Skip to content
AptifoldGet in touch

Aptifold Research · Derivation and replay

Verify the
scattering result

Aptifold’s AI research agent derived two exact local solutions in this Schwarzschild scattering channel. Follow its proof or rerun the checks against the saved inputs.

Fix the channel and normalization

Two parent waves share angular mode (ℓ, ℓ) and positive frequency ω. The daughter has mode (2ℓ, 2ℓ) and frequency 2ω. Set G = c = M = 1, take k = iω, and use the time dependence e−iωt. Thus w = Mω is numerically equal to ω below. The Schwarzschild exterior has r > 2 and x = r + 2 log(r/2 − 1).

Let E and O be the even and odd linear masters, J = iO, and Dx = (1 − 2/r)∂r. The transformed even solution supplies a common reference odd parent. These definitions match the physical boundary problem in Bucciotti et al. (2026).

F = 1 − 2/r
λⱼ = (j − 1)(j + 2)/2
γⱼ = (j − 1)j(j + 1)(j + 2)/12
Wⱼ = γⱼ + 3F/[r(λⱼr + 3)]

g = γℓ,   Γ = γ₂ℓ/2
Jref = (Wℓ − Dₓ)E/(g − k)
J = ρ Jref,   p = O_in/E_in = −iρ(g + k)/(g − k)

The two ratios are

ρ₁ = 1
ρ₂ = [(Γ + g)/(Γ − g)] [(g − k)/(g + k)]

p₁ = −i(g + k)/(g − k)
p₂ = −i(Γ + g)/(Γ − g) = −i(9ℓ² + ℓ − 4)/(7ℓ² − ℓ)

The theorem concerns rational local quadratic particular solutions of one projected daughter equation for integer ℓ ≥ 2 and positive real ω. The separate physical outgoing strain also requires the ingoing future-horizon solution and zero freely specified incoming daughter strain at infinity.

Construct both solutions

Use the projected daughter (W2ℓ − Dx)E2/(2Γ − 2k) − J2. Its odd wave operator is Dx2 − 4k² − V2ℓ,O, with Vj,O = F[j(j + 1)/r² − 6/r³].

Set J = Jref when defining the jet basis, and set Z = r²DxJ. The linear parent equation reduces higher radial derivatives to J and Z. Consequently the unknown is a vector (a,b,c) of rational functions multiplying J², JZ and Z².

A = r²(k² + Vℓ,O),   B = 2F/r,   C = 1/r²
DₓJ = C Z,   DₓZ = A J + B Z

T(a,b,c) = (Fa′ + Ab,
            Fb′ + 2Ca + Bb + 2Ac,
            Fc′ + Cb + 2Bc)

L(a,b,c) = T(T(a,b,c)) − (4k² + V₂ℓ,O)(a,b,c)

Expand the quadratic Einstein source into the even–even, mixed, and odd–odd coefficients SEE, SEO, and SOO. The mixed coefficient includes both orderings. The archived odd source was built by reconstructing the pure-odd metric, taking the quadratic curvature, and projecting the (2ℓ, 2ℓ) harmonic. The replay begins with those projected coefficients in data/assembly.json.

S(ρ) = S_EE + ρ S_EO + ρ² S_OO
L(P₁) = S(1)
L(PΔ) = S_EO + (1 + ρ₂) S_OO

S(ρ) = L[P₁ + (ρ − 1)PΔ] + (ρ − 1)(ρ − ρ₂) S_OO
P₂ = P₁ + (ρ₂ − 1)PΔ

The difference identity constructs the second solution. It also makes the remaining question exact. If the wave operator cannot produce SOO from a rational local vector, the last coefficient must vanish. Pure odd incidence has the same obstruction.

Completeness

Why there is no third rational solution

A pole count turns the question into a finite matrix. A row vector that sends each legal operator column to zero and the odd source to a nonzero value certifies that the source lies outside the operator’s image.

Origin, r = 0

The source pole order is at most ten. The leading factor is (2m + a − 2)(2m + a + 6), with a = 4, −4, −12. The exceptional positive m are 3 and 7. The solution bound is m ≤ 7.

Horizon and r = −3/λ₂ℓ

A pole of order m at the horizon shifts the imaginary wave indices by −m/2. The source gives an upper bound of one. The other point can have a simple pole with a forced coefficient. Remove both forced terms before collecting Laurent powers.

Infinity and other radii

At other finite radii a pole would raise its order by two and create a pole absent from the source. The two leading kernel directions at infinity have a nonzero next term 2d(±2k) rd−1 for d > 4. The source grows at most as r³ in the unscaled jet basis.

Apply L to each allowed jet and power of r to form the 30 columns of A. Read the coefficients of the remaining odd source into s using the same 41 equations. The exact vector z satisfies zA = 0 and zs ≠ 0 throughout the stated domain. The nonvanishing check uses signs after shifting r and ℓ, fixed-label Sturm counts for positive real ω, and the absence of integer roots of nonlinear irreducible factors in ℓ. The download includes the witness and rebuilds the system.

Run a column check in your browser

These six examples use the saved matrix specialized at exact ℓ and w, then reduced modulo the prime 65,537. In this field 256² = −1, so k = iω has an exact representative. Choose a column to see the actual products being summed. The symbolic script checks the unrestricted rational functions across the whole domain.

30 / 30 operator projections are zero

38228 odd-source projection · mod 65,537

1–30 are operator columns. S is the odd source.

Odd-source column · Sum of 41 products = 38228 modulo 65,537

See the 41 terms in this sum
RowEquationWitness weightEntryProduct mod p
1Jet 1, r^31667829361661
2Jet 1, r^2628082175961748
3Jet 1, r^126241534322020
4Jet 1, r^00648810
5Jet 1, r^-10539730
6Jet 1, r^-2345044208359597
7Jet 1, r^-3614105897035228
8Jet 1, r^-4476306366614590
9Jet 1, r^-523937238429758
10Jet 1, r^-6625863489160823
11Jet 2, r^0252623873615885
12Jet 2, r^-158985326114537
13Jet 2, r^-240866502811785
14Jet 2, r^-34474838227559
15Jet 2, r^-456915275084427
16Jet 2, r^-5165034442665396
17Jet 2, r^-624071245631183
18Jet 3, r^-1606805908130106
19Jet 3, r^-271804923822262
20Jet 3, r^-3610894116908
21Jet 3, r^-4582825629942876
22Jet 3, r^-562900115561533
23Jet 3, r^-636475621565179
24Jet 1, r^-765000
25Jet 1, r^-9000
26Jet 1, r^-10147100
27Jet 1, r^-112916500
28Jet 2, r^-105498000
29Jet 2, r^-115352200
30Jet 3, r^-11000
31Jet 1, r^-8000
32Jet 2, r^-9000
33Jet 3, r^-104491100
34Jet 2, r^-8000
35Jet 3, r^-9000
36Jet 2, r^-75051600
37Jet 3, r^-85489500
38Jet 3, r^-75502400
39Jet 1, r^4000
40Jet 2, r^15824400
41Jet 3, r^0246600

Recover the outgoing strain

Define Q⁻ as the outgoing second-order spin-minus strain divided by the square of the first-order incoming spin-minus strain, with orthonormal harmonics. RE is the even parent’s horizon-ingoing reflection at ω. RO is the odd daughter’s horizon-ingoing reflection at 2ω. Both use the same tortoise origin.

For the second mixture the incoming spin-minus amplitude relative to the even incident amplitude is 2Γ/(Γ − g). The reconstructed metric satisfies ten projected Einstein equations in the saved extraction. The quadratic asymptotic coordinate change leaves a direct term with RE² and gives a zero mixed phase. Cancel the physical incoming daughter with the allowed homogeneous solution to obtain the RO term. The three phase records and the normalized coefficients are included under data/physical-response.json.

Q⁻₁ = [Kℓ k² / (4g²(k − Γ))] Rₒ

Q⁻₂ = [Kℓ / (4Γ)] {[k(k − g²/Γ)/(k + g)²] Rₑ²
                    + [k/(k − Γ)] Rₒ}

Kℓ = (2ℓ+1)!(20ℓ³+35ℓ²+ℓ−8) / [144(ℓ−1)!(ℓ−2)!]
     × √{ (2ℓ+2)!(2ℓ−2)! / [π(4ℓ+1)!] }

For ℓ = 2 at w = 3/8, the two local branches give outgoing magnitudes about 0.00480 and 0.00155. The published scattering calculation has parity ratios with zero outgoing amplitude. Both statements can be checked in the same physical normalization.

Threshold prediction

Which reflection term wins?

The two reflection phases enter the second mixture with weights that nearly cancel. The small difference fixes the first frequency correction to the normalized outgoing magnitude.

For a positive leg frequency u, the real elastic Compton projection gives the quadratic reflection phase πu²bⱼ after the common low-frequency convention is fixed. The parent and daughter combine as bℓ − 2b₂ℓ in the real cubic coefficient.

L = j(j + 1)
bⱼ = (15L² + 13L + 24) / [(2j + 1)L(4L − 3)]
Bᵣ = πKℓ (bℓ − 2b₂ℓ) / Γ²

Let Hⱼ be the jth harmonic number. The leading signed coefficient C₂ supplies the normalization and the minus sign. The first derivative of log(|Q⁻₂|/w²) at zero is Bᵣ/C₂.

Dℓ = H₂ℓ − Hℓ + 2/[ℓ(ℓ+1)]
     − 2/[(2ℓ)(2ℓ+1)] − log 2
Aℓ = (1/g − 1/Γ)² − 8Dℓ/Γ
C₂ = −Kℓ Aℓ/(4Γ)
Bᵣ/C₂ = −4π(bℓ − 2b₂ℓ)/(Γ Aℓ)

Explore the phase subtraction

Compare ℓ = 4 and ℓ = 5. For each label use its own incident ratio p₂, which is fixed as frequency varies. The parent and daughter terms remain close, but their order reverses. The chart recalculates both terms and the slope from the equations above.

Parent term 0.45339105

Twice the daughter term 0.45132439

Parent minus daughter +0.00206666

Analytic slope −0.414614

Direct lower-window slope −0.415010

Normalized magnitude slope

ℓ = 2−0.771540
ℓ = 3−1.372800
ℓ = 4−0.414614
ℓ = 5+0.396163
ℓ = 6+0.495483

The direct radial result encloses the real and imaginary strain at w = 1/2048, 1/1024 and 1/512. Divide the complex magnitude by w², then compare logarithmic secants. At the lower interval they are −0.415010 and +0.396542 at ℓ = 4 and 5. Comparing the two cuts and halving the frequency tests the direction of the approach. Inspect the computed slopes and enclosures.

For ℓ = 3, the phase-removed real cubic coefficient is +0.00127315440960. The separate loop plot integrates a clipped angular region. Its positive result persists over the tested grids. Resolving the omitted region and the combined resolution error remains open. Inspect the loop grids and cancellations.

Reproduce the checks locally

The archive includes runnable code, the projected symbolic sources, the matrix certificate and the numerical records behind the graphs. Use the included checksums to identify the data you ran.

Download code and data · ZIP

Replay the numerical evidence

Extract the archive, then run this from the extracted directory with Python 3.12. It recomputes the threshold slopes, both cutoff comparisons, both parent grids, and the 17 substitutions.

python -B replay.py

Verify the exact certificate

With python-flint 0.9.0 available, the symbolic verifier checks both projected identities, all matrix columns, and nonvanishing in the declared domain. It rebuilds the system from the saved source coefficients.

python -B prove.py

The download starts with projected sources and completed radial and loop integrations. Its outputs let a reviewer check the proof and aggregation from those inputs. An independent group can use the conventions above and the cited equations to reconstruct the original source and integrations.

Compare the prior results

Equations 19 and 39–40 fix the outgoing observable and parity convention. Figure 6 shows two quadrupole mixtures where the outgoing amplitude vanishes. The criterion here is the existence of a rational local particular solution. At ℓ = 2 and w = 3/8, our two mixtures give outgoing magnitudes of about 0.00480 and 0.00155. Bucciotti, Redondo-Yuste, Zhong, Kuntz and Cardoso

Equations 16–18 give a general reduction to master integrals. Applying that weighted-derivative method to the projected source recovers both quadrupole ratios. It is a direct precursor to the classification. The pole bounds and nonzero obstruction here establish completeness for integer ℓ ≥ 2 and positive real ω in the rational local class. Chen and Xie

The known linear phase shifts and elastic amplitudes supply inputs to the second response. The outgoing slope here comes from combining the parent and daughter phase coefficients at different frequencies in the quadratic strain. The archived exterior integrals check the predicted signs. Dolan · Bjerrum-Bohr, Chen, Eriksen and Shah, Appendix A.1

Exact quadratic profiles also exist in the algebraically special sector. Numerical work on Kerr tracks the second harmonic of quadrupolar driving as spin varies. These give related benchmarks in distinct regimes. Ben Achour and Roussille · Quadratic scattering by Kerr black holes

The 19 November 2026 literature check found no explicit proof of the full two-mixture classification or report of the opposite low-frequency slopes at ℓ = 4 and 5 in these sources. Historical priority remains open. An outside reader can test whether a previous reduction already implies either result by comparing the channel, the local solution criterion and the normalized strain derivative.