Hopf-fibration spectral explorations (2026-05-11)¶
Origin: User-curiosity dispatch — "fiddle around with Hopf fibration; could be handy for MFO spherical event horizon, also curious whether lifting 2D chess board / Hawaii-Emperor chain to 3D-4D via Hopf-like fibration gives different graph-Laplacian information." Pre-flighted as "might yield nothing."
Verdict in advance: mixed — 2 of 4 experiments yielded nothing (trivial lifts of simply-connected bases), 2 of 4 yielded something real (non-trivial topology = non-trivial spectrum). Honest results, math-doesn't-lie.
Reproduce: python -X utf8 docs/srmech/notes/hopf_fibration_explorations_script.py. Runtime ~10s. Deterministic seed 20260511. numpy + scipy only.
Experiment 1 — Flat 8×8 chess board lifted via Cartesian product with C_8¶
| Quantity | Value |
|---|---|
| Total dim | 512 = 8×8 × 8 |
| Eigenvalue range | [0, 11.70] |
| Factor-sum prediction max dev | 2.31e-14 (machine precision) |
Verdict: yielded nothing new. Cartesian product of a flat (simply-connected) base graph with a cycle is textbook Imrich-Klavžar — eigenvalues are exactly sums of factor eigenvalues. The "lift" adds no spectral information because there's no topology to twist around. Equivalent to standard 3D grid Laplacian.
Experiment 2 — Magnetic twist on TOROIDAL 8×8 (T²) — the non-trivial case¶
Flux sweep φ ∈ [0, 1] on 8×8 torus with Landau-gauge vector potential A_y = (2π·φ·i)/N:
| Flux φ | Eigenvalue range | Max dev from φ=0 |
|---|---|---|
| 0.000 | [0.000, 8.000] | 0.000 (baseline) |
| 0.125 | [0.136, 7.864] | 0.667 |
| 0.250 | [0.158, 7.842] | 0.413 |
| 0.500 | [0.368, 7.633] | 0.540 |
| 0.750 | [0.545, 7.455] | 0.545 |
| 1.000 | [0.709, 7.291] | 0.828 |
Verdict: yielded something real — Hofstadter-style flux-dependent spectrum. The toroidal base has π_1(T²) = Z²; non-trivial U(1) bundles exist and produce distinct spectra per flux value. The eigenvalue range narrows from [0, 8] at φ=0 toward [0.71, 7.29] at φ=1, with intermediate flux values producing intermediate narrowing. This is the discrete-graph analog of the Hofstadter butterfly (Hofstadter 1976) for electrons on a 2D lattice in a perpendicular magnetic field.
Math identity: the twisted Laplacian on T² is a real instance of the §3.5.3(C) motif extended to U(1) gauge connections — the flux parameter labels distinct elements of H¹(T², U(1)) = U(1) × U(1) (Wilson loops around the two cycles), and the spectrum depends on the Wilson-loop holonomy.
Why this matters: it confirms that the "Hopf-like" twist gives new spectral info precisely when the base has non-trivial fundamental group (π_1 ≠ 0). Flat 8×8 grid (Experiment 1) doesn't have this; toroidal 8×8 does. The chess-extension spike (commit 942e0a3) already validated toroidal chess at the §3.5.3(C) level — this exploration adds the U(1) gauge connection layer.
Experiment 3 — Discrete Hopf S³ → S² spectral comparison¶
200 random points sampled on S³; Hopf-projected to S². k-NN graph Laplacian on each:
| Manifold | Spectral gap λ₂ | Top-5 eigenvalues |
|---|---|---|
| S² (Hopf-projected) | 0.508 | [16.64, 16.94, 17.14, 17.46, 17.69] |
| S³ (sampled) | 1.210 | [16.87, 16.93, 17.07, 17.36, 17.56] |
Verdict: yielded something — different spectra at small λ; converging at large λ.
The continuum prediction: Δ_{S²} has eigenvalues l(l+1) for l = 0, 1, 2, ... → {0, 2, 6, 12, 20, ...}; Δ_{S³} has eigenvalues l(l+2) for l = 0, 1, 2, ... → {0, 3, 8, 15, 24, ...}. The S³ gap (λ_min ≠ 0) should be roughly 3 vs S²'s gap of 2 — in continuum.
Discrete approximations don't reproduce exact continuum eigenvalues at N=200, but the structural signal is there: S³ spectral gap (1.21) > S² spectral gap (0.51), matching the continuum ordering. The S³ "extra harmonics" from the S¹ fiber show up as more eigenvalue density per unit interval.
Experiment 4 — Hawaii-Emperor chain (1D path P₅₀) lifted to 2D via Cartesian with C_8¶
| Quantity | Value |
|---|---|
| Chain dim | 50 |
| Product dim | 400 = 50 × 8 |
| Factor-sum prediction max dev | 8.88e-15 (machine precision) |
Verdict: yielded nothing new. Same reason as Experiment 1 — 1D path P₅₀ has trivial topology (π_1 = 0); Cartesian product with C_8 gives sum-of-factor eigenvalues exactly. No information beyond what the factor graphs separately provide.
MFO event-horizon connection — what the Hopf fibration actually offers¶
The MFO §VII.4.1 stance commits to event horizon as a 2D phase boundary (the visible S²) with "no interior" — matter / information bound to the surface. Spherical compression (the §VII.4.1 operator) takes 3D bulk to inscribed 2D boundary.
The Hopf fibration provides exactly the mathematical structure needed to formalize what's lost in compression:
- Base S²: the visible event horizon
- Total space S³: a U(1)-principal bundle over the horizon
- Fiber S¹ = U(1): encodes one extra dimension of "phase information" per base point
- Bundle is non-trivial: total space is genuinely S³, not S² × S¹ (the trivial bundle); the topology twist is the first Chern class = 1
Spectral content of the lift: Δ_{S³} eigenvalues = Δ_{S²} eigenvalues + S¹-fiber harmonics. The continuum gap l(l+2) - l(l+1) = l (linear in l) is the extra spectral degree of freedom per S² mode that the bulk-to-boundary compression discards.
Holographic-principle reading: the Hopf bundle's S¹ fiber over the visible S² horizon IS the spectral incarnation of "the bulk information is encoded on the boundary with a phase degree of freedom per mode." This is consistent with — and a finite-dimensional spectral analog of — Maldacena's AdS/CFT correspondence where boundary phase data encodes bulk dynamics.
MFO §VII.4.1 candidate addition (conductor decision, not unilateral): note that the Hopf fibration provides the discrete spectral framework for the "spherical compression as 3D-bulk → 2D-boundary operator" stance: Δ_{boundary} + S¹-fiber harmonics = Δ_{lifted bulk}. The compression operator literally projects out the fiber-harmonic series.
Honest summary¶
What yielded NOTHING new (textbook math, expected):
- Flat 8×8 chess board × C_8: factor-sum exact at 2.31e-14 max dev. No new info.
- Hawaii chain P₅₀ × C_8: factor-sum exact at 8.88e-15 max dev. No new info.
What yielded SOMETHING real (non-trivial structure surfaces in the math):
- Toroidal 8×8 + magnetic flux: Hofstadter-style flux-dependent spectrum. Eigenvalue range narrows with flux. Genuine discrete U(1)-gauge instance of the chess §3.5.3(C) toroidal sub-instance, extended with a connection.
- Discrete Hopf S³ → S²: S³ spectral gap (1.21) > S² spectral gap (0.51), matching continuum
l(l+2) > l(l+1)ordering. The S¹ fiber adds spectral degrees of freedom that the base S² doesn't have.
MFO connection (load-bearing finding): the Hopf bundle's spectral decomposition (Δ_{S³} = Δ_{S²} + fiber harmonics) is the natural mathematical framework for what "spherical compression" of 3D bulk to 2D event-horizon boundary discards. Conductor-decision-worthy candidate for §VII.4.1 addendum.
Files¶
hopf_fibration_explorations_script.py— reproducible script (~10s, seed 20260511, numpy+scipy only)hopf-fibration-explorations-2026-05-11.md— these findings
Cite Hofstadter 1976 for the magnetic-lattice spectrum result; Hopf 1931 for the original fibration; continuum spherical-harmonic eigenvalues are textbook (Stein-Weiss, Sogge, etc.).