposition 2.5 are stated without proof, and the proof of (ii) relies on (i). Since these curvature identities are used in the proof of Theorem 5.5, it would be helpful to include the short Fourier-expansion verification.","section":"Section 2, Proposition 2.5"},{"comment":"Theorem 4.5 is stated as a theorem but its proof is omitted and deferred to reference [2]. The paper also notes that (4.6) follows from Theorem 1.5, so the result should be labeled as a consequence or remark rather than a new theorem.","section":"Section 4, Theorem 4.5"},{"comment":"In the display after (5.12), the denominator appears as 'sqrt(log(1 + 1/log(M(T))))'; this should be 'sqrt(log(1 + 1/M(T)))' to match the Eldan-Gross inequality and the subsequent use of the bound 1 + log(1/M(T)) ≤ 2 log(1/M(T)).","section":"Section 5.3, proof of Theorem 1.11"},{"comment":"The proof of Proposition 5.2 is lengthy but well organized; however, a short sentence in the proof after (5.34) says 't0 ≥ (4e)^{d/2}' while Lemma 5.23 only requires t0 > (2e)^{d/2}; the stronger bound follows from (5.34) and M_J(T) ≤ M(T) ≤ e^{-2d}, but this implication could be made explicit.","section":"Section 5.4, Proposition 5.2"}],"recommendation":"major_revision","confidential_remarks":"The paper's main novelty claim should be carefully positioned against the concurrent work of Blecher, Gao and Xu [2], which the authors already acknowledge. In particular, Theorem 4.5 is not proved in this manuscript and is attributed to [2] in substance; the editorial decision should ensure that the paper's contribution is clearly delimited to the results proved here, namely Theorems 1.8, 1.10, and 1.11 and their applications. The gap in (4.5) is the main technical obstacle, and it appears fixable; the revision should also verify that the constants in the repaired proof still yield the stated exponential form in Theorem 1.8."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new content here is the quantum random restriction method and the two inequalities it proves cleanly: the quantum Talagrand-type isoperimetric inequality (Theorem 1.10) and the quantum Eldan-Gross inequality (Theorem 1.11). The proofs of those two are detailed, use standard noncommutative semigroup and Fourier tools, and appear internally consistent. The paper is also honest about overlap: Theorem 1.9 recovers Rouzé--Wirth--Zhang, and the p=1 case of Theorem 1.8 was independently proved by Blecher--Gao--Xu. That is not a defect; the method is the contribution, and the recovered results are useful sanity checks.\n\nThe soft spot is real. In the proof of Theorem 1.8, equation (4.5) takes Corollary 3.10 applied to H_d, which contains a negative term proportional to sqrt(W(H_d)), and replaces that negative term by a quarter-size version involving sqrt(W(T)). That replacement requires sqrt(W(H_d)) <= sqrt(W(T))/4, i.e. W(H_d) <= W(T)/16. But Lemma 4.2(iii) only guarantees W(T)/16 <= W(H_d) <= W(T), and the lower endpoint is attainable for d=1 with T=(1+sigma_1)/2, where W(H)=W(T)/4. So the displayed inequality is false as written. The stress-test note is right that this is not fatal: using the valid upper bound W(H_d) <= W(T) on the negative term, together with Inf_p(H_d) <= 2^p Inf_p(T) and the D_>= condition, enlarges the constant rather than breaking the argument. But as written, the proof of Theorem 1.8 is incomplete and must be revised.\n\nThe rest of the paper looks sound. The Fourier-spectrum estimates in Section 5, the Buser-type inequality, the local reverse Poincaré inequality, and the derivation of the Eldan--Gross inequality from Theorem 1.10 are all coherent. The citation pattern is appropriate; the self-citation to the CAR algebra paper is for comparison, not assumption. The counterexample in Remark 4.4 correctly shows why p=2 fails even classically.\n\nWho should read this: anyone working on quantum Boolean analysis, noncommutative influences, or functional inequalities on the quantum hypercube. The random restriction method and the Eldan--Gross inequality will be useful even if Theorem 1.8 needs a corrected proof. The paper deserves a serious referee; the referee should focus on (4.5) and ask for a repaired proof or an explicit statement that Theorem 1.8 is conditional pending that fix.","headline":"A solid quantum random-restriction toolbox with two strong new inequalities, but the proof of the headline dimension-free KKL theorem has a real gap at (4.5) that is repairable in principle.","tokens_in":33430,"tokens_out":1520,"would_cite":true,"duration_ms":16937,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46L53","94D10","47D07"],"pacs":[],"model":"deepseek-v4-flash","headline":"Developing the random restriction method in the quantum setting, this paper proves a quantum Eldan-Gross inequality and related KKL-type inequalities.","keywords":["quantum KKL inequality","quantum Eldan-Gross inequality","quantum Talagrand isoperimetric inequality","random restriction","quantum Boolean functions","heat semigroup","Fourier spectrum","noncommutative Lp spaces"],"falsifier":"Compute the two sides of equation (4.5) for a concrete Fourier polynomial $T$ and a specific dyadic $d$ (for example, $T = \\frac12 + \\frac{1}{2\\sqrt{n}} \\sum_j \\sigma_j$) and check whether the relaxed inequality $A W_{\\approx d}(H_d) - B \\sqrt{\\mathrm{Inf}_p(H_d)} \\sqrt{W_{\\approx d}(H_d)} \\ge A W_{\\approx d}(T)/16 - (B/4) \\sqrt{\\mathrm{Inf}_p(H_d)} \\sqrt{W_{\\approx d}(T)}$ holds. A single violation would invalidate the proof of Theorem 1.8; if no violation is found, the relaxation may be salvageable by a sharper estimate.","tokens_in":32219,"feed_emoji":"⚛️","tokens_out":15391,"duration_ms":129227,"temperature":0.7,"pith_summary":"This paper brings the random restriction technique—a combinatorial method for bounding the influence of Boolean functions on the hypercube—into the quantum setting, where the hypercube is replaced by the algebra $M_{2^n}$ with normalized trace. The authors use it to prove a quantum Eldan-Gross inequality for projections: $\\mathrm{var}(T)/\\sqrt{\\log(1+1/\\sum_j \\|d_j(T)\\|_1^2)} \\le K \\|(\\sum_j |d_j(T)|^2)^{1/2}\\|_1$ with a universal constant $K$, together with a dimension-free quantum KKL inequality for $0\\le T\\le 1$: $\\max_j \\|d_j(T)\\|_p^p \\ge \\frac14 \\exp[ - \\frac{K}{2-p} \\frac{\\sum_j \\|d_j(T)\\|_p^p}{\\mathrm{var}(T)}]$ for $1\\le p<2$. These are quantitative influence bounds for quantum Boolean functions, the kind of control needed for the open quantum KKL conjecture. The same machinery yields a quantum Talagrand-type isoperimetric inequality and recovers recent alternative answers to the quantum KKL conjecture based on semigroup and CAR-algebra methods, marking a unification of several quantum influence inequalities under one approach.","feed_headline":"Random restrictions prove quantum KKL and Eldan-Gross bounds","feed_subtitle":"New influence bounds for quantum Boolean functions, and a fresh route to the quantum KKL conjecture.","key_machinery":"The central object is the noncommutative random restriction operator, defined for a subset $J \\subseteq [n]$ by $R^J_j(T) = E_{M_{J^c \\cup \\{j\\}}}(d_j(T))$ if $j \\in J$ and $R^J_j(T) = 0$ otherwise, where $E$ is the conditional expectation onto the subalgebra supported on $J^c \\cup \\{j\\}$ and $d_j(T)$ is the $j$-th quantum partial derivative. When $J$ is chosen randomly with selection probability $1/d$, this operator isolates Fourier weight in the band $d \\le |\\mathrm{supp}(s)| < 2d$, as expressed through the spectral weight $W_{\\approx d}(T) = \\sum_{d \\le |\\mathrm{supp}(s)| < 2d} \\hat{T}(s)^2$. The proof also uses the difference of two noise operators, $H_d(T) = (1 - 1/(2d))^L(T) - (1 - 1/d)^L(T)$, where $\\delta^L(T)$ is the noise operator that shrinks each Fourier coefficient of degree $k$ by $\\delta^k$, to extract the same frequency band, and combines the resulting estimates with hypercontractivity and log-Sobolev inequalities for the heat semigroup. This machinery converts the task of bounding influence into a sequence of variance-influence inequalities, yielding Theorems 1.8 and 1.11.","core_discovery":"The paper's central claim is that the random restriction method, previously used for a unified proof of the classical KKL, Talagrand, and Eldan-Gross inequalities, admits a noncommutative version on the quantum hypercube $M_{2^n}$. In that setting, the authors prove the quantum Eldan-Gross inequality (Theorem 1.11): for every projection $T \\in M_{2^n}$ there is a universal constant $K$ such that $\\mathrm{var}(T) / \\sqrt{\\log(1 + 1/\\sum_j \\|d_j(T)\\|_1^2)} \\le K \\|(\\sum_j |d_j(T)|^2)^{1/2}\\|_1$, and the dimension-free quantum KKL inequality (Theorem 1.8): for $1 \\le p < 2$ and $0 \\le T \\le 1$, $\\max_j \\|d_j(T)\\|_p^p \\ge \\frac14 \\exp[ - \\frac{K}{2-p} \\frac{\\sum_j \\|d_j(T)\\|_p^p}{\\mathrm{var}(T)}]$. These follow from a sequence of Fourier-spectrum estimates for the restriction operator, combined with hypercontractivity and log-Sobolev inequalities for the heat semigroup. If correct, the inequalities give quantitative control of the influence of quantum Boolean functions by their variance and provide an alternative partial answer to the quantum KKL conjecture.","pith_inferences":["If the proof of Theorem 1.8 can be completed (the step at (4.5) currently rests on an unproven relaxation), the theorem would provide a dimension-free quantum KKL inequality that is uniform in $n$; if the step fails, the theorem may still be true but needs a different argument.","The random restriction machinery likely extends to higher-order analogues, such as the quantum Talagrand influence inequality with constant $(2-p)^{-1}$ rather than $(2-p)^{-2}$, as suggested by the independent quantum results cited in the paper.","The Fourier-spectrum estimate of Theorem 5.1 is a noncommutative version of a classical lemma and may be applicable to quantum noise sensitivity and Fourier tail bounds beyond the present inequalities.","A testable extension is to check whether the quantum Eldan-Gross inequality holds for all self-adjoint contractions, not just projections, and whether the universal constant $K$ can be made explicit."],"forward_implications":["For every projection on the quantum hypercube, the quantum Eldan-Gross inequality gives a universal trade-off between variance, the sum of squared $L^1$-influences, and the $L^1$ norm of the gradient vector, extending the classical Eldan-Gross inequality to noncommutative Boolean analysis.","The dimension-free quantum KKL inequality (Theorem 1.8) bounds the maximum $L^p$-influence of any bounded element from below by an exponential decay in the total $L^p$-influence divided by variance; for $p < 2$ this is strong enough to imply the $L^p$-influence KKL bound of Theorem 1.9, $\\max_j \\|d_j(T)\\|_p^p \\ge C (2-p) \\mathrm{var}(T) \\log(n)/n$.","Combined with Fourier-spectrum estimates, the Eldan-Gross inequality yields a quantum Talagrand-type isoperimetric inequality and the balanced-projection bound $\\max_j \\|d_j(T)\\|_1 \\ge C \\sqrt{\\log n}/n$, as well as recovering the CAR-algebra results that were previously obtained by different methods.","The method unifies several recent quantum influence inequalities under one technique, indicating that random restriction is as effective in the quantum hypercube as it is classically."],"supporting_citations":[{"why":"supplies the random restriction scheme for the classical KKL and Eldan-Gross proofs that the quantum method extends.","marker":"[14]"},{"why":"proves the quantum Talagrand-type inequality and the $L^p$-influence KKL bound that Theorem 1.9 recovers.","marker":"[22]"},{"why":"establishes the CAR-algebra Eldan-Gross and KKL inequalities recovered by the present method.","marker":"[11]"},{"why":"defines quantum Boolean functions and poses the quantum KKL conjecture that motivates the results.","marker":"[18]"},{"why":"supplies the Fourier-spectrum lemma that Proposition 5.2 adapts to the noncommutative setting.","marker":"[13]"},{"why":"provides the simplified isoperimetric and deviation estimates used in Theorem 5.3 and Corollary 5.16.","marker":"[8]"},{"why":"states the classical Eldan-Gross inequality whose quantum analogue is Theorem 1.11.","marker":"[6]"},{"why":"states the classical KKL inequality whose quantum counterparts are the paper's main targets.","marker":"[12]"}],"fun_headline_variants":["Quantum random restriction method proves KKL-type bounds","Quantum Eldan-Gross and KKL inequalities via restrictions","Random restriction method extends to quantum Boolean functions","New quantum KKL inequalities from random restriction method","Noncommutative proof of quantum KKL and Talagrand inequalities"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument's load-bearing premise is that in equation (4.5) the negative correction term can be taken as small as one quarter of its maximum possible value, even though the proved bounds only constrain the relevant spectral weight within a factor of four.","fun_headline_variants_meta":{"raw":{"variants":["Quantum random restriction method proves KKL-type bounds","Quantum Eldan-Gross and KKL inequalities via restrictions","Random restriction method extends to quantum Boolean functions","New quantum KKL inequalities from random restriction method","Noncommutative proof of quantum KKL and Talagrand inequalities"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000193,"raw_usage":{"total_tokens":1339,"prompt_tokens":922,"completion_tokens":417,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":538,"completion_tokens_details":{"reasoning_tokens":340}},"tokens_in":538,"tokens_out":417,"duration_ms":5001,"temperature":1.0,"reasoning_tokens":340,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T17:36:59.589855+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the two sides of equation (4.5) for a concrete Fourier polynomial $T$ and a specific dyadic $d$ (for example, $T = \\frac12 + \\frac{1}{2\\sqrt{n}} \\sum_j \\sigma_j$) and check whether the relaxed inequality $A W_{\\approx d}(H_d) - B \\sqrt{\\mathrm{Inf}_p(H_d)} \\sqrt{W_{\\approx d}(H_d)} \\ge A W_{\\approx d}(T)/16 - (B/4) \\sqrt{\\mathrm{Inf}_p(H_d)} \\sqrt{W_{\\approx d}(T)}$ holds. A single violation would invalidate the proof of Theorem 1.8; if no violation is found, the relaxation may be salvageable by a sharper estimate.","supporting_citations":[{"cited_title":"Kelman, S","cited_arxiv_id":null,"evidence_quote":"supplies the random restriction scheme for the classical KKL and Eldan-Gross proofs that the quantum method extends."},{"cited_title":"Rouz´ e, M","cited_arxiv_id":null,"evidence_quote":"proves the quantum Talagrand-type inequality and the $L^p$-influence KKL bound that Theorem 1.9 recovers."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"establishes the CAR-algebra Eldan-Gross and KKL inequalities recovered by the present method."},{"cited_title":"Montanaro and T","cited_arxiv_id":null,"evidence_quote":"defines quantum Boolean functions and poses the quantum KKL conjecture that motivates the results."},{"cited_title":"Keller and G","cited_arxiv_id":null,"evidence_quote":"supplies the Fourier-spectrum lemma that Proposition 5.2 adapts to the noncommutative setting."},{"cited_title":"Eldan and R","cited_arxiv_id":null,"evidence_quote":"states the classical Eldan-Gross inequality whose quantum analogue is Theorem 1.11."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"states the classical KKL inequality whose quantum counterparts are the paper's main targets."}],"review_version":1}