Computational research
Calculations and proof diagnostics
Interactive models and reproducible calculations test finite spectra, Bell correlations, symmetry arguments, and fluid scaling.
Change a parameter and inspect the consequence.
Does a finite separation survive as the chain grows?
Sparse diagonalization of H = −Σ ZᵢZᵢ₊₁ − hΣ Xᵢ, with periodic boundaries and Pauli eigenvalues ±1. These points are computed results for N = 4, 6, 8, 10, 12, 14.
The dashed reference is 2 tan(π/4N). Every finite value is positive, while the exact expression tends to zero.
Inspect the numerical results.
| N | E₀ | E₁ | E₁−E₀ | Residual |
|---|---|---|---|---|
| 4 | -5.22625186 | -4.82842712 | 0.397825 | 3.40e-15 |
| 6 | -7.72740661 | -7.46410162 | 0.263305 | 4.95e-12 |
| 8 | -10.25166179 | -10.05467898 | 0.196983 | 1.74e-14 |
| 10 | -12.78490644 | -12.62750303 | 0.157403 | 1.51e-12 |
| 12 | -15.32259515 | -15.19150823 | 0.131087 | 1.61e-14 |
| 14 | -17.86280854 | -17.75049095 | 0.112318 | 1.22e-13 |
Watch the computed quantities change.
These silent films are rendered from the same model definitions and recorded data. Each includes captions, a transcript, and its frame parameters.
The first separation changes with chain size.
Six finite periodic Ising chains compare three field strengths. The exact critical reference and the ground-state doublet distinction stay visible.
Read the explanation and method
The first separation changes with chain size. This periodic transverse-field Ising chain has H = -sum Z_i Z_(i+1) - h sum X_i, with J = 1 and Pauli eigenvalues plus or minus one.
The film reveals the notebook's computed first-level separation, E1 minus E0, for N = 4, 6, 8, 10, 12 and 14, comparing h = 0.5, 1 and 1.5. The vertical axis is logarithmic. Lines join computed points and do not add measurements.
At h = 1, the exact finite-size reference 2 tan(pi/(4N)) agrees with the numerical values. Ringed orange points mark this comparison.
For h < 1, the first pair approaches a nearly degenerate ground-state doublet. Its splitting is not the bulk excitation gap above that manifold.
These finite one-dimensional spin calculations do not construct four-dimensional Yang–Mills theory or establish a Millennium solution.
Method. Periodic transverse-field Ising chain, Pauli normalization, J=1; first-level separation E1−E0 at finite N.
The settings determine the CHSH value.
An angle sweep shows the Bell-state CHSH value, followed by an explicitly defined white-noise visibility sweep with the classical bound held at two.
Read the explanation and method
The settings determine the CHSH value. The chosen Bell state is |Phi+> = (|00> + |11>)/sqrt(2).
Alice measures A0 = Z and A1 = X. Bob measures B0 = cos(theta) Z + sin(theta) X and B1 = cos(theta) Z - sin(theta) X.
For this state and these settings, S(theta) = 2(cos(theta) + sin(theta)). The first half sweeps theta from zero to ninety degrees through the notebook's computed one-degree samples.
At theta = 45 degrees, S = 2 sqrt(2), about 2.828. The local hidden-variable bound is |S| <= 2, assuming local outcomes and a setting-independent shared hidden-variable distribution.
The second half fixes theta at 45 degrees and adds white noise: rho_v = v|Phi+><Phi+| + (1-v)I/4. Visibility v increases from zero to one.
Here S = 2 sqrt(2) v, so violation occurs only for v > 1/sqrt(2), about 0.7071. This is specific to the stated family and settings; it is not a universal entanglement threshold.
The film presents theoretical correlations, not experimental trials. These modeled correlations do not enable signalling.
Method. Bell state |Φ+>, A0=Z, A1=X, B0=cosθ Z+sinθ X, B1=cosθ Z−sinθ X; then ρ_v=v|Φ+><Φ+|+(1−v)I/4 at θ=45°.
A uniform gap can be proved for independent spins.
Take N spin-½ sites, use Pauli matrices with eigenvalues ±1, and let h>0. Operators Xᵢ on different sites commute, so their tensor-product eigenvectors form a common eigenbasis.
H₀ = −h Σᵢ Xᵢ
Eₖ = −hN + 2hk, k = 0, …, N
Δ(H₀) = E₁ − E₀ = 2h > 0A basis vector with k negative X eigenvalues has the displayed energy. The state with every X eigenvalue positive is the unique ground state; changing one eigenvalue raises the energy by exactly 2h. The same lower bound works for every finite N. This proves the finite-volume statement uniformly in N for this decoupled model.
What happens when interactions are added?
For the periodic interaction V=−JΣᵢZᵢZᵢ₊₁, the operator norm obeys ‖V‖≤N|J|. The variational characterization of ordered eigenvalues gives λ₁(H₀+V)≥λ₁(H₀)−‖V‖ and λ₀(H₀+V)≤λ₀(H₀)+‖V‖. Subtraction yields:
Δ(H₀+V) ≥ 2h − 2N|J|.For fixed nonzero J, the estimate loses its positive lower bound as N grows. A negative lower bound does not prove that the physical separation closes; it shows that this global norm argument cannot establish the desired uniform result. A stronger route would need to exploit locality or a more precise structure of the ground state.
The next obligation is uniform control.
Any proposed Yang–Mills construction must still establish the interacting four-dimensional continuum theory and a positive spectral gap after the required limits. An exactly solved independent-spin system is a benchmark for reasoning about quantifiers, not a transfer of its theorem to gauge theory.
The finite-chain calculation separates three different questions.
The script constructs the interacting Hamiltonian directly as a sparse matrix and computes the two lowest algebraic eigenvalues for six even chain lengths and three field strengths. It retains eigenpair residuals, a fixed random seed for starting vectors, and the numerical library versions.
At J=h=1, the exact periodic-chain first separation is 2 tan(π/(4N)). For the even periodic chains and Pauli normalization used here, the fermionic sector energies at criticality are −2 csc(π/(2N)) and −2 cot(π/(2N)); their difference gives this expression. The six numerical comparisons agree within 4.83e-14. The expression tends to zero although each finite value is positive.
For h<1, the first separation instead tracks the ordered phase’s nearly degenerate ground-state doublet. A gap above the resulting ground-state manifold is a separate quantity. For h>1, the points approach the paramagnetic bulk value 2(h−1), but a finite list of points alone cannot prove that limiting statement.
Small residuals establish approximate eigenpairs. They do not, by themselves, certify that no eigenvalue was missed. The exact critical comparison and independent small-chain checks provide additional validation.
The Bell calculation makes the assumptions algebraic.
For predetermined local outcomes a₀,a₁,b₀,b₁∈{−1,+1}, write S=a₀(b₀+b₁)+a₁(b₀−b₁). Exactly one bracket is zero and the other is ±2, so S is ±2. A probability mixture over these assignments remains between −2 and 2, provided the hidden-variable distribution does not depend on the chosen settings.
For |Φ⁺⟩=(|00⟩+|11⟩)/√2, the displayed Pauli settings give S=2(cosθ+sinθ), reaching 2√2 at θ=45°. Mixing in the maximally mixed state with visibility v gives S=2√2v at that angle. The threshold v>1/√2 is specific to this state family and these settings; it is not a general threshold for entanglement.
The reduced state on either side is I/2. The outcome distribution at one side therefore remains unchanged by the remote setting. Bell inequality violation and controllable faster-than-light communication are different claims.
Symmetry alone does not place every zero on a line.
Consider the polynomial P(s)=[(s−¼)²+1][(s−¾)²+1]. It has real coefficients, its two factors exchange under s→1−s, and its roots are ¼±i and ¾±i. Thus both reflection and conjugation preserve the zero set while every zero lies off Re(s)=½.
P(s) = s⁴ − 2s³ + (27/8)s² − (19/8)s + 425/256
P(1−s) − P(s) = 0The reproduction script verifies both the symmetry identity and all four roots using exact rational and symbolic arithmetic. This eliminates a symmetry-only proof strategy. It does not refute the Riemann hypothesis: the polynomial does not satisfy the zeta function’s full analytic and arithmetic requirements.
The next obligation is a property that symmetry does not supply.
A spectral approach would need a well-defined self-adjoint operator and a proved correspondence between its real spectrum and every relevant zero. Naming such an operator or observing a resemblance to quantum spectra does not establish that correspondence.
Energy control leaves a scale-dependent quantity uncontrolled.
For a smooth, decaying, divergence-free solution on ℝ³, taking the inner product with u removes the transport and pressure terms by integration by parts. The unforced energy identity is:
(1/2) d‖u‖₂²/dt + ν‖∇u‖₂² = 0.
uλ(x,t) = λu(λx,λ²t), ‖uλ‖p = λ^(1−3/p)‖u‖p.The squared L² norm scales as λ⁻¹, while the maximum speed scales as λ. The L³ norm is unchanged. This is why the energy identity alone does not supply the stronger pointwise or critical-scale bound a global smoothness argument would need. These scaling multipliers are not a computed singular fluid trajectory.
With forcing, the energy identity includes ∫f·u. The September 2026 announced construction uses smooth forcing. A later regularity result assumes real-analytic forcing together with particular core symmetry and anisotropic bounds. Comparing those different hypotheses is essential; the latter result does not by itself refute the smooth-forcing construction or establish general unforced regularity.
Calculation records and reproduction
| Calculation | Verification | Boundary |
|---|---|---|
| 18 sparse spin spectra | Residuals retained; six exact critical comparisons. | Finite systems; approximate eigenpairs. |
| 16 classical Bell assignments | Every deterministic assignment has S=±2. | Local outcomes and setting independence. |
| 91 quantum measurement angles | Matrix expectations agree with 2(cosθ+sinθ). | Specified state and observables. |
| Polynomial reflection and four roots | Exact symbolic identities. | A countermodel to one inference. |
| Nine fluid scaling factors | Norm exponents follow change of variables. | Scaling relations, not solution trajectories. |
Run the script from the project’s scripts directory with Python, NumPy, SciPy, and SymPy available. The output records their versions and writes the full numerical dataset. The quantum experiments use one CPU thread and do not require an external generation service.
