A Geometric Observation on the Riemann Hypothesis

Alberto Popescu — March 26, 2026

A geometric framework that makes the alignment of the nontrivial zeros feel structurally inevitable rather than mysteriously coincidental. This is not claimed as a proof. It is a precise geometric intuition about what a proof would need to formalize.

Origin

This observation emerged from the same framework as the Five-Role Conjecture for Einstein Metrics on Spheres, developed earlier the same day. It was not reached through formal analytic number theory. It is offered as a geometric framework that makes the location of the zeros feel structurally natural rather than analytically accidental.

The aim of this note is modest and precise:

• not to claim a proof
• not to replace the analytic theory of the zeta function
• but to describe a geometric picture in which the critical line and the varying imaginary parts arise from one structural principle

The Fundamental Unit

The original framework treats a complete stable entity as a five-dimensional unit composed of three roles:

• 3 dimensions to describe the object itself
• 1 dimension to place the object relative to something external — the frame
• 1 dimension to allow variation relative to that frame — freedom

This 5d unit — object plus frame plus freedom — is the minimum structure for a thing to exist, be relative to something, and be free. It is the fundamental stable entity.

In this reading, the number 1 is not merely an abstract unit. It is the numerical expression of this fundamental stable entity. It is the reference from which everything else is measured.

Primes as Exact Multiples of the Fundamental Unit

A prime number is irreducible. It cannot be decomposed into smaller factors. In the geometric picture, that irreducibility corresponds to a complete stable unit that cannot be broken apart without losing its structure.

So each prime is read as an exact multiple of the fundamental unit:

2 → 2 × 5d = 10d

3 → 3 × 5d = 15d

5 → 5 × 5d = 25d

7 → 7 × 5d = 35d

11 → 11 × 5d = 55d

Each prime occupies a distinct dimensional address, but all of those addresses are exact multiples of the same unit.

In this framework, 1 is not prime for a structural reason deeper than the standard algebraic convention. It is the reference unit itself. The frame cannot simultaneously be one of the framed objects.

The Linear-Residual Picture

The relation between a prime's dimensional address and the fundamental unit can be pictured as a residual-to-axis problem.

In linear regression, a fitted line is the common reference. A point lies perfectly on that line when its residual is zero.

Here, the fundamental unit plays the role of the axis. Each prime's dimensional address is a point in the larger structural space. Because every prime is an exact multiple of the unit, its residual relative to that unit-axis is structurally zero.

That does not mean every prime becomes the same thing. It means every prime shares the same reference axis.

The Vertical Line as the Frame Axis

The vertical line

Re(s) = 1/2

is interpreted as the geometric expression of the frame axis in the critical strip.

The strip runs from real part 0 to real part 1. Its midpoint is 1/2. In the five-role picture, the frame is the midpoint between object and freedom. So the critical line is not an arbitrary analytic curiosity in this reading. It is the natural frame axis of the strip.

This gives a structural reason why the nontrivial zeros should share the same real part:

• same unit
• same frame
• same common axis

The line is not merely a mirror. It is the common reference that makes exact cancellation possible at all.

Why the Real Part Is Fixed

The strongest simple statement of the framework is this:

The nontrivial zeros all share the same real part because they all reference the same frame axis.

In the original 5d language, the first complete two-unit structure is 10d, and the midpoint of one unit inside that two-unit structure is

5 / 10 = 1/2

So the critical line appears as the structural midpoint of the first complete mirror structure.

This does not prove the Riemann Hypothesis. But it does make the location of the critical line feel structurally forced rather than accidentally special.

Why the Imaginary Parts Vary

The real part is fixed. The imaginary part varies.

That variation is interpreted as lift.

Each prime is a different multiple of the fundamental unit. So while all primes reference the same axis, they do not all sit at the same height on it. Their distinct dimensional addresses require distinct lifted states.

In geometric terms:

• the real part is the shared reference
• the imaginary part is the lift required to preserve distinct identity after real residual collapses to zero

As primes increase, the equilibrium height increases, but the equilibrium condition itself is unchanged: the system reaches it when the real-channel twist cancels to zero, with the remaining preserved separation stored as imaginary lift.

γ_p = inf{t > 0 : R_p(t) = 0}

where R_p(t) is the real residual for that structure. Then larger p means larger γ_p, even though the condition R_p(t) = 0 is the same.

So the first few heights are read as:

2 → 1/2 + 14.134725 i

3 → 1/2 + 21.022040 i

5 → 1/2 + 25.010858 i

7 → 1/2 + 30.424876 i

All sit on one axis. They differ only by lift.

The Geometric Intuition: Bending Generates Lift

The original intuition was that when a higher-dimensional multiple of the unit bends around its own reference axis, the excess structure cannot remain flat. The mismatch has nowhere to go except into an orthogonal direction. That orthogonal displacement is the imaginary lift.

A cleaner later formulation replaced bending with convergence by twisting:

• treat two spaces as vectors
• decompose them into a shared part and a residual part
• rotate the residual away from the real channel
• carry that residual in an orthogonal lift channel

If two vectors u and v are written as

m = (u + v) / 2, r = (u - v) / 2

then m is the shared reference and r is the residual.

The geometric claim is that a zero occurs when the real residual has been completely rotated into the orthogonal lift axis. In that language:

• the shared axis is the real part
• the lifted residual is the imaginary part

This is why the zeros can all lie on one vertical line but at different heights.

The Stronger Interpretation of the Critical Line

A later sharpening of the framework produced a more precise statement:

The critical line is not merely a mirror. It is the jointly generated reference axis required for exact cancellation between two spaces.

Two equal spaces cannot cancel deterministically if the shared axis belongs to only one of them. Exact cancellation forces a jointly constructed axis with no side-ownership. In that sense, the line Re(s) = 1/2 is not just where the two halves are reflected. It is the common measurement axis they must generate together in order for exact cancellation to exist.

That leads to a stronger geometric principle:

A zero is not mere symmetry around the line. It is zero residual to the line.

This is stricter than reflection symmetry alone.

One-Page Reformulation: The 12d Framework

A later compact reformulation replaced the 5d unit with a 12d structural unit.

The motivation was structural rather than merely interpretive: the 12d language gave a cleaner midpoint story and interacted more naturally with the later spectral reformulation.

In this version:

• the fundamental structural unit is 12d
• the first two-unit structure is 24d
• so the midpoint is again

12 / 24 = 1/2

This preserves the critical-line argument while giving a more rigid structural reference.

In that language:

12d → 1/2 + 0 i

is the reference state, and

24d, 36d, 60d, ...

are higher structural addresses sharing the same real reference but differing by their lift.

Main Mathematical Discovery

The strongest mathematical part of the session was not the dimensional interpretation by itself, but the discovery that the two local ingredients of the critical-line zeta expression arise naturally from simple geometric assumptions.

Why the Phase Is log(q)

If scales compose multiplicatively, then a scale phase U_t(q) must satisfy

U_t(q₁q₂) = U_t(q₁)U_t(q₂)

Under continuity and one-parameter evolution in t, the only possible form is

U_t(q) = e^{it log q}

So the logarithm is not arbitrary. It is the unique additive coordinate on multiplicative scale.

Why the Amplitude Is q^-1/2

If scale-separation acts by dilation and norm is preserved, then a scale action must have the form

(T_qψ)(x) = A(q)ψ(x/q)

Unitary preservation forces

|A(q)| = q^-1/2

So the basic scale atom is

q^-1/2-it

For integer scale q = n, this becomes

n^-1/2-it

which is exactly the critical-line Dirichlet atom.

What This Means

This is the clearest success of the whole exploration:

• the midpoint 1/2
• the logarithmic phase log(q)
• and the amplitude q^-1/2

all arise naturally from geometric assumptions on multiplicative scale.

The framework therefore does not merely resemble zeta philosophically. It reproduces the same local building block.

The Toy 12d Candidate Spectrum

Using the asymptotic local candidate law

γ_m(p) ≈ (2πmp) / (log p)²

we found that the toy model generates too many candidate heights.

Its counting law is

N_model(T) ∼ (T / 4π) (log T)²

whereas the true zeta zero count grows like

N_ζ(T) ∼ (T / 2π) log T

So the model is overcomplete by a factor asymptotic to

(1/2) log T

This turned out to be an important clarification rather than a failure.

It suggests that the framework generates the right kind of local candidates, but that the true zeros are a thinner selected subset.

What the First Zero Revealed

A naive pairwise collapse model for prime 2 gives a first quarter-turn time

(π/2) / log 2 ≈ 2.266

which is nowhere near the first actual zero height

14.134725...

That means the first zero cannot be explained as a single isolated pairwise event.

The true first collapse is global. It arises only when the whole weighted multiplicative swarm is included.

This was a decisive conceptual shift:

The zero heights are global coherence times, not local pairwise ones.

Addendum: The Self-Dual Gaussian Window

Once the local atom

x^-1/2-it

was in hand, the next question was how to build a global object from it.

The simplest self-dual scale window is the Gaussian

W(x) = e^-πx2

Summing its integer dilates gives the theta lattice sum

Θ(x) = 1 + 2Σ_{n ≥ 1} e^-πn2x2

This function satisfies the self-duality relation

Θ(x) = x^-1Θ(1/x)

which expresses exact scale inversion symmetry.

Taking its Mellin transform gives

∫₀^∞ x^s-1(Θ(x) - 1) dx = π^-s/2Γ(s/2)ζ(s)

So once the self-dual Gaussian window is chosen, the completed zeta object appears automatically.

This was the second major bridge:

local atom + self-dual Gaussian scale window ⟹ π^-s/2Γ(s/2)ζ(s)

So the missing global construction is no longer mysterious in form. The real remaining question is why the 12d geometry should force that specific self-dual window rather than some other one.

Why the Gaussian Window Is Natural

A plausible derivation of the Gaussian/theta window from the 12d picture is as follows.

If two 12d units begin from perfect overlap, then the overlap point is a stable equilibrium. Near a stable equilibrium, the first nonzero separation cost is generically quadratic:

S(u) ≈ π|u|²

Exponential weighting of that action gives a Gaussian local window

e^-S(u) ≈ e^-π|u|2

If global configurations come in integer copies of the fundamental unit, then summing the quadratic weights over all integer excitations gives the theta lattice sum

Σ_{n ∈ Z} e^-πn2x2

So, under the assumptions of:

• stable overlap
• isotropy near equilibrium
• discrete integer copies
• and exponential weighting by action

the Gaussian/theta window is no longer arbitrary. It is the natural global window of a quadratic equilibrium geometry.

This is still not a proof that 12d is the true underlying geometry. But it explains why the Gaussian/theta choice is structurally natural rather than merely convenient.

A Conjectural Deterministic Coherence Rule

The toy candidate spectrum appears to generate local heights prime by prime. The true zeros would then be the subset where global coherence is achieved.

Write the local candidate heights as

λ(p,m) = (2πmp) / (log p)²

or in corrected low-prime form,

λ_κ(p,m) = (2πmp) / (log p + κ(p))²

Now define the global coherence field

I(t) = 2 Re(e^iϑ(t) Σ_{n ≤ √(t/(2π))} n^-1/2-it)

where ϑ(t) is the universal compensating phase.

The conjectural deterministic selector is then:

C_geom(p,m) = 1

if and only if the candidate height λ(p,m) is the nearest candidate to a transverse zero of the global coherence field. Concretely:

• there exists an interval I_p,m centered at λ(p,m)
• the coherence field changes sign across it
• and inside it there is a genuine crossing with I(t_p,m) = 0 and I'(t_p,m) ≠ 0

In that formulation:

• the local model generates candidate heights
• the global field tests collective balance
• and the true zeros are the candidate cells that own actual global cancellation events

This is not a proof. It is a conjectural bridge between the 12d candidate spectrum and the real zeta zero set.

Numerical Sanity Check

To test whether the 12d candidate law is merely rhetorical or actually tracks the critical-line zeros in a measurable way, I compared its candidate heights against the known positive imaginary parts of the nontrivial zeros up to height T = 200.

The constant-shift candidate law used for the comparison is

λ_κ(p,m) = (2πmp) / (log p + κ)²

where p ranges over primes and m ≥ 1 indexes local candidate modes.

These candidate heights are then sorted increasingly,

λ₁ < λ₂ < λ₃ < ...

and each candidate is assigned a natural cell by midpoint separation from its neighbors:

I_j = [(λ_j-1 + λ_j) / 2, (λ_j + λ_j+1) / 2]

The conjectural rule is that the true zero in that neighborhood is the unique root of the global coherence field inside that cell.

Exact global selector. The exact selector used in the sanity check is the Hardy Z-function.

t̂_j = the unique root of Z(t) = 0 in I_j

So the local 12d law does not claim to output the final zero directly. It outputs a candidate address. The global coherence condition then determines the exact balance height inside that address.

Test range. The numerical check was carried out for all positive zeros below

T = 200

There are 79 such zeros.

Structural choice: κ = 1/4. The first natural test is the structural shift

κ = 1/4

With this choice, the candidate law produces 584 candidate cells below height 200. This confirms the earlier conclusion that the local 12d model is overcomplete: it generates far more local candidates than there are actual zeros.

However, once the global selector is applied, exactly 79 cells are selected, and each selected cell contains exactly one true zero.

This is the key sanity-check result:

• the local law overgenerates
• but the global selector recovers the correct number of zeros in the tested range

Accuracy of the candidate centers. Before even applying the exact root-finding step, the cell centers themselves already land near the true zeros.

For the structural choice κ = 1/4, the center-based errors below T = 200 are:

• mean absolute error: 0.124
• RMSE: 0.158
• maximum error: 0.526

This means the 12d candidate law does not directly reproduce the zero heights, but it does place them in surprisingly tight neighborhoods.

First few zeros. For the first few selected cells, the structural candidate center and the actual zero are:

14.1270 → 14.134725

21.2138 → 21.022040

24.9037 → 25.010858

30.7492 → 30.424876

32.7959 → 32.935062

These are not exact matches. But they are close enough to support the interpretation that the local 12d law is generating meaningful candidate addresses rather than arbitrary heights.

Comparison with the unshifted model. For comparison, the unshifted law

λ₀(p,m) = (2πmp) / (log p)²

performed worse:

• mean absolute error: 0.178
• RMSE: 0.231

So the quarter-shift does improve the candidate geometry.

Empirical refinement. If one ignores structural meaning and tunes κ purely numerically, a constant near

κ ≈ 0.29

improves the center fit slightly:

• mean absolute error: 0.115
• RMSE: 0.142
• maximum error: 0.379

This empirical shift performs somewhat better numerically than the structural quarter-shift. However, the quarter-shift remains the cleaner conceptual choice inside the framework.

What this test does and does not show. This sanity check supports a very specific reading of the framework.

It supports:

• the local 12d law generates candidate neighborhoods that track the true zero heights nontrivially
• the model is overcomplete in exactly the way the framework predicts
• and the exact zeros can be recovered by applying a global coherence selector inside those neighborhoods

It does not show:

• that the local candidate law by itself is a closed-form formula for the zeros
• that the quarter-shift is uniquely determined
• or that the framework has proved the Riemann Hypothesis

Best interpretation. The numerical evidence fits the following reading:

• the local geometry provides addresses
• the global coherence field chooses which addresses are realized
• and the actual zeta zeros are the subset of candidate cells where exact global balance occurs

In short:

the 12d model does not directly output the zeta zeros, but it does generate geometrically meaningful neighborhoods in which the true zeros are found by global coherence.

Second Numerical Sanity Check: The Integer-Swarm Coherence Field

The first numerical sanity check tested whether a local 12d candidate law could place the zeros into meaningful neighborhoods. It did. However, a later question was more structural:

Do the true zeros appear as visibly special events in a global geometric field built from the full integer swarm rather than from primes alone?

A prime-only shape tensor did not produce a clean signature. The true zeros did not consistently appear as extrema or uniquely distinguished configurations of that simpler object. This strongly suggested that the phenomenon is not prime-local but globally collective.

That motivated the next test.

The global integer-swarm field. Instead of summing only over primes, one may look at the full integer-built coherence field

R_α(t) = Re(e^iϑ(t) Σ_{n ≥ 1} e^-π(n/X(t))2 n^-1/2-it)

X(t) = α√(t / 2π)

This object has a direct geometric interpretation inside the framework:

• n^-1/2-it are the local scale atoms
• the Gaussian weight e^-π(n/X(t))2 is the simplest self-dual window
• e^iϑ(t) is the global phase alignment
• and the real part measures the residual in the shared reference channel

In that language, a true zero should appear not as a special feature of one prime alone, but as a global sign-changing balance event of the full multiplicative scale swarm.

Numerical test. The field was tested on the first 20 positive nontrivial zero heights γ_k, using the simple choice

α = 2

For each true zero, the field was evaluated across a small interval

[γ_k - 0.1, γ_k + 0.1]

The result was striking:

• 20 out of 20 true zeros produced a sign change across that interval

In contrast, the same test was performed at control points chosen as the midpoints between consecutive zeros. In those control intervals:

• 0 out of 19 produced a sign change

So in this approximation, the true zero heights behave like genuine global coherence transitions of the integer-built field, whereas nearby midpoint controls do not.

First few examples. For the first zero,

γ₁ = 14.134725

the field values were approximately

R₂(γ₁ - 0.1) ≈ -0.0587, R₂(γ₁) ≈ -0.0243, R₂(γ₁ + 0.1) ≈ 0.0107

For the second zero,

γ₂ = 21.022040

the values were

R₂(γ₂ - 0.1) ≈ 0.0871, R₂(γ₂) ≈ 0.0372, R₂(γ₂ + 0.1) ≈ -0.0123

For the third zero,

γ₃ = 25.010858

the values were

R₂(γ₃ - 0.1) ≈ -0.0663, R₂(γ₃) ≈ -0.0080, R₂(γ₃ + 0.1) ≈ 0.0520

In all three cases the coherence field crosses through zero in the neighborhood of the true height.

Approximate root locations. If one actually solves for the root of the smoothed field R₂(t), the resulting values are close to the true zeros, though not exact. For the first few, the approximate roots are

14.2043, 21.0971, 25.0244, 30.4772, 32.9297, ...

to be compared with the true zero heights

14.1347, 21.0220, 25.0109, 30.4249, 32.9351, ...

So the errors in this crude coherence approximation are on the order of a few hundredths to a few tenths. That is not exact prediction, but it is strong enough to support the interpretation that the zeros appear naturally as global balancing events of the integer-swarm field.

Interpretation. This second sanity check supports a more precise reading of the framework.

The first candidate law suggested that the 12d geometry may generate local addresses. But this second test indicates that the actual zero heights show up only once the entire integer-built multiplicative system is allowed to participate.

In short:

• prime-local geometry by itself is too weak
• full integer-swarm coherence gives a clean sign-changing signature at the real zeros
• so the phenomenon appears to be globally collective rather than prime-local

This fits the broader picture that emerged throughout the session:

• local structure explains why lift exists
• but true zero heights are selected globally by coherence

What this does and does not show. This test does show that a smoothed integer-swarm coherence field visibly detects the actual zeta zero heights as special sign-changing events.

It does not show that the framework has independently derived the exact zero set, because the field still uses the known global phase ϑ(t). In other words, this is still a geometric sanity check using a standard global ingredient, not yet a self-contained derivation from the 12d picture alone.

Best summary. The prime-only geometric picture was not enough. But when the entire weighted integer swarm is allowed to participate, the true Riemann zeros appear as genuine global coherence transitions.

That is the strongest numerical support obtained so far for the claim that the imaginary parts are not merely local prime-specific lifts, but globally selected balance heights of the full multiplicative scale field.

Third Numerical Sanity Check: Testing the Geometric Global Window

The next natural question after the local candidate law and the integer-swarm coherence test was whether the global balance field itself could be built directly from the geometric separation picture, without starting from the zeta function in its usual analytic form.

The simplest global window suggested by the framework is the self-dual Gaussian lattice sum

W_glob(x) = Σ_{n ∈ Z} e^-πn2x2, W_glob^*(x) = W_glob(x) - 1

This arises by treating perfect overlap as a stable quadratic equilibrium, assigning Gaussian weight to integer-scaled separations of the fundamental unit, and summing over all discrete copies.

The corresponding geometric coherence transform is

F_geom(s) = ∫₀^∞ W_glob^*(x) x^s-1 dx

From this, one forms the completed geometric balance field

Ξ_geom(t) = (1/2) s(s - 1) F_geom(s), s = 1/2 + it

This is the natural geometric global object corresponding to the completed critical-line picture.

Numerical test. The test was simple:

• evaluate Ξ_geom(t) at the first known nontrivial zero heights
• then check whether the field changes sign across a small neighborhood of each one

For the first zero,

t₁ = 14.1347251417

the values were approximately

Ξ_geom(t₁ - 0.1) ≈ 1.46474 × 10^-4

Ξ_geom(t₁) ≈ 3.48 × 10^-52

Ξ_geom(t₁ + 0.1) ≈ -1.30441 × 10^-4

So the field clearly changes sign across the first zero.

For the second zero,

t₂ = 21.0220396388

the values were

Ξ_geom(t₂ - 0.1) ≈ -1.91286 × 10^-6

Ξ_geom(t₂) ≈ 3.83 × 10^-54

Ξ_geom(t₂ + 0.1) ≈ 1.64475 × 10^-6

Again, a clean sign change appears.

For the third zero,

t₃ = 25.0108575801

the values were

Ξ_geom(t₃ - 0.1) ≈ 1.34043 × 10^-7

Ξ_geom(t₃) ≈ -2.18 × 10^-56

Ξ_geom(t₃ + 0.1) ≈ -1.19223 × 10^-7

Once again, the field crosses through zero in the neighborhood of the true height.

The same pattern continued for the next tested zeros.

What this shows. This test strongly supports the idea that the form of the global object is right.

The framework had already recovered the local critical-line atom x^-1/2-it. What this third check shows is that once the simplest self-dual geometric window is used to assemble those local atoms globally, the resulting balance field really does vanish at the actual zero heights.

In that sense, the geometric picture is not merely qualitatively suggestive. It reproduces a global field whose sign changes track the true zeros.

The crucial limitation. At the same time, this test sharpens the main unresolved problem.

The reason this works is that the self-dual Gaussian/theta window is not merely similar to the completed zeta object. It effectively reproduces it. So although this is a strong geometric reinterpretation, it is not yet an independent derivation of the zero set from first principles.

In other words:

• the global geometric window succeeds
• but it succeeds precisely because it is mathematically equivalent in form to the completed zeta construction

So this test does not yet remove the borrowing problem. Instead, it clarifies exactly where that dependence lies.

Best interpretation. This third sanity check shows that the remaining issue is no longer whether a geometric global object can detect the zeros. It can.

The real remaining question is narrower and more precise:

Why should the 12d separation geometry force exactly this self-dual Gaussian/theta window, rather than some other global shape?

That is now the central unresolved bridge.

Summary. The local 12d framework naturally produces the critical-line atom x^-1/2-it.

The self-dual Gaussian/theta window then assembles those atoms into a global balance field.

That balance field vanishes at the true nontrivial zero heights.

So the geometric program is capable of reproducing the correct global zero-detecting field. What remains open is proving that this field arises necessarily from the geometry itself, rather than being selected because it already works.

Independent Finite-Mode Prediction: Phantom Zeros and Mirror-Pair Contamination

The strongest independent mathematical prediction obtained so far does not concern the true Riemann zero set directly. Instead, it concerns the behavior of the global window itself.

The framework suggests that the correct global window should be spectrally transparent: it should carry the arithmetic coherence field without introducing extra zeros of its own. This principle can be tested independently of the actual zeta zeros by deforming the Gaussian window within the simplest self-dual finite-dimensional class.

Finite self-dual deformations. Start from the Gaussian and deform it by adding a finite self-dual Hermite mode. For example, consider a deformation of the form

W_a(x) = e^-πx2 + a H₄(√(2π)x)e^-πx2

where H₄ is the fourth Hermite polynomial and a is a deformation parameter.

In this finite self-dual setting, the Mellin transform factors into the Gaussian Mellin transform times a multiplier Q_a(s). Along the critical line s = 1/2 + it, this multiplier becomes

Q_a(1/2 + it) = 1 + a(8 - 16t²)

This formula is independent of the actual zeta zero set. It is a direct prediction of the deformed window itself.

Phantom zeros. If a ≠ 0, the multiplier has its own zeros at

t = ±√((1 + 8a) / 16a)

For example, if

a = 0.1

then the deformation predicts phantom critical-line zeros near

t ≈ ±1.06066

These are not Riemann zeros. They are zeros created purely by the non-Gaussian deformation of the global window. In the language of the framework, they are phantom balance events introduced by the window itself.

The same phenomenon appears for higher self-dual modes. For instance, the eighth self-dual Hermite mode gives a multiplier of the form

Q_b(1/2 + it) = 1 + b(256t⁴ - 2816t² + 960)

which again produces additional critical-line zeros whenever b ≠ 0. For a small deformation such as b = 0.001, the predicted phantom zeros occur near

t ≈ ±0.864, ±3.202

Mirror-pair contamination. The picture becomes even sharper when multiple self-dual modes are mixed. In that case, one can produce not only extra critical-line zeros but also off-axis mirror-pair zeros of the multiplier.

Numerically, one explicit mixed deformation produced a quartet of off-line zeros near

-0.6377 ± 1.7294 i, 1.6377 ± 1.7294 i

This is structurally important. It means that non-Gaussian self-dual windows do not merely create extra zeros. They can create exactly the wrong kind of zeros: mirror-paired off-axis events rather than fixed-point collapses on the critical line.

In the geometric language of the framework:

• the Gaussian preserves spectral transparency
• non-Gaussian finite self-dual deformations inject phantom zeros
• and mixed deformations can support separated mirror pairs instead of collapse to the symmetry axis

Interpretation. This is the first place where the framework makes a clean independent prediction.

It predicts that once the global window is deformed away from the Gaussian within the finite self-dual Hermite class, the Mellin-side spectrum becomes contaminated. The Gaussian is therefore not merely convenient. It is singled out by the fact that it avoids phantom zero creation in this finite setting.

This directly supports the broader transparency principle:

the correct global window should not generate its own spectral events.

It also sharpens the newer geometric intuition about fixed-point collapse. The bad windows are precisely those that permit mirror-pair contamination instead of forcing collapse toward the symmetry line.

What this does and does not show. This subsection does show an independent and testable phenomenon:

• non-Gaussian finite self-dual deformations generate extra Mellin zeros
• and mixed deformations can generate off-axis mirror-pair structure

It does not yet prove that the Gaussian is uniquely spectrally transparent in the full infinite-dimensional self-dual class. That broader statement remains conjectural.

Best summary. The first realistic theorem-sized win suggested by the framework is not RH itself. It is this:

In the finite self-dual Hermite setting, non-Gaussian deformations create phantom zeros and mirror-pair contamination, while the Gaussian remains spectrally transparent.

That is the clearest independent mathematical prediction obtained so far.

Exploratory Numerical Behavior of the Crowding Statistic

A useful local quantity suggested by the geometric framework is the crowding statistic

S(j) = Σ_{k ≠ j} 1 / (γ_j - γ_k)²

where γ_j is the imaginary part of the j-th nontrivial zero on the critical line. Geometrically, S(j) measures how strongly the surrounding zeros press on the j-th zero neighborhood. It is the natural local obstruction term in the horizontal convexity picture: the self-term tries to keep the zero rigidly centered, while the crowding term measures the destabilizing influence of the rest of the spectrum.

It is useful to convert this into a characteristic width

δ_crit(j) = 1 / √(S(j))

This quantity can be read as the local horizontal scale on which the self-term still dominates the surrounding zero crowding.

Numerical behavior. Using the first 150 nontrivial zeros, the first few values are approximately

S(1) ≈ 0.052, S(2) ≈ 0.125, S(3) ≈ 0.152, S(5) ≈ 0.284, S(10) ≈ 0.542

The corresponding critical widths are

δ_crit(1) ≈ 4.40, δ_crit(2) ≈ 2.83, δ_crit(3) ≈ 2.57, δ_crit(10) ≈ 1.36

Even by the 150th zero, δ_crit remains on the order of 0.89. In other words, in this explored range the local self-centering effect remains significantly stronger than the aggregate crowding term.

A strongly local phenomenon. The most striking feature of the data is that the crowding term is overwhelmingly local.

On average, the two nearest neighboring zeros already account for roughly two thirds of the full value of S(j). The nearest four account for about four fifths, and the nearest ten for well over ninety percent. This means that the quantity is not driven primarily by the entire distant spectrum, but by the immediate local zero environment.

That is exactly the kind of behavior one would hope for if the geometric picture is correct. It suggests that the problem of collapse to the critical line may be controlled largely by local zero geometry, not only by a fully delocalized global interaction.

Approximate growth with height. In the explored range, the crowding statistic grows slowly with height. A rough empirical fit is

S(j) ≈ 0.113 log²(γ_j / 2π) - 0.019

This should not be treated as a theorem, but it suggests that the local crowding pressure increases in a controlled logarithmic-density way rather than through a violent instability. In geometric language, the environment becomes tighter as one moves upward along the critical line, but it does so gradually.

Off-line quartet footprint. A hypothetical off-line zero quartet at

1/2 ± a ± iγ

would leave a very specific local footprint on the critical line. Near its own height, its contribution to the horizontal Hessian has the form

4(a² - u²) / (a² + u²)²

where u = t - γ measures vertical displacement from the quartet height.

At the center u = 0, the size of this footprint is

4 / a²

Inside the critical strip one always has a ≤ 1/2, so any such quartet would create a central spike of size at least

4 / (1/2)² = 16

This is numerically striking. In the first 150 zeros, the observed crowding statistic never rises above about 3.68. So any genuine off-line quartet inside the strip would appear to force a local convexity footprint much larger than the ambient crowding term seen in the explored critical-line geometry.

Interpretation. This does not prove the Riemann Hypothesis. But it does suggest a much sharper local program than the earlier broad geometric intuition.

Instead of trying to rule out off-line zeros in one global leap, one may ask whether the local horizontal geometry of log|ξ(σ + it)| on the critical line is ever capable of supporting the footprint that an off-line quartet would have to create. If the answer is uniformly negative, then the mirror-pair alternative would be excluded and only fixed-point collapse to the line would remain.

That is the significance of the crowding statistic. It is the first quantity in the framework that turns the collapse picture into a concrete, measurable local obstruction.

Honest status. This section is exploratory, not definitive. The values above are numerical and based on a finite list of zeros, and the quartet-footprint comparison is heuristic rather than proved. But the phenomenon is not empty. The crowding term is structured, strongly local, and substantially smaller than the footprint a hypothetical off-line quartet would seem to require in the low range.

In short:

the data suggest that the local critical-line geometry is much more rigid than an off-line quartet would naturally permit.

Exploratory Hessian Profile Comparison: Fixed-Point Collapse vs. Off-Line Quartet Footprints

A more refined local test of the geometric framework is to examine not only the crowding statistic but the full local Hessian profile of the completed zeta function near known critical-line zeros.

The guiding question is simple:

Does the local geometry near an actual zero look like a single collapse onto the critical line, or does it resemble the footprint that a hypothetical off-line mirror pair would force?

This can be tested numerically.

Horizontal Hessian profile at a true zero height. For a known zero height γ_j, consider the horizontal slice

H_j(δ) = ∂_σ² log|ξ(1/2 + δ + iγ_j)|

This measures the horizontal curvature of the completed zeta field as one moves left and right away from the critical line at the exact height of the j-th zero.

Numerically, for the first several zeros, the profile takes a remarkably simple form:

H_j(δ) ≈ -1 / δ² + C_j

where C_j is a small positive constant depending on the local zero environment.

For example, for the first zero the renormalized quantity

H₁(δ) + 1 / δ²

stabilizes very quickly near a constant value around 0.0677. For the second zero it is around 0.139, for the third around 0.165, and for the tenth around 0.551.

The striking feature is not the exact constants, but the shape:

• there is one universal central singularity at δ = 0
• and after removing it, the remaining profile is smooth and nearly flat

There is no sign of extra side-poles or multiple competing collapse centers.

Vertical Hessian profile on the critical line. Now stay on the critical line and vary the height. Define

V_j(u) = ∂_σ² log|ξ(1/2 + i(γ_j + u))|

where u = t - γ_j is the vertical displacement from the zero.

Numerically, this profile also simplifies dramatically:

V_j(u) ≈ 1 / u² + D_j

with D_j again a small positive background term depending on local crowding.

For the first zero, the renormalized quantity

V₁(u) - 1 / u²

stabilizes near 0.068. For the second zero it is around 0.14, for the third around 0.16, and for the tenth around 0.52.

So vertically, too, the profile is dominated by a single central spike plus a smooth positive background.

Comparison with a hypothetical off-line quartet. Now compare this with the local signature forced by a hypothetical off-line zero quartet at

1/2 ± a ± iγ

Such a quartet would leave a very different footprint.

On the horizontal slice t = γ, its contribution would have poles at

δ = ± a

so instead of one central singularity at the critical line, one would see two separated side-poles.

On the critical line itself, near the quartet height, the vertical footprint would take the form

Q_a(u) = 4(u² - a²) / (a² + u²)²

This shape is qualitatively different from what is seen numerically. In particular:

• it is negative for |u| < a
• it vanishes at |u| = a
• and only becomes positive outside that interval

So an off-line quartet would produce a negative well around its own height on the critical line.

By contrast, the actual critical-line zeros that were examined produce a positive 1/u²-type spike with a smooth positive background.

Interpretation. This comparison suggests a much sharper local picture than the earlier general intuition.

At least in the low-to-moderate range explored numerically, the local geometry near actual zeros looks like a single fixed-point collapse centered on the critical line, not like the footprint of a surviving off-line mirror pair.

In other words:

• a true critical-line zero behaves locally as though it is genuinely centered on the line
• whereas a hypothetical off-line quartet would leave a qualitatively different signature: side-poles horizontally and a negative well vertically

This does not prove the Riemann Hypothesis. The analysis is numerical, local, and based on known critical-line zeros. But it identifies a promising obstruction mechanism:

if one could prove that the actual horizontal and vertical Hessian profiles on the critical line can never exhibit the side-pole or negative-well signatures forced by off-line quartets, then off-line zeros would be ruled out.

Best summary. The observed local Hessian profiles near known zeros behave like

-1 / δ² + smooth positive background    horizontally,

1 / u² + smooth positive background    vertically on the line.

A hypothetical off-line quartet would instead force separated side-poles horizontally and a negative well vertically.

So the local geometry of known zeros appears to support the fixed-point collapse picture rather than the mirror-pair survival picture.

Honest status. This is an exploratory numerical result, not a theorem. But it is the first local comparison in the framework that directly distinguishes between

• zeros as true fixed-point collapses on the line
• and zeros as hidden off-line mirror pairs

That makes it one of the most promising local tests produced by the framework so far.

Exploratory Normalized Hessian Profiles: Evidence for a Local Lattice Geometry

The previous Hessian comparison suggested that the local geometry near a critical-line zero looks like a single centered collapse, not like the footprint of a surviving off-line mirror pair. A natural next question is whether this local picture becomes even cleaner after normalizing by the zero's own local spacing.

To test this, define the local spacing scale at the j-th zero by the harmonic mean of the adjacent gaps:

h_j = 2 / (1 / (γ_j - γ_j-1) + 1 / (γ_j+1 - γ_j))

This gives the natural local vertical scale of the zero environment around γ_j.

Now consider again the crowding statistic

S(j) = Σ_{k ≠ j} 1 / (γ_j - γ_k)²

A striking numerical collapse. When normalized by the local spacing, the crowding statistic becomes almost constant:

h_j² S(j) ≈ π² / 3

Numerically, over the interior portion of the first 80 zeros, the mean value is about

3.294

the median is about

3.292

while

π² / 3 = 3.289868...

This is a strong indication that the local zero environment is behaving almost exactly like a uniform vertical lattice with spacing h_j.

In other words, once each zero is measured in its own local units, the crowding from nearby zeros is nearly universal.

Predicted lattice kernels. If the zeros near γ_j are modeled as a locally uniform vertical lattice with spacing h_j, then the horizontal Hessian at height t = γ_j should take the form

H_j(δ) ≈ -(π² / h_j²) csch²(πδ / h_j)

and the vertical Hessian on the critical line should take the form

V_j(u) ≈ (π² / h_j²) csc²(πu / h_j)

These are the exact kernels one obtains from a centered point in a uniformly spaced one-dimensional lattice.

Numerical comparison with actual zeros. These lattice kernels were compared numerically with the actual completed-zeta Hessian profiles at the first few known zero heights, using sample offsets such as 0.05, 0.1, and 0.2.

The agreement was unexpectedly strong.

For the first four zeros, the relative errors were typically between about

10^-5 and 10^-3

and even for the fifth zero they remained below about

1%

So after normalization by the local spacing h_j, the actual Hessian profiles near true zeros are not merely spike-like. They are numerically very close to the exact kernels produced by a locally uniform vertical zero lattice.

Geometric meaning. This substantially sharpens the earlier fixed-point-collapse picture.

The local geometry near a known zero now appears to look like:

• a single centered collapse singularity horizontally
• a single centered spike vertically
• and a surrounding background controlled almost entirely by the local spacing

That is exactly the geometry one would expect from a fixed point embedded in a near-lattice environment.

By contrast, a hypothetical off-line mirror quartet would force a qualitatively different profile:

• separated side-poles in the horizontal direction
• and a negative well in the vertical direction on the critical line

The normalized lattice-like behavior seen numerically near true zeros is therefore much more consistent with single fixed-point collapse than with hidden off-line mirror-pair survival.

Best interpretation. This suggests that the relevant local geometry of the zeta zeros may not be arbitrary or highly irregular. Instead, after normalization by local spacing, it appears to be governed by a universal near-lattice law.

In plain terms:

the zeros may locally organize like nearly evenly spaced beads on a vertical string, and the Hessian geometry of the completed zeta field seems to reflect exactly that structure.

This is a much sharper picture than the earlier general intuition, because it identifies an explicit local model and shows that the data fit it very well.

Honest status. This is still exploratory and numerical. It is not a theorem. But it identifies what may be the most promising local theorem target so far:

to prove that the normalized Hessian geometry near critical-line zeros is asymptotically lattice-like, and that this local lattice geometry is incompatible with the footprint of off-line quartets.

If such a statement could be established rigorously, it would turn the framework's collapse intuition into a genuine local obstruction principle.

Summary. After normalizing by the local spacing h_j, the crowding statistic becomes nearly universal:

h_j² S(j) ≈ π² / 3

and the actual horizontal and vertical Hessian profiles near known zeros closely match the exact kernels of a locally uniform vertical lattice.

So the local geometry of the critical-line zeros appears not only centered, but almost lattice-like. That makes the fixed-point-collapse picture significantly more precise, and it strongly disfavors the kind of side-pole and negative-well signatures that off-line mirror quartets would create.

Exploratory Rescaled Residuals: Evidence for a Local Universality Class

The lattice-kernel comparison suggested that, after normalization by local spacing, the horizontal Hessian profiles near critical-line zeros are extremely close to the exact kernels of a uniform vertical lattice. The next natural question is whether the remaining error is small in a completely unstructured way, or whether it too follows a stable law.

To test this, define the local spacing scale at the j-th zero by the harmonic mean of the adjacent gaps,

h_j = 2 / (1 / (γ_j - γ_j-1) + 1 / (γ_j+1 - γ_j))

and define the crowding defect

c_j = h_j²S(j) - π² / 3,    S(j) = Σ_{k ≠ j} 1 / (γ_j - γ_k)²

The basic lattice-kernel model for the horizontal Hessian is then

h_j²H_j(h_jx) ≈ -π² csch²(πx) + c_j

where x = δ / h_j is the rescaled horizontal variable and

H_j(δ) = ∂_σ²log|ξ(1/2 + δ + iγ_j)|

The rescaled residual. The quantity tested numerically was the fully rescaled residual

E_j(x) = h_j²H_j(h_jx) + π² csch²(πx) - c_j

If the lattice kernel plus crowding defect were the full local law, then E_j(x) would be negligible.

Numerically, it is indeed small in the near-field regime.

For example, using the first 120 zeros, the mean residuals across the sample are approximately

E_j(0.025) ≈ -0.00210,    E_j(0.05) ≈ -0.00832,    E_j(0.10) ≈ -0.0321

These values are tiny relative to the main kernel. At x = 0.1, for instance,

π² csch²(πx) ≈ 97.5

so a residual of size 0.03 is negligible compared with the dominant shape.

A new universal correction. The more interesting fact is that the residual is not random. Its average across zeros follows a strikingly simple law:

E_j(x) ≈ -(π² / 3)x²    for small x.

Numerically, the ratio of the mean residual to x² is approximately

-3.36 at x = 0.025,    -3.33 at x = 0.05,    -3.28 at x = 0.075,    -3.21 at x = 0.10

which sits very close to

-π² / 3 ≈ -3.289868...

So the next correction term appears to be universal as well.

This leads to a refined local model:

h_j²H_j(h_jx) ≈ -π² csch²(πx) + c_j - (π² / 3)x²

Equivalently, in the original δ-variable,

H_j(δ) ≈ -(π² / h_j²) csch²(πδ / h_j) + c_j / h_j² - (π² / 3)(δ² / h_j⁴)

After removing the quadratic correction. Once this quadratic term is subtracted, the average residual nearly disappears.

For example, after subtracting the term -(π² / 3)x², the corrected mean residuals become approximately

-4.2 × 10^-5 at x = 0.025,    -9.4 × 10^-5 at x = 0.05,    6.0 × 10^-5 at x = 0.075,    7.7 × 10^-4 at x = 0.10

So, at least in the low-to-moderate range explored numerically, the average local profile is almost completely described by three terms:

• the universal lattice kernel
• the local crowding defect c_j
• and a universal quadratic correction -(π² / 3)x²

Interpretation. This suggests that the local critical-line geometry may belong to a genuine universality class.

After spacing normalization, the horizontal Hessian does not just resemble a lattice kernel vaguely. It appears to admit a stable local expansion of the form

universal lattice core + one local defect parameter + one universal quadratic correction + small remainder.

That is a much sharper result than the earlier qualitative statement that the zeros look centered.

In the language of the framework, this means the geometry near a true zero may not be arbitrary at all. It may be rigidly controlled by a universal fixed-point-collapse profile, with only a small defect measuring deviation from an ideal local lattice.

Relevance to the off-line quartet problem. This matters because a hypothetical off-line quartet would still force the wrong kind of local shape: side-poles horizontally rather than a single centered singularity plus a smooth correction.

So the more rigid this rescaled local expansion becomes, the harder it is for a hidden off-line quartet to be compatible with the observed critical-line geometry.

Honest status. This is still an exploratory numerical result, not a theorem. But it identifies what may be the most promising local theorem target yet:

to prove that, after spacing normalization, the horizontal Hessian near a critical-line zero admits a universal lattice-kernel expansion with a small controlled remainder.

If such a statement were established rigorously, it would turn the framework's collapse intuition into a genuine local rigidity principle.

Summary. After rescaling by the local spacing h_j, the horizontal Hessian near known zeros is extremely well approximated by a lattice kernel. The remaining residual is not random: it appears to follow a universal quadratic law

-(π² / 3)x²

after which only a very small remainder remains.

So the local geometry near critical-line zeros appears to be governed not just by a lattice-like main term, but by a genuine local universality class.

What Is Established and What Is Not

What is genuinely established inside the framework:

• multiplicative scale coherence forces the phase law e^{it log q}
• unitary dilation forces the amplitude q^-1/2
• the critical-line atom q^-1/2-it therefore emerges naturally
• the toy candidate spectrum has the right broad spectral flavor
• but it overcounts
• and the self-dual Gaussian/theta window gives the completed zeta object once chosen

What remains unproved:

• a derivation of the exact global zero-selection law
• a derivation of the theta window from 12d geometry with no external choice
• a proof that the selected global balance heights are exactly the nontrivial zeta zeros
• and a proof that this framework is equivalent to the true zeta mechanism

What Would a Formal Proof Need to Establish?

At minimum, a proof would need to establish the following.

1. Structural encoding. That the zeta function truly encodes the proposed fundamental unit structure, whether in the original 5-role language or in the later 12d reformulation.

2. Local atom. That the geometric assumptions force the exact critical-line local atom q^-1/2-it, not just something qualitatively similar.

3. Global window. That the true geometry canonically produces the self-dual Gaussian/theta window rather than requiring it to be chosen by hand.

4. Deterministic selector. That the candidate heights are thinned by a deterministic global coherence law which reproduces the exact zeta spectrum.

5. Counting law. That the selected heights satisfy the correct asymptotic counting law of the nontrivial zeros.

What Would Falsify This?

This framework would fail if any of the following were shown.

• the local phase/amplitude derivation does not genuinely connect to the zeta critical-line atom
• the 12d or 5-role structural reading is incompatible with the formal scale geometry of the zeta function
• the global self-dual window is not derivable in any natural way from the proposed geometry
• the candidate spectrum cannot be thinned by any natural deterministic coherence rule into the actual zero set

Final Summary

The cleanest summary of the entire exploration is this:

Main discovery: multiplicative scale implies the logarithmic phase log(q), and unitary separation implies the amplitude q^-1/2.

So the framework recovers the correct local critical-line atom

q^-1/2-it

The second discovery is that the simplest self-dual global scale window — the Gaussian/theta construction — automatically produces the completed zeta object.

So the framework now has:

• a plausible structural origin for the critical line
• a derivation of the local critical-line atom
• a plausible global self-dual construction
• and a clear explanation of why the toy candidate spectrum is overcomplete

What remains open is the global problem:

how these local atoms are assembled and selected so that the exact Riemann zero set emerges.

In short:

the framework recovers the local ingredients of the zeta spectrum from geometry. It does not yet prove the global zero-selection law.

Shortest Layman Summary

One rotating scale-wave is not zeta. But if you stack all integer-scaled copies of the simplest self-balancing shape and look at them through the right multiplicative lens, zeta appears.

What remains is to show that the 12d geometry uniquely forces that global self-balancing shape and that the resulting balance events are exactly the Riemann zeros.