{"id":"34ba2e72-6e8d-4f1a-b090-b166625f2490","arxiv_id":"2502.00037","paper_version":4,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Superstate Quantum Mechanics generalizes the quantum eigenvalue problem to a matrix-eigenproblem for unitary operators, unifying inverse problems, higher-order maps, and quadratic optimization.","lead":"The paper proposes Superstate Quantum Mechanics, a framework in which quantum states are unitary operators constrained by multiple quadratic relations and energy is a quadratic function of these operators. It derives a matrix-valued eigenvalue problem for the stationary case and sketches two possible dynamical equations, positioning the framework as a bridge between inverse problems and machine learning.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (51) is not a consistent non-stationary extension: for generic S it neither preserves unitarity (22) nor evolves stationary solutions of (28) by the expected phase, so the claimed 'bridge' is unsupported.","rationale":"The stationary algebraic core of the paper—the Lagrange derivation of (28), the trace identity F = Trλ, the orthogonality relation (30), and the canonical-form theorem—is internally sound, and I do not object to it. The available code and the honest discussion of the quadratic-fidelity restriction are supporting. The load-bearing weakness is in the dynamical half of the claimed bridge. The reader's weakest assumption (quadratic fidelity) is a real but explicitly acknowledged applicability limit; it narrows the domain of the algebraic theorem but does not threaten the theorem itself. The Section VI dynamics, by contrast, are presented as part of the SQM framework and underwrite the abstract's 'bridge' claim, yet no proposed equation is both nontrivial and consistent with the unitary state space. Eq. (51) fails unitarity for generic S, and even its action on stationary solutions is wrong: the phase factor does not commute with the superoperator S. Eq. (57) is consistent but trivial, and Eq. (59) does not fix the problem. This is not a disagreement with current consensus; it is an internal consistency check of the proposed dynamics. A numerical residual test would settle it directly. Because the paper labels the exact dynamics as future work in places, the appropriate verdict remains CONDITIONAL, not rejection; the mathematical core can stand, but the physical/dynamical claims should be explicitly demoted or supported.","tokens_in":30574,"tokens_out":22154,"duration_ms":215621,"concrete_test":"Compute a stationary pair (λ[s], U[s]) for a random generic Hermitian tensor S with D = n = 4 using the authors' code [36]. Then set U(t) = exp(−iλ[s]t/ℏ)U[s] and evaluate the residual R(t) = ||S U(t) − λ[s]U(t)|| for several t > 0 (e.g., t = 0.1, 1, 10). If R(t) is nonzero at machine precision, Eq. (51) does not evolve its own stationary solutions by the expected phase. As a secondary check, integrate Eq. (59) with a = b = 1 from a random unitary initial condition and monitor ||U(t)†U(t) − I||; growth confirms the proposed nontrivial dynamics leaves the state space.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section VI proposes iℏ∂U/∂t = SU (Eq. 51) as the non-stationary counterpart of the stationary problem (28), and the abstract claims this 'naturally bridges' direct and inverse quantum mechanics. For a generic Hermitian superoperator S, Eq. (51) does not preserve the unitary constraint (22); the paper's own exponential solution (52) satisfies only the simplified constraint (26). More tellingly, (51) is inconsistent with its own stationary states in the usual quantum-mechanical sense. If SU[s] = λ[s]U[s], the expected phase evolution is U(t) = exp(−iλ[s]t/ℏ)U[s]. Because S is a tensor superoperator, S(U(t)) is not generally exp(−iλ[s]t/ℏ)SU[s]; only for the degenerate forms (53) or (64) does the phase factor pass through. Thus R(t) = ||S U(t) − λ[s]U(t)|| is generically nonzero for t > 0. The alternative (57) preserves unitarity but reduces to a common phase, and the GPE combination (59) inherits the non-unitarity of (51) unless a = 0. The paper concedes that the exact dynamic equation is future work, but the abstract's bridge claim overstates what is actually demonstrated.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper introduces 'Superstate Quantum Mechanics' (SQM), a framework in which quantum states are matrices (typically unitary or partially unitary operators) subject to multiple quadratic constraints, and the 'energy' is a quadratic form of the state with a Hermitian tensor S. The main formal result is the derivation, via Lagrange multipliers, of a stationary equation SU = λU, where λ is a Hermitian matrix of multipliers (an 'eigenmatrix'), and the claim that this algebraic problem generalizes the eigenvalue problem and captures the global maxima of the fidelity in the constrained optimization. The paper also proposes two non-stationary equations, iℏ∂U/∂t = SU and iℏ∂U/∂t = ⟨U|S|U⟩ U, discusses a hierarchy of solutions and orthogonality conditions, gives a canonical-form transformation, and outlines applications to quantum channel learning and classical computation via density-matrix networks. The paper explicitly acknowledges limitations: exact quadratic fidelity is available only for pure-state mappings, the numerical algorithm's global convergence is an open question, and the exact form of the dynamics is left for future research.","tokens_in":30859,"tokens_out":8879,"duration_ms":75435,"significance":"The stationary derivation is clean and self-contained: constructing the Lagrangian (27) and varying it does yield the algebraic problem (28), and the orthogonality relation (30) follows from Hermiticity of S. The paper is also honest about its scope, flagging the quadratic-fidelity restriction, the open convergence question, and the speculative status of the dynamics. If the algebraic problem (28) and its hierarchy are established rigorously, the result would be a useful reformulation of a class of quadratically constrained unitary optimization problems, with potential applications in unitary learning and quantum tomography. However, the advertised 'bridge' between direct and inverse quantum mechanics rests on the non-stationary equation (51), which is shown in the paper (and acknowledged there) to violate unitarity for generic S; the bridge is therefore not established. The unproven solution count (Dn) and the mixed-state fidelity proxies further delimit the claims. Overall, the paper contains a defensible core result (the stationary equivalence) embedded in a much broader and more speculative narrative.","major_comments":[{"comment":"The proposed linear dynamic equation iℏ∂U/∂t = SU is not a consistent time-dependent generalization for generic Hermitian tensors S. For a stationary solution SU[s] = λ[s]U[s], the expected phase evolution U(t) = exp(−iλ[s]t/ℏ)U[s] is not a solution of (51) because S(U(t)) ≠ exp(−iλ[s]t/ℏ)SU[s] in general, and the formal exponential solution (52) violates the unitarity constraints (22) except for the special forms (53) and (64). Since the paper's abstract and Section VII claim that SQM 'naturally bridges direct and inverse quantum mechanics problems,' this claim is not supported by the demonstrated results, and the paper's own admission that the exact dynamic equation is future research (Section VII) should be reflected in the abstract.","section":"VI, Eq. (51)"},{"comment":"The statement 'The total number of solutions is up to Dn' (s = 0 . . . Dn−1) is presented without proof or reference. This count is load-bearing for the expansion (34) and for the claim in Eq. (37) that a mixed-unitary channel can achieve the same fidelity with a full basis Ns = Dn. Because (28) is a new algebraic problem, this completeness property requires a proof or an explicit conjecture; the numerical evidence in Appendix A, which obtains only a few solutions and in two of four test cases fails to converge for the exact orthogonality constraint (A6), does not support the count as an established fact.","section":"IV, Eq. (29)"}],"minor_comments":[{"comment":"The reference [36] is titled 'The code for polynomials calculation,' but the text describes it as the software accompanying this paper; please update the reference or the description so readers can locate the SQM code.","section":"Ref. [36]"},{"comment":"Eq. (59) combines (51) and (57); since the aSU term alone violates unitarity for nonzero a, the text should state explicitly that unitarity is preserved only in the a=0 limit.","section":"VI, Eq. (59)"},{"comment":"The statement that the runs 'clearly demonstrate the correctness of the orthogonality of solutions in the form of Eq. (30)' is too strong: two of the four reported tensors fail to converge (fdiffSK ≈ 71.9 and ≈ 172.5, flagOK=false). Please rephrase to indicate a partial demonstration.","section":"Appendix A"},{"comment":"The 'Theorem' in Section IVB is an identity transformation rather than a structural simplification; the proof selects U and V using the solution itself. Calling it a definition or observation would be more precise.","section":"IV.B"},{"comment":"The notation '− − → U max' in Eq. (27) is unclear; standard notation such as 'max_{U}' would improve readability.","section":"Eq. (27)"}],"recommendation":"major_revision","confidential_remarks":"The paper's core stationary result is sound, but the manuscript is a mixture of a publishable algebraic finding and speculative physical claims. The dynamic equation (51) is not unitarity-preserving, and the paper concedes the exact dynamics are unknown; this weakens the 'bridge' narrative in the abstract and conclusion. The unproven solution count and the reliance on the authors' previous numerical results [30,31] also need attention. For a physics journal, I would recommend major revision with the expectation that the speculative sections be clearly separated from the proven results; the algebraic section could also be published as a standalone paper. There is no indication of misconduct or undisclosed duplication; the self-citations are relevant. The reference [36] appears to point to unrelated software; this should be corrected."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper's central stationary result holds up; its abstract's bridge claim does not. The algebraic problem SU = λU, derived from a QCQP over unitary operators, is a legitimate generalization of the eigenproblem, and the canonical-form theorem (Section IVB) is a real contribution. The λ-dependent orthogonality conditions and the two-Hamiltonian reduction to a higher-order quantum map are also new and correct as far as they go. The paper is refreshingly honest about what is speculation: no known physical process realizes these states, fidelity for general mixed states is only proxy, and the exact form of the dynamic equation is explicitly future work.\n\nThe soft spots are where claims outrun results. Equation (51), iℏ∂U/∂t = SU, does not preserve unitarity for a generic Hermitian S; the exponential solution only satisfies the simplified constraint, not the full one. Worse, (51) is not consistent with its own stationary solutions in the usual phase-evolution sense: for SU[s] = λ[s]U[s], the phase factor does not generally pass through S, so the residue grows. The stress-test note is right. The paper's own text concedes much of this, but the abstract still claims a 'natural bridge' — an overstatement. A referee should ask for that language to be toned down and the dynamics section reframed as exploratory.\n\nTwo smaller issues. The claim that the number of solutions is 'up to D^n' is stated without proof; if it matters, it needs an argument or counterexample. And the numerical algorithm's superiority over standard QCQP solvers is asserted, not demonstrated; no benchmark is given. Since the code is provided, this is easy to fix.\n\nWorth engaging? Yes. The stationary core is a clean, reproducible piece of mathematics with potential use in unitary learning and constrained optimization. The physical interpretation is speculative but clearly labelled. I'd send it to a referee who knows QCQP and higher-order quantum maps, with instructions to focus on the dynamics section and the solution-count claim. The paper deserves a serious referee, but it needs revision before publication.","headline":"The stationary algebraic core (SU = λU) is a genuine, clean result; the abstract's 'naturally bridges' claim is the weak plank.","tokens_in":31417,"tokens_out":2208,"would_cite":false,"duration_ms":21341,"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":"A generalization of quantum mechanics with unitary operators as states reduces the stationary inverse problem to a new algebraic equation, $SU = \\lambda U$, whose eigenvalue is a Hermitian matrix.","keywords":["superstate quantum mechanics","quantum inverse problem","unitary optimization","quadratically constrained quadratic program","eigenmatrix","higher-order quantum maps","quantum channel learning","density matrix network"],"falsifier":"Take a small random Hermitian tensor $S_{jk;j'k'}$ (say $D = n = 3$), enumerate by brute force every stationary point of the quadratic fidelity on the manifold of unitary operators, compute all solutions of $SU = \\lambda U$, and check that the two sets coincide and that the solution with the largest $\\operatorname{Tr}\\lambda$ matches the brute-force global maximum; any stationary point that fails to satisfy $SU = \\lambda U$ would falsify the reduction. The paper's own reported examples (in Ref. [31]) where a quadratic proxy fidelity outranks the true channel fidelity for mixed-state mappings provide an immediate second test of where the quadratic-form requirement breaks.","tokens_in":30377,"feed_emoji":"⚛️","tokens_out":12980,"duration_ms":105030,"temperature":0.7,"pith_summary":"This paper proposes a generalization of quantum mechanics in which the state is not a unit-norm vector $\\psi$ but a unitary operator $U$, a collection of complex numbers tied together by many quadratic constraints instead of just one. The central claim is that maximizing any quadratic fidelity under those constraints — the mathematical form of the quantum inverse problem, where a system is reconstructed from observations — is equivalent to a new algebraic problem, $SU = \\lambda U$, in which the 'eigenvalue' is a Hermitian matrix $\\lambda$ rather than a scalar. This unifies the stationary and the non-stationary problems: the same Hermitian superoperator $S_{jk;j'k'}$ plays the role of the Hamiltonian, governing both the inverse problem and candidate evolution equations for the quantum channel itself. The authors derive a hierarchy of solutions with orthogonality conditions, a canonical form for each solution, and two candidate dynamical equations, and they use the framework to define a computational model that runs on a classical computer. If the framework is right, hard optimization problems in quantum tomography, unitary learning, and machine learning gain the algebraic structure of an eigenvalue problem instead of an unguided constrained search.","feed_headline":"Fidelity maximization under unitarity reduces to SU = λU","feed_subtitle":"Unitary states turn the inverse problem into an eigenproblem whose 'eigenvalue' is a matrix.","key_machinery":"The machinery is the superoperator tensor $S_{jk;j'k'}$, a Hermitian analogue of the Hamiltonian that defines the quadratic fidelity (Eq. 9), together with the algebraic problem $SU = \\lambda U$ (Eq. 28) obtained as the Lagrange stationarity condition of that fidelity under the quadratic unitarity constraints. The object that carries the argument is the eigenmatrix: a Hermitian $D \\times D$ matrix $\\lambda$ of Lagrange multipliers whose trace is the fidelity, replacing the scalar eigenvalue of the ordinary eigenproblem. Every stationary point of the quadratically constrained program is asserted to satisfy (28); a hierarchy of solutions provides the analogue of eigenstates and energy levels, and the same tensor $S$ reappears in the non-stationary equations $i\\hbar\\,\\partial U/\\partial t = SU$ and $i\\hbar\\,\\partial U/\\partial t = \\langle U | S | U \\rangle U$, making $S$ the single object that bridges the direct problem (dynamics) and the inverse problem (reconstruction).","core_discovery":"The central discovery is that a quadratically constrained quadratic program whose variables are the entries of a (partially) unitary operator $U$ — maximize $F = \\sum_{jk,j'k'} U^{*}_{jk} S_{jk;j'k'} U_{j'k'}$ subject to the unitarity constraints $\\delta_{ij} = \\sum_k U_{ik} U^{*}_{jk}$ — has as its Lagrange stationarity condition a new algebraic problem, $SU = \\lambda U$, where $S$ is the Hermitian tensor defining the quadratic fidelity, $U$ is the 'eigenstate' (a unitary operator), and $\\lambda$ is a Hermitian matrix of Lagrange multipliers, the 'eigenmatrix.' The fidelity at a solution equals the trace of the eigenmatrix, $F = \\operatorname{Tr}\\lambda$, exactly as the energy in ordinary quantum mechanics is the eigenvalue. The problem has up to $Dn$ solutions $(U^{[s]}, \\lambda^{[s]})$ forming a hierarchy; each $\\lambda^{[s]}$ can be read as a 'second Hamiltonian' of a quantum system whose ordinary Hamiltonian is $i \\ln U^{[s]}$. The paper proves a canonical-form theorem (with $\\lambda$ diagonal and $U$ the identity matrix when $D = n$), derives orthogonality conditions that depend on the eigenmatrix itself, and shows that for a two-Hamiltonian form of $S$ the same tensor reproduces standard higher-order quantum maps — evidence that $S$ governs both the stationary inverse problem and the evolution of the quantum system itself.","pith_inferences":["If the equivalence of stationary points and solutions of $SU = \\lambda U$ holds, the same algebraic structure should transfer to any quadratic-constraint optimization — partially unitary learning, Kraus-operator constraints, and possibly classical dimensionality reduction — giving guaranteed-hierarchy alternatives to gradient methods beyond the quantum examples in the paper.","The eigenmatrix-dependent orthogonality condition suggests that superpositions of stationary solutions are generically not physical states; the paper's density supermatrix (a mixed unitary channel, Eq. 55) is then the natural state space, and one could test whether repeated 'measurements' of $\\lambda$ follow a probabilistic rule based on $\\operatorname{Tr}\\lambda$ or something new.","A numerical test the authors flag but leave open: compute the density of states over fidelity for tensors $S$ built from samples of a known Hamiltonian and check their conjecture that the distribution is shaped by the observation sample rather than by the Hamiltonian; if true, this becomes a diagnostic tool for quantum tomography.","The speculative step — a physical process that solves the inverse problem in a single measurement and returns a Hermitian matrix — would have a distinctive signature (a matrix-valued measurement outcome with no POVM realization); if no such process exists, the framework still stands as a classical computational model, which is the paper's immediately available application."],"forward_implications":["Every stationary point of a quadratic fidelity under unitarity constraints satisfies $SU = \\lambda U$, so inverse problems that can be posed in quadratic form are solvable by eigenproblem-type methods that produce a hierarchy of candidate solutions rather than a single gradient guess.","The fidelity of solution $s$ is $F = \\operatorname{Tr}\\lambda^{[s]}$, giving the inverse problem a spectrum and a density of states over fidelity, with the maximum-fidelity solution playing the role of the ground state.","The same tensor $S$ that defines the stationary inverse problem also defines candidate dynamics, and for the two-Hamiltonian approximation those dynamics reduce to known higher-order quantum maps, so direct and inverse quantum mechanics are governed by one object.","2D circuits with two independent sequences of unitary gates can transform one quantum system into another, generalizing 1D quantum circuits, with the ground-state solution $U^{[0]}$ as the natural initial system.","For pure-state data of the standard machine-learning form, reconstructing a general quantum channel reduces exactly to a quadratically constrained quadratic program, yielding a classical computational model — a hierarchy of quantum channels, or density matrix network — invariant under non-degenerate linear transformations of the data."],"supporting_citations":[{"why":"Supplies the partially unitary learning formulation and the iterative numerical algorithm that the algebraic problem (28) builds on.","marker":"[30]"},{"why":"Provides the quantum channel learning algorithm, the hierarchy construction, and the counterexamples where quadratic proxy fidelities outrank true fidelity.","marker":"[31]"},{"why":"The higher-order quantum operations framework that the two-Hamiltonian dynamics (66) reproduces, anchoring the non-stationary claim.","marker":"[12]"},{"why":"The optimal quantum learning of a unitary transformation that motivates the stationary inverse problem.","marker":"[13]"},{"why":"Supplies the mixed-state fidelity definition (13)-(14) and the quantum channel background used throughout.","marker":"[3]"},{"why":"Provides the expected fidelity for pure-to-mixed mappings (C8) that makes the classical computational model exactly quadratic.","marker":"[4]"},{"why":"The unitary evolution recurrent neural networks work that motivates unitary learning in machine learning and artificial intelligence.","marker":"[26]"}],"fun_headline_variants":["Superstate quantum mechanics: when fidelity becomes an eigenmatrix","Multiple constraints, one new equation: SU = λU","The inverse quantum problem as a matrix eigenproblem","Quantum states with multiple constraints yield SU = λU","From Hamiltonian to eigenmatrix: a new superstate theory"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the quantity being maximized can be written exactly as a quadratic expression in the unknowns, which holds for pure-state data but only approximately for mixed-state quantum channels; if that fails, the algebraic problem $SU = \\lambda U$ does not apply and only unguaranteed numerical search remains.","fun_headline_variants_meta":{"raw":{"variants":["Superstate quantum mechanics: when fidelity becomes an eigenmatrix","Multiple constraints, one new equation: SU = λU","The inverse quantum problem as a matrix eigenproblem","Quantum states with multiple constraints yield SU = λU","From Hamiltonian to eigenmatrix: a new superstate theory"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001238,"raw_usage":{"total_tokens":5164,"prompt_tokens":1107,"completion_tokens":4057,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":723,"completion_tokens_details":{"reasoning_tokens":3961}},"tokens_in":723,"tokens_out":4057,"duration_ms":28491,"temperature":1.0,"reasoning_tokens":3961,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T14:24:41.513038+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a small random Hermitian tensor $S_{jk;j'k'}$ (say $D = n = 3$), enumerate by brute force every stationary point of the quadratic fidelity on the manifold of unitary operators, compute all solutions of $SU = \\lambda U$, and check that the two sets coincide and that the solution with the largest $\\operatorname{Tr}\\lambda$ matches the brute-force global maximum; any stationary point that fails to satisfy $SU = \\lambda U$ would falsify the reduction. The paper's own reported examples (in Ref. [31]) where a quadratic proxy fidelity outranks the true channel fidelity for mixed-state mappings provide an immediate second test of where the quadratic-form requirement breaks.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the partially unitary learning formulation and the iterative numerical algorithm that the algebraic problem (28) builds on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the quantum channel learning algorithm, the hierarchy construction, and the counterexamples where quadratic proxy fidelities outrank true fidelity."},{"cited_title":"Bisio, G","cited_arxiv_id":null,"evidence_quote":"The optimal quantum learning of a unitary transformation that motivates the stationary inverse problem."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the expected fidelity for pure-to-mixed mappings (C8) that makes the classical computational model exactly quadratic."},{"cited_title":"Arjovsky, A","cited_arxiv_id":null,"evidence_quote":"The unitary evolution recurrent neural networks work that motivates unitary learning in machine learning and artificial intelligence."}],"review_version":1}