A review of two IT³ preprints

This is a review of two preprints by Dr. Logvinovich on the IT³ framework, the two he named as the place where its proofs are given. It sets out which results are proved, where the steps from mathematics to physics hold or fail and why, and which tests could settle the open points.

Written by Claude prompted by Lance Pollard. Most of the close reading and all of the computation were done with Claude (Anthropic). Every computed value can be rerun with the script at the end, and every quotation can be found by its LaTeX label in the source files on Zenodo.

DOI	record read	title	versions
10.5281/zenodo.21555360	22674891, `YM_V5.tex` (v5.4, 2,159 lines)	The Yang-Mills Mass Gap on ℝ⁴ Is Ill-Posed, and on a Compact Constructible Manifold It Is Automatic	5 (2026-07-25 to 09-09)
10.5281/zenodo.22655608	22882422, `Math_Topology_v17.tex` (1,512 lines)	Algebraic Origin of the 86 ↔ 86 Capacity Symmetry in the Relativistic Dirac-Fock Vacuum (with J.-C. Perez)	17 (2026-09-08 to 09-21)

Both DOIs are Zenodo concept DOIs, and each resolves to the latest version above, which is the version reviewed, read in full from the LaTeX source as published on 2026-10-01. Below, YM is the first paper and CAP the second. Theorems are cited by their LaTeX labels (thm:wall13, rem:lattice-rank), since the PDF numbering shifts between versions. A third record that both papers cite for their particle masses, "Exact Parameter-Free Precision in the IT3 Spectral-Geometric Framework" (concept DOI 10.5281/zenodo.20277693), is called the Boson paper here, following the papers' own citations. It was not part of this review.

Summary

The pure mathematics is correct. The arithmetic of K = ℚ(√2, √3, √5), the Kummer criterion that puts √13 outside K, the index theorem on S⁴ × T³, Friedrich's bound and the facing map on 1..172 all check out. YM also corrects several of its own earlier claims in print, which is uncommon and worth noting.

The steps from that mathematics to physics do not hold as written. Five are decisive:

The lattice CAP defines contains a vector of length √13, so the "Kummer wall" is not a boundary of that lattice (step 5 below).
YM states that the wall does not make space compact and does not cut off the spectrum. CAP, written twelve days later, states both as proved theorems. The two preprints contradict each other on the central mechanism.
The number 86 enters through an identification of the filled Dirac sea with the known elements through radon. It is not derived from the geometry.
1088.04 GeV is not derived in either preprint. Both cite the Boson paper.
"Mass is confined light" is stated without an equation, a cavity size, a mass value or a spin.

Much of it could likely be repaired. Some fixes are already partly written in YM, and others might only mean stating an input as an input or adding one missing calculation. See "How the framework could be repaired".

What remains is testable, and two tests are proposed at the end: the shell structure of period 8, and a pre-registered cross section for the 1088 GeV state.

Errors, problems and unclear points

The issues in plain words

Errors (false as written):

E1. CAP says its lattice has no vector of length √13. It does: (1, 0, 2√3).
E2. CAP says the √13 wall makes space compact. Any lattice of this kind makes a compact space, wall or no wall.
E3. CAP says the wall is a UV cutoff. The field K has numbers arbitrarily close to √13, so it cuts nothing off, and CAP's own spectrum is unbounded.
E4. CAP uses the 48-element cube symmetry O_h. This lattice has only 8 symmetries, because its three axes have different lengths.
E5. The confined-light picture says light without walls has no mass. Two photons flying apart still have mass E/c².
E6. CAP calls the reflected light an "antiphoton" and the reflection CP. A photon is its own antiparticle, and a mirror is not CP.
E7. CAP says matching 2, 8, 8, 18, 18, 32 has a 4⁻⁶ chance. That sequence was the input, so matching it proves nothing.

Problems (true, but they do not show what is claimed):

P1. 86 comes from counting the known elements, not from the geometry.
P2. 240² comes out of every field of this type, so it is not a sign of E8.
P3. The cascade comes back into K at 144, so 13 is not a final exit.
P4. The "FCC closure" works for any number divisible by 4.
P5. The α-decay rule is true of every possible chain, so no experiment can test it.
P6. The mass gap comes from the box size, applies to free fields too, and its size is never fixed.
P7. "Zero fitted parameters," but the measured proton mass is an input.
P8. The knot T(103, 3) changes if one term of the energy function is dropped.
P9. Photons have whole-number spin, so they cannot make an electron (spin ½).

Unclear (neither paper says):

U1. Where the wall is in space.
U2. Where 1088.04 GeV comes from, and what fixes its unit.
U3. What sets the size of the box of light.
U4. Why the field is ℚ(√2, √3, √5), and why the cascade stops at 13.
U5. Whether "no g-block in period 8" is a firm prediction.
U6. Which of the two papers holds the current position.
U7. How strongly the 1088 GeV particle couples, which is needed to test it.

Each one is walked through below in plain terms, with where it is in the text and a way to check it. The step numbers point to the derivation table further down.

Errors, walked through

E1. The lattice contains a vector of length √13 (step 5).

CAP builds its internal space from three arrows at right angles, of lengths 1, √2 and √3. The lattice is every point you can reach by taking whole-number steps along those arrows. CAP's wall theorem says no point of this lattice can sit at distance √13 from the starting point, because √13 is not in the field K.

Take one step along the first arrow and two steps along the third. You land at (1, 0, 2√3). By Pythagoras, its distance is √(1² + (2√3)²) = √(1 + 12) = √13. So the lattice does contain that distance. The same trick gives √7 (two steps along the first arrow, one along the third) and √11 (two along the second, one along the third).

The general reason: every point of the lattice has coordinates in K, but a distance is found by adding squares and taking a square root, and that step can leave K. So "every length in the lattice lies in K" fails for this lattice, and the theorem built on it fails too.

In the text: CAP thm:wall13, Λ "physically cannot contain vectors with lengths requiring the adjunction of new primes (7, 11, 13)".
Check: 1² + (2√3)² = 13. Also 2² + (√3)² = 7 and (2√2)² + (√3)² = 11.

E2. The wall does not make space compact (step 5).

"Compact" here means the internal space closes up on itself, like the surface of a donut, so it is finite. CAP says the √13 wall is what forces this.

But any three arrows pointing in independent directions make a lattice, and wrapping space around any such lattice gives a closed, finite torus. That works for arrows of length 1, √2 and √3. It works just as well for 1, √2 and π, where there is no number field at all and so no wall. So compactness comes from choosing a lattice, not from √13.

CAP's own lem:units (i) says this, and so does YM, which calls the link from the wall to compactness an open conjecture. CAP, written later, calls it "formally proved", and its proof depends on E1.

In the text: CAP thm:wall13 ("forces the tiling to close"), against CAP lem:units (i), YM rem:lattice-rank and YM conj:bridge.
Check: any three independent vectors in ℝ³ generate a lattice whose quotient is a compact torus. This is standard.

E3. The wall is not a UV cutoff (step 8).

A UV cutoff is a smallest possible length, or equivalently a largest possible energy, below which nothing exists. It is what would tame the infinite sums that appear in quantum field theory. CAP says leaving √13 out of K provides one.

But K is not a set of spaced-out lengths. It contains every fraction, and numbers as close to √13 as you like. For example (249 − 18√2)/62 is in K and differs from √13 by less than one part in a million. Taking one number out of a set that fills the whole number line leaves no gap anyone could measure, so there is nothing to act as a cutoff.

CAP's own formula for the allowed light waves on the torus, |k|² = (2π)²(n₁² + n₂²/2 + n₃²/3), has no largest value: you can always pick bigger whole numbers n. And √13 does not appear in it. YM says the same thing in rem:geomcause.

In the text: CAP thm:uv_cutoff, an "absolute UV cutoff" that "resolves the UV catastrophe".
Check: √13 − (249 − 18√2)/62 = 3.7 × 10⁻⁷. The wave with n = (1000, 0, 0) is allowed.

E4. The lattice has 8 symmetries, not 48 (step 14).

A symmetry of a lattice is a rotation or reflection that leaves it looking exactly the same. A cube lattice, with three equal arrows, has 48 of them (the group O_h), including turns that swap one axis for another. CAP uses O_h.

In this lattice the three arrows have different lengths: 1, √2 and √3. Swapping two axes would put an arrow of length 1 where the lattice only has √2, so the lattice would no longer match itself. The only moves that survive are flipping each axis on its own, which gives 2 × 2 × 2 = 8. That group is called D_2h.

This does not change CAP's shell count, because the count 2(2l + 1) comes from ordinary rotations and spin and never needed O_h. It means the stated reason for the count is wrong, not the count.

In the text: CAP lem:Oh.
Check: the script at the end counts the whole-number transformations that preserve x² + 2y² + 3z². There are 8.

E5. Light without walls still has mass ("Mass from confined light").

In relativity, the mass of a system comes from its total energy and total momentum: m²c⁴ = E² − p²c².

Put two equal light pulses in a mirrored box, bouncing back and forth. Total energy E, total momentum zero (they move in opposite directions), so the box gains mass E/c². That is a real, known effect, first worked out by Einstein.

Now remove the box and let the two pulses fly apart in opposite directions. Total energy is still E, total momentum is still zero, so the pair's mass is still E/c². The walls never created the mass. What the walls do is keep the energy in one place, so it moves around as a single object. That is confinement, and it matters, but it is not what "walls off, mass → 0" says.

In the text: CAP thm:reflection_mass, and the author's illustration labeled "WALLS OFF ⇒ mass → 0".
Check: E² − (E/2 − E/2)²c² = E².

E6. A reflected photon is still a photon (step 10).

CAP calls the light bouncing back from the wall a "mirror antiphoton", and calls the reflection CP.

An antiparticle is a particle's partner with opposite charges. The photon has no charge, and physics treats it as its own antiparticle, so "antiphoton" names nothing new. A mirror turns light around in one direction. C swaps particles for antiparticles, and P flips all three directions at once, so a bounce off a mirror is neither C nor CP.

CAP's map i(n) = 11 − n swaps rows of the periodic table. That is a different thing from light hitting a wall, so calling it CP does not connect it to the reflected light.

In the text: CAP thm:mirror_antiphoton.
Check: any quantum electrodynamics text. The photon is its own antiparticle (C = −1).

E7. The 1-in-4,096 chance measures the input ("What is fitted or chosen").

A chance of matching by accident only means something when the answer was not used to build the model.

CAP starts from the sequence 0, 1, 1, 2, 2, 3, which is the order in which shells fill in the real periodic table (the Madelung rule). The shell sizes 2, 8, 8, 18, 18, 32 follow from it by simple arithmetic. Then CAP reports that hitting 2, 8, 8, 18, 18, 32 by chance has odds of 1 in 4⁶ = 4,096.

But the sequence came from the periodic table in the first place, so getting it back was certain. The odds say nothing about whether the framework is right.

In the text: CAP guard:null, with the input in lem:Oh.
Check: add 2(2l + 1) for l = 0 up to 0, 1, 1, 2, 2, 3 in turn.

Problems, walked through

P1. 86 is put in, not derived (step 13). The index theorem CAP uses is correct: it says a certain count on the space (the index) equals a whole number (the degree of a map). But the theorem works for any whole number. Nothing in the geometry prefers 86 over 85 or 1,000. CAP gets 86 by saying the filled states are the elements up to radon and counting them: 2 + 8 + 8 + 18 + 18 + 32 = 86. That is counting the periodic table and giving the count a new name. YM itself calls this step "an epistemic bridge, not a theorem" (rem:indexwitness-status).

P2. 240² is not a sign of E8 (step 6). The discriminant of K is a number that measures how its arithmetic is built. Its square root is 57,600 = 240², and 240 is also the number of roots of E8, a famous structure in mathematics and physics. But the discriminant comes out a perfect square for every field of this kind, built from three square roots: ℚ(√2, √3, √7) gives 336², ℚ(√2, √3, √13) gives 624² and ℚ(√3, √5, √7) gives 420². So being a square is automatic, and 240 = 2⁴ × 3 × 5 is K's own primes multiplied together. Its match with E8 is a coincidence of the number 240.

P3. 13 is not a final exit (step 4). The cascade 1, 2, 3, 5, 8, 13, 21, ... adds the last two numbers each time. CAP marks 13 as where it leaves K, since √13 is not in K. Keep going, though, and it reaches 144, and √144 = 12, which is in K. So the cascade comes back. A wall that the sequence later walks back through needs an explanation, and the text gives none. (That 144 is the only square in this sequence after 1 was proved by Cohn in 1964.)

P4. The FCC closure fits any multiple of 4 (step 16). Face-centered cubic packing is the way oranges stack at a grocery store. Each repeating cube of it holds 4 spheres, and the spheres always fill the same fraction of space, π/√18, about 74%. CAP notes that 172 = 43 × 4 fills 43 cubes exactly. So does any number divisible by 4: 4, 40, 176, 1,000. The packing fraction does not depend on the number at all, so it says nothing about 172 in particular.

P5. The α-decay rule cannot fail (step 22). An alpha decay removes 2 protons, so Z goes to Z − 2, and an even Z stays even while an odd Z stays odd. CAP's rule says a chain's label keeps its parity, landing on lead for even chains and bismuth or thallium for odd ones. Because all of CAP's shell totals are even, its label has the same parity as Z, so the rule reduces to "subtracting 2 keeps parity". That is true of every chain that could ever be observed, so no experiment can test it. Observed superheavy chains usually end in fission, and CAP assigns that in advance to nuclear physics (guard:tfi_nuclear).

P6. The mass gap comes from the box (step 25). The Yang-Mills mass gap problem asks why the carriers of the strong force behave as if they have a minimum energy, in infinite open space. YM and CAP show that on a finite closed space, every wave has a minimum energy. That is true, but it is true of any field in a finite space, including plain light, which has no mass gap in open space. A guitar string has a lowest note because of its length, not because of what it is made of. The size of the gap depends on the size of the space, and neither paper fixes that size. So this shows the box has a gap, not that Yang-Mills does.

P7. A measured value goes in (step 18). The confinement energy 0.966 TeV is computed as 10.0951 × 102 × 0.93827 GeV, and 0.93827 GeV is the proton's measured mass. A measured number used as an input is a parameter, and CAP's ledger lists zero of them. Listing it would be accurate and would cost little.

P8. The knot depends on the formula chosen (step 17). CAP picks the knot T(103, 3) as the lowest point of an energy formula with three terms. Drop one term and the lowest point moves to 101. Drop another and it moves to 102. The paper says the terms come from the spectral action but does not compute them from it. So the choice of formula decides the answer.

P9. Light cannot make an electron ("Mass from confined light"). Every particle has a spin, either whole (0, 1, 2) or half (½, 3/2). Light has spin 1. Combine any number of light waves, standing or moving, and the total spin is still whole. An electron has spin ½. So trapped light can give a body mass, but it cannot by itself make an electron. The "Bose-Fermi translator" in the text does not do this either: the spectral action starts with the electrons already in its equations.

Unclear points, walked through

U1. Where the wall is. If light reflects off a wall, the wall has to be somewhere, at some distance, with a rule for how waves behave there. After E1 the wall is not an edge of the lattice. Neither paper says where else it is, which leaves the reflection, and so the mass mechanism, with nothing to reflect from.

U2. Where 1088 GeV comes from. Both papers state 1088.04 GeV for a predicted particle and point to the Boson paper for the derivation. CAP's code only checks 8 × 136 = 1088, which has no units. There is also a striking match: |Λ₁² − 103| = 1.0883, a number with no units, equals the predicted mass written in TeV to four digits. If that is the derivation, the choice of TeV, rather than GeV or the proton's mass, is doing the work, and it would need a reason.

U3. What sets the box size. Light trapped in a box of size R has energy of about ħc/R: a smaller box means more energy and more mass. So a mass prediction needs R. The MIT bag model gets R from a pressure balance with one measured constant. IT³ names nothing that fixes R, so the mechanism does not yet give a number.

U4. Why this field, and why 13. K is built from √2, √3 and √5. Why stop at √5 and not include √7? And why does leaving K at 13 count, but returning at 144 does not? Neither choice is explained, and both shape every result.

U5. Whether period 8 is a firm prediction. CAP's tables imply that the 8th row of the periodic table has 18 elements, no g-block, and a noble-gas-like closed shell at element 136. Standard relativistic chemistry calculations expect g orbitals in that row instead. This is a real, checkable difference, and neither paper says plainly that IT³ stands behind it.

U6. Which paper is current. YM and CAP say opposite things about whether the wall makes space compact and cuts off high energies (see the table further down). It would help readers to know which one the author holds now.

U7. How strongly the 1088 GeV particle interacts. A new particle at 1088 GeV that interacted as strongly as the Z boson would already have shown up at the LHC, so this one must interact weakly. To be tested, the prediction needs to say how weakly: the production rate times the chance of decaying to electron or muon pairs. Without that, any LHC result can be explained away.

How the framework could be repaired

These are only suggestions, offered in case they are useful. Many of the issues above might be addressed by stating a little less, or by adding one missing piece, and some of the fixes are already partly written in YM.

The wall and compactness (E1, E2, U1, U6). One option might be to carry the YM position over into CAP: the torus is compact because the lattice has full rank, which is a choice of lattice, and could be stated as one. thm:wall13 could then perhaps be narrowed to what it does prove, a fact about which square roots K contains, rather than a statement about Λ. If the √13 idea is meant to carry physics, it would probably need an object that really excludes √13, since Λ does not (E1). A natural first step might be to name that object and prove the exclusion for it.

The UV cutoff (E3). A cutoff needs a length scale, and a number field on its own cannot supply one. One way to get one could be to give the internal space a physical size (fixing R_S⁴ and the scale of Λ in meters or GeV) and show which modes are removed and why. Otherwise it may be simplest to set the claim aside, as YM rem:geomcause already does.

The symmetry (E4). It might work to replace O_h with D_2h, the group this lattice has. Since the 2(2l + 1) counting comes from rotations and spin rather than from O_h, the shell count would likely stay the same. If O_h matters elsewhere, another option could be a lattice that has it, though that would mean equal generator lengths and giving up 1, √2, √3.

Mass from confined light (E5, P9, U3). This could perhaps be written as a bag-model calculation: a field, a boundary condition at a stated radius, the energy of the lowest mode, and a pressure balance that fixes the radius. The mechanism would then produce a number that could be compared with a measured mass. The claim might also read more accurately as "confinement turns field energy into the rest mass of a body", rather than "walls create mass". For spin ½, one route could be to confine a fermion field, as the bag model does with quarks, since photons alone would not give it.

The antiphoton (E6). It might be clearer to drop the word, and to describe i(n) = 11 − n as what it is, a mirror on the floor labels of the periodic table, kept separate from the description of light.

The odds (E7, P2, and "p = 0 by taxonomy"). A count could possibly replace the taxonomy: a list of every identification that was tried, including the ones that did not work, and how often the same matches appear for alternatives such as other triquadratic fields, other starting pairs for the cascade, or other mirror maps. It may also help to note that every triquadratic field gives a square discriminant, and then either set the E8 link aside or show something about K that is specific to E8 beyond the integer 240.

The number 86 (P1). One option might be to list it among the postulates openly ("the degree is set to 86 by the periodic table"). Another could be to look for a mechanism that selects the degree, such as an energy minimum, an anomaly condition or a quantization rule, and show that it picks 86 without counting elements.

The particle masses (P7, U2). It could help to include the 1088.0439 GeV derivation in the paper that uses it, with every input listed and the step that fixes the unit. The proton mass could perhaps be counted as an input in the ledger. That would cost little, since one input is still very few.

The knot (P8). Either the three terms of E(N) and their coefficients could be computed from the spectral action, or E(N) could be listed as a chosen function, along with how far the minimum moves when the function changes.

The mass gap (P6). The claim might be narrowed to what is proved: on this compact space every field, free or interacting, has a gap of order 1/R. From there, fixing R, or addressing the infinite-volume limit that the Clay problem asks about, could be possible next steps.

The α-decay rule (P5). This might be replaced with a prediction that could fail, such as half-lives, branching ratios between α decay and fission, or which isotope of element 120 lives longest.

The predictions (U5, U7). It could be valuable to commit, in public and hashed as ETNO-P1 was, to two things: the period-8 shell structure (no g electrons, a closed shell at 136), and the 1088 GeV state's spin, width and σ × BR into lepton pairs. Each could then pass or fail against data nobody has seen yet, which would likely do more for the framework than further identities.

The claim, as the preprints state it

Space is not ℝ⁴. The arena is the compact 7-manifold M = S⁴ × T³, where S⁴ is spacetime closed up at infinity and T³ = ℝ³/Λ is an internal vacuum torus whose lattice Λ is generated by the orthogonal lengths 1, √2, √3, all in K = ℚ(√2, √3, √5). The field is "self-generated" by the identities 1 + 2 = 3 and 2 + 3 = 5, and the continuation 3 + 5 = 8, 5 + 8 = 13 leaves K at √13. That exit, the "Kummer wall", is read as the reason space is compact, as an absolute UV cutoff, and as a mirror. Light reflecting off the wall forms a standing wave, and that confined energy is rest mass. On a compact space every elliptic operator has a gap, so the Yang-Mills mass gap is "automatic". An odd index on M counts the periodic table: 2 + 8 + 8 + 18 + 18 + 32 = 86 per hemisphere, doubled by a mirror involution to Z_max = 172. The square root of K's discriminant is 57,600 = 240², read as a link to E8. Predictions: a resonance at 1088.04 GeV, Z_max = 172, an α-decay "grammar" for Z = 119 and 120, and a density of 246 g/cm³ for element 120. Zero fitted parameters are claimed.

The objects, as the text defines them

"Arena ℳ = 𝕊⁴ × 𝕋³ over 𝕂". A product of ordinary real smooth manifolds (YM def:M). S⁴ is a round sphere of radius R, and R is never fixed. T³ = ℝ³/Λ is a flat torus with Λ = ℤ·(1,0,0) + ℤ·(0,√2,0) + ℤ·(0,0,√3) (CAP lem:units (i), thm:dual_lattice). "Over K" means that the three generators have coordinates in K. It is not a variety over a number field in the sense of algebraic geometry, and the manifold's points, metric and topology do not depend on the field. The text says so: "This compactness is independent of which field the generators are written in" (CAP lem:units (i)), and "a torus over ℤ and a torus over K with the same generators are the same torus" (YM rem:lattice-rank). The generators already lie in ℚ(√2, √3), so √5 never enters Λ.

"Discrete vacuum lattice". Λ is a discrete subgroup of ℝ³ because it has full rank, which makes the torus compact. Momenta lie in the dual lattice, so the photon spectrum is discrete but unbounded: |k|² = (2π)²(n₁² + n₂²/2 + n₃²/3), n_i ∈ ℤ (CAP thm:dual_lattice). There is no lattice spacing in space and no dynamics on Λ. Here "discrete" means "compact".

The Kummer wall (CAP thm:wall13). For a positive integer n, √n ∈ K exactly when the square-free part of n is in {1, 2, 3, 5, 6, 10, 15, 30}. The cascade 1, 2, 3, 5, 8, 13 first leaves that set at 13. The step from there to a boundary in space is the "physical postulate of constructibility": "the lengths of all lattice vectors ... belong to the field K". From it the proof concludes that Λ "physically cannot contain vectors with lengths requiring the adjunction of new primes (7, 11, 13)" and that this "forces the tiling to close".

Light, the mirror and mass (CAP sec:light). At the wall "the topological refractive index becomes effectively infinite (space ceases to exist)", the field reflects totally, and the reflected wave is the "mirror antiphoton", defined as the image of the photon under the floor involution i(n) = 11 − n, read as CP. Incident and reflected waves form a standing wave, and "localized, confined wave-packet energy is exactly the definition of rest mass" (thm:reflection_mass).

The index. M has odd dimension 7, so a unitary u: M → U(N) gives a Toeplitz operator PuP with an integer index. Since Â(S⁴ × T³) = 1, Ind(PuP) = ∫ch₇(u) = deg(u) (CAP lem:Ahat, thm:deg), with u = β ⊠ v, β the Bott class on S⁴ and v ∈ K¹(T³) a "knotted holonomy" (pos:twist).

The postulates counted (CAP sec:ledger): metric pinning by the inradius (√2 + √3 − √5)/2, existence of the holonomy class [v], and a pair potential U(r) = −σN₁N₂/r for gravity. Three postulates, "continuous fitted parameters: 0", "numerology risk: p = 0".

The derivation, step by step

Kinds: theorem (proved in the text, proof checked), definition, identification (a physical reading given to a mathematical object), chosen (a constant, function or sequence picked, or a measured value used as input), asserted (stated as proved, with no proof given).

#	step	where	kind	verdict
1	K has degree 8, is totally real, Galois group (ℤ/2)³, Δ_K = 2¹⁶3⁴5⁴ = 3,317,760,000, class number 1, Minkowski bound 138.4	CAP `thm:disc`, YM `thm:discroot`	theorem	true (computed). The earlier order-discriminant error was corrected in print
2	1² + (√2)² = (√3)², (√2)² + (√3)² = (√5)², so K is "self-generated from 1"	CAP `thm:pythagorean_closure`	identification	the identities are true. The proof is a narrative (1D, then a plane, then 3D), and it does not say why the field stops at √5. The next identity, 3 + 5 = 8, adds nothing because √8 = 2√2
3	the inradius of the (√2, √3, √5) triangle is (√2 + √3 − √5)/2 ≈ 0.4551	CAP `prop:inradius`	theorem	true. Pinning the metric to it is `pos:metric`, so it is an input
4	√n ∈ K iff the square-free part of n is in {1,2,3,5,6,10,15,30}. The cascade 1, 2, 3, 5, 8 stays in K and 13 leaves	CAP `thm:wall13`	theorem	true (computed). Two qualifications. The cascade returns to K at x₁₁ = 144 = 12². Over 300 terms the only members in K are 1, 2, 3, 5, 8 and 144, which matches Cohn's 1964 result that 1 and 144 are the only square Fibonacci numbers. And the wall moves with the starting pair and the field, as YM `app:subfield` shows
5	therefore Λ cannot contain a vector of length √13, so T³ must be compact	CAP `thm:wall13` proof	asserted, false	Λ contains 1·(1,0,0) + 2·(0,0,√3) = (1, 0, 2√3), of length √(1 + 12) = √13. It also contains (2, 0, √3) of length √7 and (0, 2√2, √3) of length √11 (computed). The form x² + 2y² + 3z² represents 7, 11, 13, 14, 17, 19, 21, 22, 23 and more. So the constructibility postulate fails for the lattice the paper defines. Compactness comes from full rank, as CAP `lem:units` (i) and YM `rem:lattice-rank` both say. YM records the wall-to-compactness step as an open conjecture (`conj:bridge`). CAP, written later, calls it "formally proved"
6	√Δ_K = 57,600 = 240², the square of the E8 kissing number, so the vacuum covolume is "the tensor square of the E8 kissing configuration"	CAP `thm:e8_resonance`	identification	the arithmetic is true and the link is generic. √Δ is a perfect square for every triquadratic field tried: ℚ(√2,√3,√7) gives 336², ℚ(√2,√3,√13) gives 624², ℚ(√3,√5,√7) gives 420² (computed). Each odd ramified prime divides exactly four of the seven quadratic discriminants, so the square is structural. Its root, 240 = 2⁴·3·5, is a power of 2 times the generating primes. That it equals the E8 root count is a property of the integer 240
7	the photon spectrum on T³ is discrete	CAP `thm:dual_lattice`	theorem	true, and standard for any flat torus
8	√13 ∉ K is an "absolute UV cutoff" that "resolves the UV catastrophe"	CAP `thm:uv_cutoff`	asserted, false	step 7's formula is unbounded in n_i. K contains ℚ and is dense in ℝ, so it removes no length scale: (249 − 18√2)/62 ∈ K is within 4 × 10⁻⁷ of √13 (computed). YM `rem:geomcause` says this itself: "There is no theorem asserting that the absence of √13 in K bounds the resolvent norm or truncates the spectral density"
9	at the wall the refractive index is infinite, light reflects totally, and the standing wave is rest mass	CAP `thm:reflection_mass`	identification, stated as a theorem	no proof is given. The wall has no location in space (step 5), so there is nothing to reflect from. No cavity size, boundary condition, mass value or spin follows. See "Mass from confined light" below
10	the reflected wave is a "mirror antiphoton", and i(n) = 11 − n is CP	CAP `thm:mirror_antiphoton`	definition	i acts on floor indices of the periodic table, not on space or on fields. The photon is its own antiparticle, so a reflected photon is a photon
11	annihilation is an XOR on the 16-bit string of 57,600 = 1110 0001 0000 0000	CAP `thm:binary_photon`	identification	the binary string is true. Its physical reading depends on writing the integer in base 2
12	Â(S⁴ × T³) = 1 and Ind(PuP) = deg(u), multiplicative under u = β ⊠ v	CAP `lem:Ahat`, `thm:deg`	theorem	true (S⁴ is stably parallelizable, T³ is flat)
13	deg(u) = 86	CAP `thm:deg86`, `thm:pump`	asserted	not derived. K¹(T³) ≅ ℤ⁴, and the top-degree part of [v] can be any integer, so deg(u) is whatever v one picks. The proof sets the spectral flow equal to "the number of states in the filled bulk", then sets the filled Dirac sea equal to the elements through radon ("Because the physical elements of the periodic system are exactly the ... eigenmodes"). That identification is where 86 enters. YM `rem:indexwitness-status` calls it "an epistemic bridge, not a theorem"
14	C(n) = 2⌊(n+3)/2⌋² = 2, 8, 8, 18, 18, 32 from O_h branching plus Madelung	CAP `lem:Oh`	arithmetic true, inputs chosen	the input l_max(n) = 0, 1, 1, 2, 2, 3 is the Madelung order read off the real table. Restricting SO(3) to a finite subgroup never changes dimensions, so 2(2l + 1) comes from SO(3) and spin, and O_h adds nothing to the count. O_h is also not the symmetry of this lattice. With orthogonal generators of lengths 1, √2, √3 the point group is D_2h, of order 8 (computed: x² + 2y² + 3z² has exactly 8 integer isometries). And the maximal crystallographic point group is not unique, since D_6h (order 24) is also maximal
15	Z_max = 86 + 86 = 172 via the fixed-point-free i(n) = 11 − n	CAP `lem:inv`, `thm:facing`	definition	the involution is chosen to mirror six floors onto six. The facing map is a correct fixed-point-free involution on 1..172, and its proof checks. The agreement with the Dirac-Fock critical charge is discussed below
16	172 = 43 × 4 is "exact FCC closure", packing fraction π/√18 "with zero fitting"	CAP `thm:fcc_closure`	definition	true for any number of spheres divisible by 4. N spheres in N/4 cells give π/√18 for N = 4, 40, 176 or 1000 alike
17	the knot T(103, 3) is the unique minimum of E(N) = (N − Λ₁²)² + Λ₁Ω(N) + Λ₁⁴/N, "derived from the Connes-Chamseddine spectral action"	CAP `thm:knot_derivation`	chosen	the minimum is 103 as published. Without the Λ₁⁴/N term it is 101, and without the Ω term it is 102 (computed). The three-line derivation names a source for each term and computes none of them from the spectral action
18	confinement threshold ΔE = Λ₁ · g(T(103,3)) · m_p ≈ 0.966 TeV	YM `def:anchors`, `prop:knot`, citing the Boson paper Eq. 8	chosen	10.0951 × 102 × 0.93827 GeV = 966.14 GeV (computed). It uses the measured proton mass as an input
19	M_Z′ ≈ 1088.0439 GeV	YM `def:anchors` (cites Boson Eq. 7), CAP ledger (cites "wv36")	asserted here, derived elsewhere	neither preprint derives it. Both cite the Boson paper (`w_boson_v36.tex`), which this review has not read. CAP's verification code checks it as `8*136 == 1088`. Numerically, the writhe deficit \|Λ₁² − 103\| = \|51 + 36√2 − 103\| = 1.0883 is dimensionless and equals M_Z′ in TeV to four digits (M_Z′ / \|Wr\| = 999.75 GeV, computed)
20	gravity: merging lowers σN^(2/3), and U(r) = −σN₁N₂/r	CAP `thm:coalescence`, `pos:pair_potential`	postulate	sub-additivity gives no distance law. The 1/r is a postulate, and the ledger counts it as one
21	α⁻¹ is "the macroscopic measure of the discrete holonomy sum", with 137 = 136 ∨ 1 = 240 − 103 = 13² − 32 and 960/7 ≈ 137.14	CAP `prop:photon_holonomy`, `sec:alpha_register`	asserted, then flagged	no holonomy sum is computed. The integer identities are true and are flagged in the text as "resonances"
22	the parity of the node index μ is conserved along α chains, so even chains land on Pb, odd ones on Bi/Tl	CAP `lem:lsb_parity`, `thm:alpha_decay`	theorem, but cannot fail	true because every closure Σ_n is even and α decay is Z → Z − 2, so Z keeps its parity under any labeling. Observed superheavy chains end in spontaneous fission (²⁹⁴Og → ²⁹⁰Lv → ²⁸⁶Fl → SF), and `guard:tfi_nuclear` assigns branching to "independent nuclear variables" in advance
23	the Topological Friction Index is the Hamming distance between Z and Z − 2 in binary	CAP `def:tfi`	definition	depends on base 2, and does not use the lattice, the field or the index
24	Pauli exclusion is "a Diophantine routing constraint": the spinor suffix 10₂ has no third bit	CAP `sec:deep_resonances`	asserted	a property of how 2 is written in binary
25	the Yang-Mills gap is automatic on M by Cheeger, Friedrich and Lichnerowicz	YM `thm:cheeger`, `thm:lich`, `thm:reductioM`, `thm:secquant`, CAP `thm:gap`	theorems, applied beyond their reach	each inequality is stated correctly (Friedrich's n/(4(n−1)) = 7/24 for n = 7 is right). They hold for a free field, so the same argument shows free Maxwell theory is gapped on S⁴ × T³. The gap scale is set by 1/R_S⁴ and 1/\|Λ\|, and neither is fixed. The Clay problem asks for a gap in infinite volume, which a compact space cannot address, and YM's own scope remarks say it is not a Clay solution

What is correct

The arithmetic of K: degree, Galois group, the discriminant by the conductor-discriminant formula, the index-64 order, the Minkowski bound, class number one, unit rank 7.
The Kummer criterion for which √n lie in K, and √13 ∉ K.
The inradius of the (√2, √3, √5) triangle.
Discrete spectrum of elliptic operators on a compact manifold, Cheeger's inequality, Friedrich's bound, the Weitzenböck formula.
Â(S⁴ × T³) = 1, the odd index equal to the degree, and multiplicativity of the degree under the external product.
The facing map F(Z) = Z + 172 − Σ_n − Σ_{n−1}, a fixed-point-free involution on 1..172 that preserves block and node indices.
The integer identities in the appendices: j((1 + √−43)/2) = −960³, the Heegner primes, Ogg's supersingular primes, the Golay parameters, 43 << 2 = 172. All true, and all flagged in the text as resonances rather than derivations.
YM's corrections of itself, which are real and unusual: rem:lattice-rank (K is dense, compactness comes from rank), rem:geomcause (the wall does not touch the spectrum), rem:singer-compact (Gribov copies persist on S⁴), rem:nogo-scope (the no-go is scoped to the canonical apparatus), rem:88 (8 = 8 is a coincidence), and the twenty-two items of app:honesty.
Pre-registration. In an earlier paper the author published ETNO-P1, a predicted sky position for a distant solar-system object, as a hash before any search. That is the right practice, and the tests proposed below follow it.

The two preprints disagree on the central mechanism

question	YM v5.4 (2026-09-09)	CAP v17 (2026-09-21)
does √13 ∉ K make T³ compact?	no. Compactness comes from the rank of Λ, and the bridge is an open conjecture (`rem:lattice-rank`, `conj:bridge`)	yes, "formally proved" (`thm:wall13`)
does the wall cut off the spectrum?	no. "There is no theorem asserting that the absence of √13 in K bounds the resolvent norm or truncates the spectral density" (`rem:geomcause`)	yes, an "absolute UV cutoff" (`thm:uv_cutoff`)
does the cascade terminate?	"does not terminate over ℝ" (`rem:numbertheory`)	terminates at the wall

Step 5 shows the YM position is the correct one: the lattice contains vectors of length √7, √11 and √13, so the wall cannot be a property of Λ. A statement of which column is the author's current position would settle much of what follows.

What is fitted or chosen

CAP's ledger (sec:ledger) counts three postulates and zero continuous parameters. The text also uses:

the field K itself, and the decision to stop the cascade at its first exit (13) rather than at x₁₁ = 144, where it returns to K
the generators 1, √2, √3 (not √5), and the radius of S⁴
the Madelung sequence l_max = 0, 1, 1, 2, 2, 3 and the cap at l = 3
twelve floors and the mirror i(n) = 11 − n
the three terms of E(N) and their weights
the measured proton mass in ΔE_conf
the external derivation of 1088.0439 GeV, and GeV as its unit
r_eff = 71.0 pm for element 120 (CAP prop:density, flagged in the text as a relativistic-chemistry input)
the exponent 7/2 in P = 4φ^(−7/2) (YM rem:cascade-status: "remains empirical input")
base 2, for every binary argument

"p = 0" is reached "by taxonomy" (CAP's guard on the fine-structure register: "Numerology risk remains p = 0 by taxonomy, not by assertion"). That assigns the risk by a label rather than by counting the identities that were tried. The null model in guard:null computes 4⁻⁶ for reproducing 2, 8, 8, 18, 18, 32, but that sequence is the observed period lengths, which were the input.

Mass from confined light

The physical idea is sound, and it is known physics. A photon in a perfect mirror box adds E/c² to the box's rest mass (Einstein 1906). Most of the proton's mass is field energy of nearly massless quarks and gluons confined in about 1 fm. The MIT bag model (Chodos, Jaffe, Johnson, Thorn and Weisskopf 1974) is the closest textbook version of the IT³ picture: massless fields held by a boundary condition at a wall, with energy about 2.04/R per quark mode plus a volume term BV, and the radius fixed by pressure balance, dE/dR = 0. The bag constant B is the one parameter that sets the mass scale. For "mass is trapped light" to produce a number, something in IT³ has to play the role of B and R. Neither preprint names one. For a box of light, relativity also requires the radiation pressure to be balanced by tension in the walls before the energy transforms as the rest energy of a body (von Laue 1911), so the walls need a physical description, not only a position.

Three points on the standing-wave picture as stated.

Removing the walls does not make the mass zero. Two photons of energy E/2 moving in opposite directions have total energy E and total momentum 0, so their invariant mass is E/c² with or without walls. Each photon alone is massless in both cases. The walls keep the energy in one place, so it behaves as a body with a rest frame. "Walls off, mass → 0" conflates the invariant mass of the system with its confinement.
A cavity mode is a boson. Photons carry integer angular momentum, and a standing wave of them cannot be a spin-½ fermion. The spectral action is invoked as a "Bose-Fermi translator", but it computes a bosonic action from a Dirac operator that already contains the fermions. It does not turn photons into electrons.
A reflected photon is a photon. The photon is its own antiparticle, and reflection reverses momentum, which is neither C nor CP.

The two experiments shown in support. Stodolna et al. (PRL 110, 213001, 2013) imaged the nodal structure of hydrogen Stark states by photoionization microscopy, and the pattern agrees with the Schrödinger equation in a static field. Piazza et al. (Nat. Commun. 6, 6407, 2015) imaged a standing plasmon wave on a silver nanowire, with the light confined between the two metal ends of the wire. Both agree with standard quantum theory, and in both the confining boundary is physical matter.

Where the framework is strongest

Z = 172 and the Dirac-Fock critical charge. For a Dirac electron around an extended nucleus, the 1s level reaches −mc² near Z_cr ≈ 170 to 173, depending on the nuclear radius model (Pieper and Greiner 1969, and later refinements). Pyykkö's 2011 table runs to 172 for that reason. IT³ reaches 172 as 2 × (2 + 8 + 8 + 18 + 18 + 32), from a different route. That agreement is real and is the strongest single fact in the framework's favor. Two caveats: past Z_cr the vacuum becomes charged rather than the element ceasing to exist, and Z_cr itself is not sharp, so the agreement cannot be made sharper than about ±2.

Period 8 is a sharp prediction. The CAP ledger lists group-18 closures at 2, 10, 18, 36, 54, 86, 118, 136, 154, 162, 170, 172. That fixes period 8 at 18 elements, filled s (119, 120), d (121 to 130) and p (131 to 136), with no g-block and no closure at 126. The extended Madelung rule, with l unbounded, gives period lengths 2, 8, 8, 18, 18, 32, 32, 50, and Dirac-Fock calculations place a 5g series in period 8. So IT³ and standard relativistic quantum chemistry make different predictions about the same computable object.

Proposed tests

prediction	as stated	what makes it falsifiable	how it would fail
period 8 has 18 elements, closed shells at 136, 154, 162, 170, 172, no g electrons	CAP ledger	already sharp	published relativistic ground configurations (Dirac-Fock, Dirac-Coulomb-Breit coupled-cluster) place g electrons in a ground state, or find no closed shell at Z = 136. The cleanest test available, because both sides compute the same object
no electronic closure at 126	CAP ledger	already sharp	a relativistic calculation finds a closed shell at 126
a resonance at 1088.04 GeV	"first higher resonance", "HL-LHC projection Z ≈ 10.07σ" (YM `rem:fals`)	the state's spin, width, and σ × BR into ee and μμ at √s = 13.6 TeV, stated before new data and hashed as ETNO-P1 was	the LHC dilepton limit at 1088 GeV falls below the stated σ × BR. If a peak appears there, that is a real result
Z_max = 172	"absolute geometric end"	a statement about chemistry that can exist	not testable in the laboratory. Testable in silico through the shell structure above
α chains of 119 and 120 land on Bi/Tl or Pb	CAP `thm:alpha_decay`	a claim the parity rule does not already guarantee	as stated it cannot fail (step 22)
ρ(120) ≈ 246 g/cm³	CAP `prop:density`	an IT³ value of r_eff	the formula is hard-sphere FCC packing, so it tests the 71 pm input

Questions for the author

Which column of the table above is the current position: the YM conjecture or the CAP theorem? If the wall is not a place in space, what does the light reflect from?
Which equation in the Boson paper gives 1088.0439 GeV, which inputs does it use, does any measured mass enter, and what fixes GeV as the unit?
What sets the radius of S⁴ and the scale of Λ, and so the size of the gap and of any confined-light mass?
Can the standing-wave mechanism produce one known mass, the electron's, with spin ½?
Is there a route to deg(v) = 86 that does not count the known elements?
Is "no element has g electrons in its ground state, and Z = 136 is a closed shell" a prediction IT³ stands behind?
Why does the cascade stop at its first exit, 13, and not count the return at 144 = 12²?

What this review did not check

The Boson paper, where 1088.0439 GeV and ΔE_conf are said to be derived. Its answer to question 2 may change the verdicts on steps 18 and 19.
The CMS 2.45σ dielectron excess and the "blind aggregation of 26 datasets" cited in YM rem:fals. These were not checked against CMS publications.
The dilepton exclusion. A Z′ with sequential-Standard-Model couplings is excluded to roughly 5 TeV by the 13 TeV searches cited below. The exact limit at 1088 GeV for a narrow state with smaller couplings was not read off those papers, which is why the test above asks for a stated σ × BR.

Rerunning the computations

Every "computed" value above comes from this script. Python 3.8 or later, standard library only.

from math import isqrt, sqrt

def squarefree(n):
    out, p = 1, 2
    while p * p <= n:
        while n % (p * p) == 0:
            n //= p * p
        if n % p == 0:
            out *= p
            n //= p
        p += 1
    return out * n

IN_K = {1, 2, 3, 5, 6, 10, 15, 30}  # square-free parts whose root lies in Q(√2,√3,√5)

def root_in_K(n):
    # √n ∈ K exactly when n = m·k² for some m in IN_K
    return any(n % m == 0 and isqrt(n // m) ** 2 == n // m for m in IN_K)

# Step 4: the cascade 1, 2, 3, 5, 8, 13, ... and which terms have a root in K.
a, b, inside = 1, 2, [1]
for _ in range(300):
    if root_in_K(b):
        inside.append(b)
    a, b = b, a + b
print("cascade terms with root in K:", inside)  # [1, 2, 3, 5, 8, 144]

# Step 5: vectors x(1,0,0) + y(0,√2,0) + z(0,0,√3) have squared length x² + 2y² + 3z².
for target in (7, 11, 13):
    hits = [(x, y, z) for x in range(4) for y in range(4) for z in range(4)
            if x * x + 2 * y * y + 3 * z * z == target]
    print(f"length √{target} in Λ:", hits)

# Step 14: integer isometries of the form x² + 2y² + 3z² (the point group of Λ).
G = (1, 2, 3)
def form(u, v):
    return sum(G[i] * u[i] * v[i] for i in range(3))
small = [(x, y, z) for x in range(-2, 3) for y in range(-2, 3) for z in range(-2, 3)]
cols = [[v for v in small if form(v, v) == G[i]] for i in range(3)]
count = sum(1 for c0 in cols[0] for c1 in cols[1] for c2 in cols[2]
            if form(c0, c1) == form(c0, c2) == form(c1, c2) == 0)
print("point group order:", count)  # 8, which is D_2h, not O_h (48)

# Steps 1 and 6: discriminant of Q(√p,√q,√r) as the product of its 7 quadratic subfields.
def quad_disc(m):
    m = squarefree(m)
    return m if m % 4 == 1 else 4 * m
def triquad_disc(p, q, r):
    d = 1
    for m in (p, q, r, p * q, p * r, q * r, p * q * r):
        d *= quad_disc(m)
    return d
for f in ((2, 3, 5), (2, 3, 7), (2, 3, 13), (3, 5, 7)):
    d = triquad_disc(*f)
    s = isqrt(d)
    print(f"Q(√{f[0]},√{f[1]},√{f[2]}): Δ = {d}, √Δ = {s} = {isqrt(s)}², square: {s * s == d}")

# Step 1: the Minkowski bound (n!/n^n)·√Δ for a totally real field of degree 8.
print("Minkowski bound:", 40320 / 8**8 * sqrt(triquad_disc(2, 3, 5)))  # 138.43

# Step 8: K is dense. An element of K close to √13.
print("√13 − (249 − 18√2)/62 =", sqrt(13) - (249 - 18 * sqrt(2)) / 62)

# Steps 18 and 19.
L1sq = 51 + 36 * sqrt(2)
print("Λ₁ =", sqrt(L1sq), " |Λ₁² − 103| =", abs(L1sq - 103))
print("Λ₁ · 102 · m_p =", sqrt(L1sq) * 102 * 0.93827, "GeV")
print("1088.0439 / |Wr| =", 1088.0439 / abs(L1sq - 103), "GeV")

# Mass from confined light, point 1: two photons of E/2 moving apart.
E = 1.0
p1, p2 = E / 2, -E / 2
print("invariant mass of the pair:", sqrt(E**2 - (p1 + p2) ** 2))  # E, with no walls

# Period 8 by the extended Madelung rule, l unbounded.
def madelung_periods(k):
    order = sorted(((n + l, n, l) for n in range(1, 12) for l in range(n)))
    lengths, cur = [], 0
    for _, n, l in order:
        if l == 0 and cur:
            lengths.append(cur)
            cur = 0
        cur += 2 * (2 * l + 1)
    return lengths[:k]
print("period lengths:", madelung_periods(8))  # [2, 8, 8, 18, 18, 32, 32, 50]

Sources

Logvinovich, V. 2026. YM v5.4, Zenodo record 22674891 (concept DOI 10.5281/zenodo.21555360).
Logvinovich, V. and Perez, J.-C. 2026. CAP v17, Zenodo record 22882422 (concept DOI 10.5281/zenodo.22655608).
The Boson paper, concept DOI 10.5281/zenodo.20277693, identified by its landing page only.
Einstein, A. 1906. "Das Prinzip von der Erhaltung der Schwerpunktsbewegung und die Trägheit der Energie." Ann. Phys. 20, 627.
von Laue, M. 1911. "Zur Dynamik der Relativitätstheorie." Ann. Phys. 35, 524.
Chodos, A., Jaffe, R. L., Johnson, K., Thorn, C. B. and Weisskopf, V. F. 1974. "New extended model of hadrons." Phys. Rev. D 9, 3471.
Pieper, W. and Greiner, W. 1969. "Interior electron shells in superheavy nuclei." Z. Phys. 218, 327.
Pyykkö, P. 2011. "A suggested periodic table up to Z ≤ 172, based on Dirac-Fock calculations on atoms and ions." PCCP 13, 161.
Stodolna, A. S. et al. 2013. "Hydrogen atoms under magnification: direct observation of the nodal structure of Stark states." PRL 110, 213001.
Piazza, L. et al. 2015. "Simultaneous observation of the quantization and the interference pattern of a plasmonic near-field." Nat. Commun. 6, 6407.
Cohn, J. H. E. 1964. "On square Fibonacci numbers." J. London Math. Soc. 39, 537.
ATLAS Collaboration 2019. "Search for high-mass dilepton resonances using 139 fb⁻¹ of pp collision data collected at √s = 13 TeV." Phys. Lett. B 796, 68. CMS Collaboration 2021, JHEP 07, 208.
Oganessian, Yu. Ts. et al. 2006. "Synthesis of the isotopes of elements 118 and 116 in the ²⁴⁹Cf and ²⁴⁵Cm + ⁴⁸Ca fusion reactions." Phys. Rev. C 74, 044602.

lancejpollard/model.md

Select an option

No results found