SuperMatics research

Research

The software and intelligence layer to build the future of superconductivity.

Papers and a certificate on the mathematics of superconductivity. Each entry states a result, its assumptions, and how it was checked.

Read the results

Mathematical physics · 22 research records

Results by model

Where we used computer checks, we identify the exact parts of the proof that they cover.

  1. The critical-temperature ceiling

    Submitted (J. Stat. Phys.)

    Sharp bounds on the Eliashberg critical temperature: a comparison principle, convexity in the spectrum, and the Kiessling-Altshuler-Yuzbashyan conjectures

    We give an independent proof of the sharp ceiling on the linear critical temperature in isotropic Eliashberg theory, anticipated by Leavens in 1977.

    Why it matters
    The ceiling gives Eliashberg solvers, including workflows using EPW, a rigorous reference bound within the covered isotropic model.
    Scope
    Phonon-mediated, linearized isotropic Eliashberg theory with zero Coulomb pseudopotential and finite second spectral moment. The ceiling is Tc < C√(λ⟨ω²⟩), with sharp C ≈ 0.1827. The upper bounds also persist with a symmetric Coulomb cutoff. This is not a universal bound on superconductors or a solver-convergence theorem.
    • Interval-arithmetic certificates

      Two computer-assisted proofs confirm the Einstein bound independently of the comparison principle.

    • Lean 4

      The finite-dimensional algebraic core, including the third KAY conjecture; not the complete infinite-dimensional theorem.

    • Independent check

      Separate implementations check the Einstein bound certificates.

  2. The superconducting energy minimum

    Submitted (J. Stat. Phys.)

    Hidden convexity of the Eliashberg condensation energy: existence, uniqueness and continuity of the superconducting state

    Hidden convexity proves that the covered planar Eliashberg chain has one energy-minimizing state up to gauge, varying continuously with temperature.

    Why it matters
    It connects linear instability to the thermodynamic transition in the covered isotropic model, rather than relying on a linear test alone.
    Scope
    Zero Coulomb pseudopotential; admissible real planar chains with positive, non-increasing interactions and power-law decay. Temperature continuity needs pointwise continuity and locally uniform decay bounds. Phonon applications use separate spectral input; nonplanar spins are outside this theorem.
    • Lean 4

      Finite-dimensional algebraic and convex-analytic core. Compactness, existence, and the infinite-chain limit are proved on paper; numerical examples are illustrations only.

  3. Multiband and anisotropic models

    Preprint

    Global uniqueness of the superconducting state in multiband and anisotropic Eliashberg models without Coulomb repulsion

    The covered attractive multiband and anisotropic Eliashberg models have at most one thermodynamic superconducting state, apart from a common phase.

    Why it matters
    For EPW and other anisotropic Eliashberg solvers, this rules out a second global energy minimum in the covered zero-Coulomb model.
    Scope
    Zero Coulomb pseudopotential; identical pairing and mass-renormalization kernels, with attraction and connectivity hypotheses. Infinite chains require positive-lag summability. The non-atomic Fermi surface additionally requires uniformly summable integrated rows, compact angular operators, and measurable irreducibility. No claim about every stationary branch or iterative convergence.
    • Interval-arithmetic certificates

      An interval certificate orders two real stationary branches of a three-band model.

    • Lean 4

      Finite hidden-convexity, attainment, density-uniqueness, and alignment algebra, plus supporting infinite and continuum statements under explicit hypotheses.

    • Independent check

      Two interval implementations check the three-band branch certificate. Neither formalizes the full analytic theorem.

  4. Surface and boundary superconductivity

    Submitted

    Boundary enhancement of the BCS critical temperature in three-dimensional half- and quarter-spaces

    At sufficiently weak coupling, Dirichlet boundaries raise the linear BCS critical temperature in three dimensions: the quarter-space temperature is strictly above the half-space temperature, which is strictly above the bulk temperature.

    Why it matters
    It shows how a surface, and the corner where two surfaces meet, can favor the onset of superconductivity within linear BCS theory.
    Scope
    Dirichlet boundaries and nonzero radial nonnegative interactions that satisfy integrability, first-moment, and full Fermi-sphere spectral hypotheses, at every fixed positive chemical potential. The result concerns the linear instability only. It does not cover the nonlinear transition or sign-changing interactions.
    • Interval-arithmetic certificates

      The kernel-positivity inequalities on a compact interval, joined to analytic estimates near zero and infinity.

    • Lean 4

      The elementary inequalities, the tail bound, the cell covering, and the series truncation. The analytic, spectral, and interval-library inputs are assumptions of the formalization.

    • Independent check

      Two Arb implementations and a separate mpmath interval implementation check the kernel inequalities.

    Public release links are not yet available.

  5. Vortex instability directions

    Technical reportSM-TR-2026-01

    Computer-assisted Morse indices of nonmagnetic Ginzburg-Landau vortices of degrees four through eight

    Conditional on the stated analytic inputs, vortices of degrees 4 to 8 have 12, 20, 30, 36, and 48 independent energy-lowering directions, respectively.

    Why it matters
    It sharpens the mathematical study of vortex stability relevant to type-II superconductors, in the nonmagnetic planar Ginzburg-Landau model.
    Scope
    Standard planar nonmagnetic radial vortices, conditional on the paper's analytic spectral and variational inputs. Bounded-kernel nondegeneracy is proved only at degree 4. This is not magnetic vortex-lattice stability.
    • Exact-rational certificates

      Profile barriers and matching variational bounds, with coupled cell estimates and interval pivot inertia at higher degrees.

    • Lean 4

      Degree-4 polynomial statements; only 15 of 97 generated higher-degree integer-list modules are checked. Analytic spectral inputs and variational arguments are not formalized.

    • Independent check

      A separate FLINT implementation checks the finite certificates.

  6. Lattice spectral stability

    Technical reportSM-TR-2026-02

    Certified non-hexagonal spectral stability in the lowest-Landau-level equation

    Two non-hexagonal magnetic-periodic states of the Hamiltonian lowest-Landau-level equation have real spectrum in every linearized Bloch fiber.

    Why it matters
    It distinguishes spectral stability from stronger dynamical stability in a lattice model related to vortex arrays in type-II superconductors.
    Scope
    The Hamiltonian lowest-Landau-level equation, not full superconducting vortex dynamics. The zero fiber has a nonzero nilpotent block, allowing at-most-linear growth. No uniform boundedness, dispersive stability, nonlinear stability, or full lattice-shape classification is proved.
    • Interval-arithmetic certificates

      Gaussian-symbol determinant positivity and quartic leading terms; the lattice-shape map remains partial.

    • Lean 4

      Conditional moment algebra, scaling, threshold arithmetic, and subdivision coverage, not the Gaussian symbol or interval implementation.

    • Independent check

      A second arithmetic implementation checks the two-point certificates. The stable-box and transition-curve certificates each use one interval implementation; a second covers only 12 of 4,178 stable leaves. The repository’s automated tests rerun a documented subset of the partial-map checks.

  7. The effect of Coulomb repulsion

    Technical reportSM-TR-2026-03

    Linear versus global stability in a Coulomb-penalized Matsubara spin chain

    A certified strong-Coulomb example has a linearly stable normal state but a lower-energy unrestricted spin configuration; in the mirror sector, the two stability notions coincide.

    Why it matters
    It shows why zero-Coulomb arguments cannot simply be carried over to unrestricted spin models with Coulomb repulsion.
    Scope
    Einstein kernel with a symmetric both-index Coulomb penalty. The counterexample uses coupling λ = 8, index window N_c = 5, and μ* = 20, with longitudinal spin reversals. It is outside the conventional causal/no-flip class and does not predict a causal superconducting state or physical first-order transition.
    • Interval-arithmetic certificates

      Exact energy decomposition and infinite-tail enclosures certify the stated finite-window stability inequalities.

    • Lean 4

      Finite-support and block identities, sufficient positivity, mirror comparison, and conditional threshold implications. Infinite-sum, compactness, attainment, and coexistence arguments are not formalized.

    • Independent check

      A separate interval implementation checks the certificate under the stated roundoff assumptions for the numerical library.

  8. Energy-dependent electronic density of states

    Preprint

    Sharp Eliashberg pairing bounds for an energy-dependent density of states

    The sharp scalar Eliashberg ceiling extends to bounded energy-dependent densities of states on the covered self-consistent normal branches.

    Why it matters
    It gives a reference bound that accommodates asymmetric electronic bands within the stated model.
    Scope
    Linearized isotropic Migdal-Eliashberg theory with the paper's self-consistent normal equations and bounded density of states. This is not an exact-Hamiltonian or general superconducting transition bound.
    • Replay certificates

      Exact and interval checks cover specified inequalities and bounds; numerical cross-checks are not proofs.

    • Lean 4

      Finite algebraic components only, not the full infinite-dimensional theorem.

  9. Thermal on-site pair order

    Preprint

    Thermal eta-pair order at exact half filling with on-site repulsion

    Positive-temperature staggered on-site pair order is proved for a prescribed pair-hopping model with on-site repulsion.

    Why it matters
    The theorem extends a reflection-positive pair-hopping construction to include on-site repulsion.
    Scope
    Dimensions three and higher, prescribed flux hopping, and the coupling and temperature restrictions in the paper. This is not superconductivity in the ordinary repulsive Hubbard model, nor a proof of stiffness or transport.
    • Replay certificates

      Exact checks and a certified integral bound support specified constants in the argument.

    • Lean 4

      Finite algebraic components only; the full thermal-order theorem is not formalized.

  10. Gamma-model constants

    Certificate

    Certified Eliashberg gamma-model linear-instability constants: replay certificate

    A software and data record gives interval enclosures for five linear-instability constants in the isotropic gamma model.

    Why it matters
    The versioned certificate supplies a reproducible input for the continuum Yukawa-SYK bound.
    Scope
    Enclosures depend on the analytic tail estimates in the record. The independent baseline does not match the narrowest widths at gamma one-half and two. These are model constants, not measurements or thermodynamic transition temperatures.
    • Interval-arithmetic certificates

      Replay checks the stated enclosures, conditional on the analytic tail estimates.

    • Lean 4

      Finite algebra only; the infinite tail is not formalized.

  11. Continuum large-N pair formation

    Preprint

    A sharp bound on pair formation in the continuum Yukawa-SYK saddle

    A sharp upper bound controls the linear pair-formation temperature in the covered continuum large-N saddle.

    Why it matters
    It identifies a limiting scale for the model while keeping the onset distinct from an equilibrium transition.
    Scope
    The regularized infinite-band continuum saddle at infinite N, with the assumptions stated in the paper. The decimal enclosure depends on the linked gamma certificate. This is not a measurement or a general superconducting ceiling.
    • Replay certificates

      Exact rational checks cover specified factors and the interval conversion; the gamma replay uses the companion certificate.

    • Lean 4

      Five finite rational identities only, not the spectral argument.

  12. Electronic pair order with slow phonons

    Preprint

    Suppression of electronic pair order for slow lattice phonons

    In the covered lattice models, an explicit slow-phonon condition gives exponential decay of local electronic charged correlations and an upper bound on the corresponding pair-order temperature.

    Why it matters
    It constrains electronic pair order in two exact electron-phonon models without treating the linearized Eliashberg onset as a many-body transition.
    Scope
    Pure spinful square-lattice Holstein and bond SSH models, with the paper's ensemble and parameter hypotheses. The positive-mass result does not bound helicity or phonon-dressed order. It is not a general superconducting ceiling or a proof of a nonzero transition.
    • Replay certificates

      Exact rational checks and 25 mutant rejections cover specified finite factors, not the analytic arguments.

    • Lean 4

      Finite rational lemmas only; the infinite-volume and analytic arguments are not formalized.

  13. Pair order on cylinders

    Preprint

    Pair density matrices and axial helicity for quantum bond phonons on cylinders

    On fixed-width cylinders and chains at positive temperature, the quantum bond SSH model has a length-uniform bound on the pair density matrix and exponentially vanishing axial isothermal helicity.

    Why it matters
    The estimates are uniform in phonon energy and canonical sector. The density-coupled Holstein result is of known type.
    Scope
    Chains and finite-width cylinders under the stated geometry and coupling assumptions. This is not a square-lattice BKT bound or a general superconducting ceiling.
    • Replay certificates

      Exact checks reject 19 mutants covering selected finite factors and logic, not the analytic proof.

    • Lean 4

      Finite lemmas only; the functional-analytic and thermodynamic arguments are not formalized.

  14. Dot large-N pair formation

    Preprint

    A sharp bound on pair formation in the Yukawa-SYK dot

    For the centered, causal, equal-flavor large-N Yukawa-SYK dot saddle with real Yukawa couplings and full phonon polarization, the strong-coupling limiting temperature is a strict upper bound at every positive finite coupling.

    Why it matters
    It resolves an intermediate-coupling overshoot question in a single-level dot where density-of-states height bounds are vacuous.
    Scope
    Centered causal equal-flavor large-N saddles with real Yukawa couplings and full dynamical polarization; the decimal enclosure uses the companion gamma-two certificate. This is not a claim about first-order transitions, noncentered saddles, finite-N transitions, or materials, and it is not a general superconducting ceiling.
    • Replay certificates

      Exact rational identities, closed flavor-interval enclosures, and nine expected mutation refusals; the gamma-two enclosure is conditional on analytic tail estimates.

    • Lean 4

      Finite rational identities for folding and flavor guards only, not the infinite-dimensional or interval arguments.

  15. Retarded glue and linear pairing onset

    Preprint

    What a glue moment bounds in linear Eliashberg theory

    In matched scalar linear Eliashberg theory, the sharp ceiling uses the total retarded glue moment, not its phonon part alone. A strict moment-norm ceiling, with a coefficient not asserted to be sharp, covers centered finite-channel reciprocal reduced models and qualifying causal shifted branches.

    Why it matters
    It separates the moment controlled by a pairing equation from a phonon-only material scale and identifies what a shifted normal branch must satisfy.
    Scope
    The shifted extension requires channel-wise radial-normal causal continuation and bounded normal frequency rows at each tested temperature. Relative-real sharpness and shifted-versus-centered excess remain open. This is a linearized onset bound, not an exact Hamiltonian or measured transition temperature, and not a general superconducting ceiling.
    • Exact replay checks

      Eleven base replays and 70 guarded mutant rejections cover finite factors. Edge-matrix and Harnack samples are diagnostic checks, not universal certificates.

    • Lean 4

      Six finite algebra files only; the infinite-dimensional analytic proofs and imported scalar theorem are not formalized.

  16. A two-thirds-power bound for electronic pair order

    Preprint

    A two-thirds-power adiabatic bound for electronic pair order

    For a fixed local electronic charge-two field and its conjugate, the covered lattice models have exponential clustering above an explicit two-thirds-power adiabatic threshold.

    Why it matters
    The result constrains electronic pair order in exact Holstein and bond SSH models while leaving the linear phonon-energy bound open.
    Scope
    Spinful square-lattice Holstein and bond SSH models with independent harmonic phonons, fixed hopping t, and no direct electron interaction. The threshold is tau³ ≥ 2²⁵ U Omega²[1+log(1+E/tau)], where tau = k_B T, U = F²/K, and E = t for Holstein or t+U for bond SSH. The theorem covers grand-canonical states and periodic thermodynamic canonical limits at fixed interior density. The linear bound tau ≤ C Omega remains open. Atomic-cutoff sharpness applies only to the paper's rare-history quadratic-reference method class, not a physical transition exponent.
    • Replay certificates

      Exact rational identities, seven interval covariance points, and 22 expected mutant rejections cover specified finite factors and constants only.

    • Analytic proof

      Clustering, thermodynamic limits, ensemble transfer, and the method-class sharpness proposition are proved in the paper and companion notes, not formalized by the finite replays.

  17. A sharper moment-norm ceiling for coupled channels

    Preprint

    A √(377/250)/(2π) moment-norm ceiling for causal multichannel Eliashberg branches

    For finite-channel reciprocal reduced Eliashberg models with bounded channel densities of states, the linear pairing temperature obeys k_B T_lin < √(377/250)/(2π) · √ρ, where ρ is the norm of the reciprocal moment matrix, on centered branches and on shifted branches with channel-wise radial-normal causal continuation.

    Why it matters
    It tightens the coefficient of the predecessor moment-norm ceiling while the coupled sharp coefficient remains open.
    Scope
    Linearized pairing onset on centered normal branches and on shifted branches with channel-wise radial-normal causal continuation. Normal frequency rows are bounded separately at each tested temperature, not uniformly in temperature. The decimal 0.19544330695 and the relative gap of about 7.0% to the sharp scalar constant are numerical, not interval certified. The full coupled sharp endpoint, the full shifted-versus-centered comparison, and any general many-body or material transition ceiling remain open. This is a linearized onset bound, not an exact Hamiltonian or measured transition temperature, and not a general superconducting ceiling. The preprint is not peer reviewed.
    • Replay certificates

      Two exact rational coverings, a primary one with 537 leaves and an independent one with 437 leaves, and 14 expected mutant rejections cover the finite inequality with certified edge reserve 247/1000 against the reported 123/500. The 72 supplementary finite forms are samples, not universal certificates.

    • Lean 4

      One file of finite polynomial identities and rational budgets only; the infinite operator, the covering algorithm, Harnack, and the Poisson rearrangement are not formalized.

  18. Imaginary-time estimates and method barriers for pair order

    Preprint

    Imaginary-time estimates and method barriers beyond the two-thirds adiabatic bound

    For the spinful square-lattice Holstein and bond SSH models, the paper proves exact imaginary-time estimates that hold at the linear scale and eight method barriers, each paired with a certified counterexample. These are estimates and method barriers, not a linear ceiling.

    Why it matters
    It records which routes to a linear phonon-energy pairing bound fail and why, and which positive estimates survive, while the fixed-scale question stays open.
    Scope
    Pure spinful square-lattice Holstein and bond SSH models with independent harmonic phonons, fixed hopping t and coupling U, no direct electron interaction, and regular density. The linear question k_B T_pair ≤ C Omega, or C Omega log^p(1/Omega), remains OPEN. No physical counterexample, transition-temperature lower bound, material or measurement claim is made. The barriers concern the stated proof routes only.
    • Replay certificates

      Exact rational and multi-precision interval replays with 36 expected mutant rejections cover specified finite certificates and counterexamples only.

    • Analytic proof

      The positive estimates and barrier propositions are proved in the paper and companion notes; the finite replays do not formalize them and no Lean check is claimed.

  19. A uniform-frequency bound on Holstein pair order

    Preprint

    Strong-coupling suppression of electronic pair order in the Holstein model, uniformly in phonon frequency and without a hopping cutoff

    For the original spinful square-lattice Holstein model with U = F²/K > 0, equal-time on-site electronic pair correlations decay as (240/59)(5/64)^d whenever beta t²/U ≤ 2⁻²⁷, with no restriction on t/U. This gives k_B T_pair ≤ 2²⁷ t²/U, uniformly in positive phonon frequency.

    Why it matters
    It bounds electronic pair order using the virtual-hopping energy scale t²/U without a hopping cutoff or a phonon-frequency restriction.
    Scope
    Grand-canonical correlations on periodic square tori of side at least four, for every real common chemical potential. Canonical transfer concerns translation-invariant thermodynamic limits at interior total density and balanced interior separately fixed spin densities, not uniform finite-volume canonical estimates. This is not a linear-in-phonon-frequency ceiling. No material prediction, synthesis, or measured superconducting transition is claimed. Separate AI agents reviewed the proof and manuscript. This preprint has not been peer reviewed.
    • Replay certificates

      Finite exact-rational and interval witnesses with 27 expected mutant rejections cover the sixteen-state bond algebra, sector sums of squares, fugacity identity, and explicit cluster budgets.

    • Analytic proof

      The manuscript and companion notes give the pair-clustering argument. The finite replay does not formalize the oscillator trace identity, infinite-volume cluster expansion, or canonical pressure argument; no Lean check is claimed.

  20. Finite-energy multichannel pairing ceilings

    Preprint

    Moment-norm ceilings √(41/30)/(2π) and √(45/32)/(2π) for multichannel Eliashberg branches

    For reciprocal multichannel Eliashberg branches with bounded energy-dependent channel densities of states, k_B T_lin < C√ρ, where ρ is the norm of the glue moment matrix. The constants are √(41/30)/(2π) on centered branches and √(45/32)/(2π) on the stated causal Harnack class.

    Why it matters
    A two-stage tail-frequency cascade tightens the moment ceiling while retaining unequal channel densities of states and finite-energy glue. A certified obstruction shows that the stated edge-wise method cannot reach the sharp scalar constant.
    Scope
    Linear instability of the reduced reciprocal Eliashberg model under the manuscript's spectral, density-of-states, and normal-branch hypotheses. The constants are upper bounds, not claimed sharp. The edge-wise method floors are greater than 0.185411 on the centered weight class and greater than 0.188201 on the Harnack relaxation, not lower bounds on a physical transition temperature; the second floor is not asserted for actual causal branches. The sharp scalar constant for general coupled channels and any general many-body or material transition ceiling remain open. No material, synthesis, or measurement is claimed. Separate AI agents reviewed the proof, certificates, and manuscript. This preprint has not been peer reviewed.
    • Exact and interval certificates

      Two independent rational coverings certify the finite inequalities with 99/107 and 245/287 causal/centered leaves, 21 expected mutant rejections, and 15 byte-exact source pins. A separate Arb interval certificate checks the edge-wise method floors and rejects nine mutants. The 72 supplementary trial forms are samples, not universal certificates.

    • Lean 4

      Lean 4.35.0 checks finite polynomial identities and rational budgets only. The infinite operator, Harnack and Poisson estimates, width cone, and soft-limit locking are proved in the manuscript, not formalized by the replay.

  21. Hubbard-Holstein pair-correlation bounds

    Preprint

    Pair correlations in the Hubbard-Holstein model at strong coupling, uniform in the phonon frequency

    For the spinful Hubbard-Holstein model on square tori, when U_eff = U - U_c > 0 and beta t²/U_eff ≤ 2⁻²⁷, equal-time on-site pair correlations at distinct sites are bounded by (240/59)(5/64)^d, uniformly over positive phonon frequency, real chemical potential, and torus sides M ≥ 4. No separate t/U cutoff is imposed. The paper also gives bounds in the gapped repulsive and dilute regimes and at high temperature.

    Why it matters
    The estimates express pair suppression through the effective interaction scale while making the covered regimes and unresolved cases explicit.
    Scope
    Grand-canonical results are for periodic square tori of side at least four with a real common chemical potential. The canonical transfer is for translation-invariant thermodynamic limits at densities in (0,2), not uniform finite-volume canonical estimates. At half filling, the combined decay criterion is tau ≥ min{1024 t, 2²⁷ t²/|U-U_c|}, with the second threshold treated as infinity when U=U_c. Low-temperature decay inside the Mott window, near its edges, and at the degenerate point below tau = 1024 t remains unresolved. Failure of a sufficient cluster criterion in these regimes is not evidence that correlations fail to decay. These are mathematical bounds for the specified model, not experimental claims. Separate AI agents reviewed the proof and manuscript. This preprint has not been peer reviewed.
    • Replay certificates

      Finite exact-rational and interval witnesses reject 20 guarded mutants and cover the listed finite checks; the replay does not formalize the oscillator trace identity, infinite-volume cluster expansion, or canonical pressure argument.

    • Analytic proof

      The manuscript and companion notes give the arguments for the stated regimes; the finite replay does not formalize the thermodynamic-limit or pressure-transfer steps.

  22. Frequency-uniform bounds with several optical modes

    Preprint

    A frequency-uniform pair bound for a multi-mode Holstein-Hubbard model

    For spinful electrons on square tori with finitely many independent on-site optical modes, when U_eff = u - U_c > 0 and beta t²/U_eff ≤ 2⁻²⁷, the paper bounds distinct-site equal-time on-site pair correlations by (240/59)(5/64)^d, uniformly over every positive mode frequency and the mode count. The canonical pair-temperature bound is 2²⁷ t²/U_eff for translation-invariant thermodynamic limits at interior densities.

    Why it matters
    It extends the strong-coupling local pair-correlation bound to several optical modes without restricting their positive frequencies, and states the exact condition for a nonlocal reduction.
    Scope
    The grand-canonical bound is for periodic square tori M ≥ 4, real chemical potential, and distinct sites. The canonical transfer applies only to translation-invariant thermodynamic limits at interior density, not uniform finite-volume canonical estimates. A nonlocal reduction requires an exact mass-weighted Gram identity separately in each equal-frequency group; static cancellations among unequal frequencies are insufficient. Counterexamples show failure of one product-factorization route, not a no-go theorem for generic nonlocal models. A frequency-uniform bound for generic summable nonlocal forces without exact orthogonal reduction remains unresolved. These are mathematical results for the stated model, not material, synthesis, or measurement claims. Separate AI agents reviewed the proof and manuscript. This preprint has not been peer reviewed.
    • Finite replay

      Exact-rational and 70-digit outward-interval checks cover the listed finite witnesses and reject 20 guarded mutants; the replay does not formalize the infinite-dimensional oscillator trace identity, infinite-volume cluster expansion, or canonical pressure-transfer argument.

    • Analytic proof

      The manuscript proves the stated bounds under its hypotheses. The nonlocal counterexamples obstruct one product-factorization method only and do not imply a no-go result for the model.

Methods

How we check the proofs

  1. Numerical steps used in a proof have interval-arithmetic or exact-rational certificates. Readers can rerun the checks once the supplement is available.
  2. Lean 4 checks the finite parts identified in each entry. The papers state the analytic hypotheses and give the infinite-dimensional arguments.
  3. Each paper undergoes an independent review before release. Where an entry lists an independent check, a separate implementation tests the certificates specified there.

Agreement between models and floating-point illustrations can inform the work; neither is a proof.

The entries link to the public proofs and certificates that are available. An entry without links has not been released yet.

Use of AI

Under Davis Rens’s direction, AI research agents wrote the proofs and generated the certificates, Lean and other code, and manuscript drafts.

Scope of the research

These theorems concern standard mathematical models of superconductivity. They report no new materials or measurements.