{"id":"13445f50-8bdc-4d25-9b96-33353cff6bba","arxiv_id":"2507.11204","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"low","formal_verification":"none","parameter_count":2,"one_line_summary":"A deterministic construction using the computational basis, the Fourier basis, and d-1 phase-shifted Fourier bases is shown to determine any d-dimensional density matrix, with a d=6 photonic demonstration.","lead":"The paper presents a quantum state tomography scheme that reconstructs any d-dimensional quantum state from only d+1 projective measurement bases, equal to the information-theoretic minimum. It demonstrates the scheme on a silicon photonic chip for six-dimensional states, a case where complete mutually unbiased bases do not exist.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The sufficiency proof leaves the existence of phases satisfying Eq. (S14) unproved for arbitrary d; T invertibility, and hence the whole d+1 construction, rests on that condition.","rationale":"The reader's weakest assumption is the same as the one identified here: the proof assumes suitable phases theta^(1)_m satisfying Eq. (11) can be chosen in every dimension, without providing an explicit construction or existence proof. I agree with that assessment. The underlying Vandermonde reduction is valid conditional on the phase condition, and the condition is easy to satisfy with an irrational quadratic phase, but the manuscript leaves this implicit, so the theorem is not fully self-contained. The experimental concern about missing calibration and raw data is genuine but does not affect the mathematical core. Since the reader's CONDITIONAL verdict already captures exactly these gaps, my stress-test does not change the verdict. If the authors add one explicit existence sentence (e.g., theta_m = alpha m^2 with irrational alpha/pi) and release data or a reproducible experimental script, the conditional status can be resolved.","tokens_in":9104,"tokens_out":22778,"duration_ms":295389,"concrete_test":"Set theta^(1)_m = alpha m^2 with alpha/(2pi) irrational, e.g., alpha = sqrt(2). Verify Eq. (S14) by noting e^{2 i alpha c (t-t')} != 1 for all 0 <= t < t' <= d-1 and c=1,...,d-1, since alpha/pi is irrational, so the Vandermonde blocks N(c) are nonsingular for every d. Additionally, compute the symbolic or high-precision determinant of T for d=6 and d=7 using this phase choice to confirm full rank. If these checks pass, the existence gap is closed and the sufficiency theorem is complete.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central reconstruction step is Eq. (10), q = T g, and the claim that d+1 projective bases are sufficient in every dimension reduces to invertibility of T. The Supplemental proves invertibility only under Eq. (S11)/(S14): one must exhibit phases theta^(1)_m for which, for each c=1,...,d-1, the d cyclic differences theta^(1)_t - theta^(1)_(t+c mod d) are pairwise distinct modulo 2pi. The paper states this condition and then gives one numerical instance, phi=0.5671 for d=6, but it never proves that such a phase sequence exists for arbitrary d. If no such sequence existed in some dimension, T would be singular and the proposed d+1 bases would fail. This is the most load-bearing unresolved point in the theoretical argument. The concern is weak in practice because a generic irrational quadratic phase sequence works, but the manuscript does not say this, and the theorem as stated is incomplete without an explicit existence argument. The experimental calibration issue flagged by the reader is real but secondary for the mathematical claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes a quantum state tomography protocol using d+1 projective measurement bases: the computational basis, the Fourier basis, and d-1 variants obtained by applying diagonal unitary operators R^(j) with phases θ_m^(j)=jθ_m^(1). The authors derive a linear system q = T g that connects the measured basis probabilities to the off-diagonal density-matrix elements and show, in the Supplemental Material, that T decomposes into Vandermonde blocks and is invertible when the first-order phase differences in Eq. (S14) are distinct modulo 2π. The state is then reconstructed as g = T^{-1}q. The paper also reports a silicon-photonic-chip experiment for d=6, reconstructing three states with fidelities above 0.96.","tokens_in":9214,"tokens_out":16969,"duration_ms":217259,"significance":"If the missing existence argument is supplied, the scheme gives an explicit family of d+1 bases saturating the parameter-counting lower bound in every dimension, including composite dimensions such as d=6 where a complete set of MUBs is not known. The algebraic derivation is transparent: the Supplemental Vandermonde reduction is sound, and the conditional invertibility criterion is clearly stated. The experiment is a useful proof-of-principle, though its quantitative claims are weakened by the absence of calibration and systematic-uncertainty analysis. The main theoretical gap is that the paper never proves that phases satisfying Eq. (S14) exist for arbitrary d; this is readily fixable by an explicit construction.","major_comments":[{"comment":"The invertibility theorem is conditional: T is invertible only when phases satisfying Eqs. (S11) and (S14) are chosen, but the manuscript never proves that such phases exist for every dimension d. The main text supplies only the numerical d=6 choice θ_m^(1)=0.5671 m^2 in footnote [42]. Since the claim that the protocol 'applies to arbitrary dimension d' rests on this existence, the theorem is incomplete as stated. The gap is easily repaired: take θ_m^(1)=πα m^2 with α irrational; then, for each c, the difference θ_t^(1)-θ_(t+c mod d)^(1) = -πα(2ct+c^2), and equality of two such differences modulo 2π would force α c(t-t') to be an integer, impossible for α irrational and t≠t'. Please add this existence argument explicitly and phrase the theorem as 'for every d there exist phases satisfying Eq. (11)'.","section":"Supplemental Material, Eqs. (S11) and (S14); main text Eq. (11)"},{"comment":"The experimental verification does not include a calibration or systematic-uncertainty analysis for the seven implemented unitaries. The reconstruction assumes that the chip realizes exactly the computational basis, the Fourier basis, and the diagonal-phase-varied Fourier bases; residual phase and transmission errors in the MZI network directly bias the inferred density matrices, and this bias is not reflected in the quoted statistical uncertainties. Please characterize the implemented bases (e.g., through interferometric calibration or tomographic characterization of the unitaries) and propagate systematic errors, or explicitly limit the experimental claim to a proof-of-principle demonstration.","section":"Experimental section, Fig. 2 and text near Eq. (14)"}],"minor_comments":[{"comment":"The abstract states that in d=6 'a complete set of mutually unbiased bases does not exist,' but the introduction correctly says that 'only three MUBs have been found to date' and that the construction remains an open problem. The nonexistence of a complete set of seven MUBs in dimension six is not an established fact; please rephrase the abstract (e.g., 'for which a complete set is not known').","section":"Abstract and Introduction"},{"comment":"The condition written as 'θ_t^(1)-θ_(t+c mod d)^(1) ≠ θ_t'^(1)-θ_(t'+c mod d)^(1)+2πn' is clearer as a congruence statement: the two differences should be non-congruent modulo 2π. Please adjust the notation.","section":"Eq. (11) and Eq. (S14)"},{"comment":"The maximum-likelihood estimation step is invoked but not described; please specify the likelihood model, the parameterization of the density matrix, and the constraints (e.g., positivity and trace) used to obtain the reconstructed matrices in Fig. 2, so that the reader can separate the effect of the linear inversion from that of the estimator.","section":"Experimental section, paragraph before Fig. 2"},{"comment":"The caption defines VOA, FPC, and SNSPD but not DAC; please define the acronym and state the role of the 'Classical Processing' block (e.g., whether it performs only counting or also unitary calibration).","section":"Fig. 1 caption"},{"comment":"For the numerical optimization that yielded φ=0.5671, please state the domain of the search and the tolerance, and clarify that exact optimality of the mutual-unbiasedness deviation is not required for invertibility.","section":"Footnote [42]"}],"recommendation":"major_revision","confidential_remarks":"The main theoretical gap is fixable with the irrational quadratic phase construction described in my major comment; I recommend requiring that addition. The abstract's d=6 MUB nonexistence wording should be corrected before publication. On novelty, the paper would benefit from acknowledging explicitly that generic choices of d+1 bases are informationally complete by standard frame arguments; the contribution lies in the explicit Fourier-plus-diagonal-phase family and the d=6 chip demonstration."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Core linear algebra checks out; the explicit phase-shifted Fourier family does give d+1 bases with an invertible tomography matrix, and the d=6 chip experiment is a reasonable proof-of-principle. What is genuinely new is the concrete construction, not the sufficiency statement—any d+1 generic bases are informationally complete, and the paper never acknowledges that, so the opening 'long-standing problem' framing oversells the result.\n\nThe Supplemental's Vandermonde reduction is correct: under θ_m^{(j)} = j θ_m^{(1)}, each block N^{(c)} is Vandermonde, and invertibility reduces to the distinctness of the d cyclic phase differences θ_t^{(1)} − θ_{t+c}^{(1)}. That condition (Eq. S14) is stated but never proved for arbitrary d; the paper just gives a numerical φ=0.5671 for d=6. This is a real gap but a trivial one: any irrational φ makes the differences distinct for all d, so the theorem as stated is incomplete but easily repaired. The stress-test note is right to flag it, but it is a minor patch, not a load-bearing flaw.\n\nThe experiment: three states in d=6, fidelities above 0.96. No calibration or systematic-error analysis, no raw data. That limits the claim 'practical feasibility' but it is acceptable for a Letter. The MLE step is standard.\n\nI agree with the reader's conditional verdict. The paper should be reviewed; it is a reasonable contribution to the tomography toolbox. For revision, I'd require: (1) a one-line existence argument for S14 for all d (e.g., irrational quadratic phase), (2) a remark that generic d+1 bases are already known to be informationally complete, positioning the novelty as explicit and practical, and (3) ideally some calibration numbers from the chip.\n\nBottom line: worth sending to a referee, likely publishable as a Letter after those clarifications. Not a breakthrough, but a solid, correct piece of work.","headline":"A correct and clean explicit construction of d+1 tomographic bases, but the sufficiency claim is not new and the phase-existence gap is easy to fix.","tokens_in":9834,"tokens_out":4122,"would_cite":true,"duration_ms":49632,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"d+1 projective measurement bases determine an arbitrary d-dimensional quantum state, with a proof for all d and a d=6 chip demonstration.","keywords":["quantum state tomography","minimal measurement bases","mutually unbiased bases","density matrix reconstruction","Vandermonde matrix","silicon photonic chip","six-dimensional quantum state","projective measurements"],"falsifier":"A direct falsifier would be to find a dimension $d$ for which no choice of $\\theta^{(1)}_0,\\ldots,\\theta^{(1)}_{d-1}$ makes the differences $\\theta^{(1)}_t-\\theta^{(1)}_{(t+c)\\bmod d}$ pairwise distinct modulo $2\\pi$; then some Vandermonde block $N^{(c)}$ in the Supplemental proof has a zero determinant, $T$ is singular, and the $d+1$ bases fail. Checking the determinant for candidate phases at $d=7,8,\\ldots$ is a concrete search.","tokens_in":8804,"feed_emoji":"⚛️","tokens_out":10798,"duration_ms":123038,"temperature":0.7,"pith_summary":"Full quantum state tomography normally requires on the order of $d^2$ measurements, and the best prior schemes that work for every dimension use $2d-1$ projective bases. This paper claims that $d+1$ projective measurement bases are sufficient to reconstruct an arbitrary $d$-dimensional quantum state, exactly matching the lower bound set by parameter counting. The construction uses the computational basis plus the Fourier basis and its $d-1$ phase-shifted variants, with phases chosen so that the reconstruction matrix is invertible. The authors demonstrate the scheme for $d=6$ on a silicon photonic chip, a dimension where complete mutually unbiased bases are unavailable, recovering two pure states and one mixed state with fidelities above $0.96$. If the construction holds, this settles the minimal number of projective measurement bases needed for complete tomography in every dimension.","feed_headline":"d+1 measurement bases fully reconstruct d-dimensional states","feed_subtitle":"Reaches the minimum bound and works in dimensions where mutually unbiased bases do not exist, shown for d=6.","key_machinery":"The central mechanism is the coefficient matrix $T$ defined elementwise by $T_{(d-1)j+k,\\,d(c-1)+a}=\\omega^{kc}e^{-i(\\theta_a^{(j)}-\\theta_{(a+c)\\bmod d}^{(j)})}$, which maps the $d(d-1)$ measured quantities $q$ to the off-diagonal density-matrix elements $g$. The proof shows that after fixed full-rank row transformations, $T$ becomes block-diagonal with blocks $N^{(c)}$; under the phase choice $\\theta_m^{(j)}=j\\theta_m^{(1)}$ each block is a Vandermonde matrix (a matrix whose columns are powers of distinct numbers, invertible exactly when those numbers are distinct). Its determinant is a product of differences $e^{-i(\\theta_t^{(1)}-\\theta_{(t+c)\\bmod d}^{(1)})}-e^{-i(\\theta_{t'}^{(1)}-\\theta_{(t'+c)\\bmod d}^{(1)})}$, so the distinct-difference condition in Eq. (11) is exactly the condition that no Vandermonde determinant vanishes, making $T$ invertible and the reconstruction $g=T^{-1}q$ possible.","core_discovery":"The paper's central claim is that the $d+1$ projective bases consisting of the computational basis, the Fourier basis $\\{|\\psi_k^{(0)}\\rangle\\}_{k=0}^{d-1}$, and the $d-1$ bases obtained by applying diagonal unitaries $R^{(j)}$ with phases $\\theta_m^{(j)}$ to the Fourier basis completely determine any $d$-dimensional density matrix $\\rho$. The computational basis gives the diagonal elements directly; the remaining $d$ bases produce the linear system $q=T g$, where $g$ collects the $d(d-1)$ off-diagonal elements $g_{a,b}$ and $q$ collects the measured quantities $q_k^{(j)}=d\\,p_k^{(j)}-1$. The paper proves that $T$ is invertible whenever $\\theta_m^{(j)}=j\\theta_m^{(1)}$ and the differences $\\theta_t^{(1)}-\\theta_{(t+c)\\bmod d}^{(1)}$ are all distinct modulo $2\\pi$, so the off-diagonal elements are uniquely recovered as $g=T^{-1}q$. It reports a $d=6$ experiment on a silicon photonic chip that reconstructs two pure states and one mixed state with fidelities above $0.96$, in a dimension where no complete set of mutually unbiased bases is known.","pith_inferences":["A generic choice of phases $\\theta_m^{(1)}$ satisfies the distinct-difference condition with probability one, so the $d+1$-basis scheme does not depend on fine-tuned values; almost any phases should work.","The remaining freedom in the phases can be used to control noise: minimizing the condition number of $T$ (extending the paper's one-parameter optimization for $d=6$) would make the reconstruction more stable for larger dimensions.","For states known to be low-rank, the same measurement design might allow reconstruction with fewer than $d+1$ bases, though this is an extension rather than a claim of the paper.","When comparing tomography schemes, reporting the condition number of $T$ alongside fidelity would sharpen the practical assessment, since measurement noise is amplified by $T^{-1}$."],"forward_implications":["If the construction is correct, full tomography of any $d$-dimensional state needs only $d+1$ projective bases, meeting the parameter-counting lower bound and improving on the previous universal $2d-1$ scheme.","The scheme applies to dimensions where complete mutually unbiased bases are not known, such as $d=6$, so it removes a long-standing obstacle to minimal-basis tomography.","The reconstruction is explicit: diagonal elements come from computational-basis probabilities and off-diagonal elements from $g=T^{-1}q$, so no iterative search is required in the ideal case.","The $d=6$ experiment shows the bases can be realized on a programmable silicon photonic chip, with fidelities above $0.96$."],"supporting_citations":[{"why":"Establishes the prior universal bound of $2d-1$ projective bases for full state reconstruction, the baseline the new scheme improves.","marker":"[15–17]"},{"why":"Shows that complete sets of $d+1$ mutually unbiased bases are only known when $d$ is a prime power, motivating a construction that works for every dimension.","marker":"[24–30]"},{"why":"Documents that for $d=6$ only three mutually unbiased bases are known, making it the test case for the experiment.","marker":"[31–37]"},{"why":"Supplies the maximum-likelihood estimation used to convert the measured probability distributions into reconstructed density matrices.","marker":"[9,10]"},{"why":"Provides the universal linear-optical network design used to implement arbitrary unitary transformations for the seven measurement bases.","marker":"[38,39]"},{"why":"Supplies the variational approach used to program the chip's unitaries to realize the required measurement bases.","marker":"[40,41]"},{"why":"Gives the mutual-unbiasedness deviation criterion used to optimize the diagonal phases for the $d=6$ experiment.","marker":"[28]"}],"fun_headline_variants":["d+1 bases suffice for complete quantum state recovery","Minimal tomography: d+1 bases determine any state","Quantum tomography with the minimum number of bases: d+1","d+1 bases: full state reconstruction in any dimension","Proof: d+1 measurement bases fully characterize states"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that for every dimension $d$ a choice of phases satisfying Eq. (11) exists—the paper gives a numerical example only for $d=6$—and that, experimentally, the chip implements those bases with negligible calibration error.","fun_headline_variants_meta":{"raw":{"variants":["d+1 bases suffice for complete quantum state recovery","Minimal tomography: d+1 bases determine any state","Quantum tomography with the minimum number of bases: d+1","d+1 bases: full state reconstruction in any dimension","Proof: d+1 measurement bases fully characterize states"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001269,"raw_usage":{"total_tokens":5191,"prompt_tokens":938,"completion_tokens":4253,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":554,"completion_tokens_details":{"reasoning_tokens":4173}},"tokens_in":554,"tokens_out":4253,"duration_ms":34270,"temperature":1.0,"reasoning_tokens":4173,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T17:16:10.185066+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct falsifier would be to find a dimension $d$ for which no choice of $\\theta^{(1)}_0,\\ldots,\\theta^{(1)}_{d-1}$ makes the differences $\\theta^{(1)}_t-\\theta^{(1)}_{(t+c)\\bmod d}$ pairwise distinct modulo $2\\pi$; then some Vandermonde block $N^{(c)}$ in the Supplemental proof has a zero determinant, $T$ is singular, and the $d+1$ bases fail. Checking the determinant for candidate phases at $d=7,8,\\ldots$ is a concrete search.","supporting_citations":[],"review_version":1}