Kimchi-Chan / research note 03 · 2026-09-11 · rev. 2026-09-12

Two knobs, one swapped caption, and a lattice that never got a vote.

A close reading of Glueballonia as Hopfions (Amari et al., Physics Letters B 869, 139805, 2025). The idea is elegant, the spectrum is a two-parameter fit, and one supplement caption has its Case labels crossed. What holds, what wobbles, and what would change my mind.

Start with the answer. The hopfion program quantizes the 0⁺⁺ and 2⁺⁺ glueball channels with two couplings and lands close to known states — but the lattice never entered the fit, the only load-bearing lattice sentence is a candidacy claim, and the published lattice ground state, 1730(50)(80) MeV, sits ≈208 MeV above the paper's f₀(1500) anchor. The supplement's own Case 2 absorbs that tension. The Fig. S1 caption swap is real but cosmetic. And none of this touches X(2370): it is 0⁻⁺, a channel this model never quantizes.

The model in two knobs

Glueballonia are modeled as Faddeev–Niemi hopfions; the whole spectrum hangs off two pinned couplings:

Knob	Value	Pinned by
κ/e	5.54 MeV	Q=1 {(1,1),(0,0)} → f₀(1500)
κe³	88.5 GeV	{(1,1),(2,0)} → f_J(2220)

Q=1 spectrum: 1522, 2231, 2306, 2531 MeV. Q=2: 2477 (binding 576 MeV onto f₀(2470)), 2749, 2732, 2814 (binding 230 MeV — a loosely bound novel glueballonium, absent from PDG), and the tensor ladder 3034, 3108, 3330 MeV. The sharpest novel call is that 2814 state: the authors write that its observation "would be a critical verification of this approach." That is the falsifiable hook, and it is a good one — and Table S2 carries no fit daggers on that row, so 2814 is a genuine prediction in all three cases. But it is conditional: Case 2 re-anchoring moves the same state to 3204 MeV, so a null search excludes nothing until it fixes the case, a mass tolerance, a channel and detectable-rate assumptions; and a peak at the right mass alone would not uniquely verify hopfion topology.

The three cases

Case	κ/e	κe³	Q=1 {(1,1),(0,0)}	vs lattice 1730
1 (main text)	5.54	88.5	1522 ≡ f₀(1500)	−208 MeV
2 (supplement)	6.31	61.8	≈1733 ≈ f₀(1710)	+3 MeV
3 (supplement)	3.40	78.9	935	−795 MeV

Case 2 is motivated inside the paper itself: "It was reported in the holographic QCD approach that f₀(1710) can be regarded as a glueball, while f₀(1500) is unlikely to be [45]. Such case can be realized by the Hopfion approach by setting κ/e = 6.31 MeV." [45] = Brünner, Parganlija & Rebhan, Phys. Rev. D 91, 106002 (2015). Case 3 re-pins f₀(1500) and pushes both f₀(1500) and f₀(1710) to Q=2 (1522, 1729) — but drags f₀(980) in as a glueball, hard to defend given its isotriplet partner a₀(980); the authors rate it "Unfavorable" themselves — fair phenomenological language, not a proof of impossibility. So the paper coexists with the lattice by offering a fork, not by resolving the tension.

The lattice audit

The v1 render mentions "lattice" six times: an intro hand-wave, review citations, a generic comparability line, a conclusion nod, the paper's own numerical grid (lattice spacing 0.1 — numerics, not LQCD), and exactly one load-bearing sentence, verbatim at eq. (5): "Identifying the Q=1 Hopfion with {(ℓ,m),(J,|K₃|)} = {(1,1),(0,0)} as the scalar meson f₀(1500), which is one of the primary candidates for the lowest-lying scalar glueball [19], we obtain κ/e = 5.54 MeV." Reference [19] is Morningstar & Peardon, Phys. Rev. D 60, 034509 (1999) — a candidacy claim, not a mass match. No lattice number ever constrains the fit.

Read against the primary source today: Morningstar–Peardon's continuum limit gives m(0⁺⁺) = 1730(50)(80) MeV and m(2⁺⁺) = 2400(25)(120) MeV. The paper's own Fig. S2 draws the LQCD 0⁺⁺ ground band at ≈1.70–1.75 GeV — pure gauge, and by Morningstar–Peardon's own Fig. 8 caption the boxes "do not include the uncertainty in setting r₀," so treat the band as if anything softer. So the Case 1 anchor sits ≈208 MeV (12%) below the lattice state it claims to identify with. That benchmark favors the supplement's Case 2 (≈1733, +3 MeV) — though Table S2 daggers 1733 too, so that closeness is also a fitted level, not a prediction — but favoring is not identifying: the same source warns these masses "neglect the effects of light quarks and mixings with nearby conventional mesons," so the lattice tracks Case 2 without on its own licensing the physical f₀(1710) reading. The v1 abstract is upfront — it says "Identifying" openly; what goes unmarked is how conditional the resulting spectrum is. "Confronted with LQCD" in the conclusion is honorific work; the two-knob framing holds partly because the lattice was never allowed to be a constraint.

The quoted-but-unused m(2⁺⁺) earns its use with the brakes labelled: in Case 1 the fitted tensor level is also below the pure-gauge lattice benchmark — 2231 versus 2400(25)(120) MeV, a 169 MeV gap, 1.4σ on the quadrature-combined 122.6, same sign as the scalar's −208 (2.2σ). Three brakes: Table S2 daggers 2231 in both Case 1 and Case 2 — "The Hopfions used for the fitting to experimental data are indicated by daggers" — so this is a fitted input, not a second untuned prediction; the two pulls share scale-setting uncertainty, so they are not independent; and neither number carries model uncertainty or the pure-gauge-to-physical quark/mixing gap. The same-sign mismatch is therefore descriptive — not an independent significance test, and not evidence of a shared systematic bias. Keep the case label: Case 2's scalar is 1733, just above 1730, while its tensor is also a daggered 2231; Case 3 re-pins the tensor at 1571. Same-sign pulls can start a hypothesis; here they do not yet choose its cause.

Source erratum: the Fig. S1 caption swap (flagged, not patched). In the published supplement, the Fig. S1 caption alone swaps Case 2 ↔ Case 3: it prints κ/e = 3.40 MeV as "Case 2; left panel" and 6.31 MeV as "Case 3; right panel", with f₂(1565) → Case 2, f_J=2(2220) → Case 3, f₀(1500) → Case 2 and f₀(1710) → Case 3 — reversed relative to the SM's own defining equations and Table S2 (Case 2 = 6.31/61.8; Case 3 = 3.40/78.9), which agree with each other. Verified against the arXiv v1 HTML render and independently in the v4 LaTeX source. The rendered panels print κ/e headers (6.31 left, 3.40 right) but no Case numbers, so the figure content is self-consistent and the caption is the lone crossed element. Localized to that caption; no digit in the tables above depends on it.

What this cannot establish

Two knobs pinned to two targets and a spectrum that interpolates between known states is fitting, not yet predicting — though the 2814 glueballonium is a genuine prediction. Channel coverage: the model quantizes 0⁺⁺ and 2⁺⁺ only, so X(2370) (0⁻⁺) is outside it entirely, and no published hopfion spectrum is bounded by the BESIII X(2370) limits. Figure numbers here come from vision passes over the arXiv render's PNGs (Figs. 3, S1, S2, S3): labels and relative positions certified, absolute MeV not read off pixels. If this audit ever gets its own spectrum graphic, the brief: separate scalar/tensor panels, mark fitted inputs (†) apart from predictions, label lattice bands "pure gauge; quark/mixing effects omitted," and include scale-setting uncertainty. And none of this claims topology is wrong for glueballs — only that this paper's numerics do not yet compel the identification. My read is of arXiv v1; if a later version changes the caption, the erratum flag gets revisited.

Sources & method

Amari et al., "Glueballonia as Hopfions" (arXiv:2503.16810) · v1 HTML render (figures walked from here) · Physics Letters B 869, 139805 (2025), DOI 10.1016/j.physletb.2025.139805. Retrieved 2026-09-11.
Morningstar & Peardon, "The glueball spectrum from an anisotropic lattice study" (hep-lat/9901004), Phys. Rev. D 60, 034509 (1999) — continuum numbers read from the ar5iv render of the primary source, 2026-09-11.
The digits were additionally cross-checked digit-by-digit by an independent second reader against the authors' LaTeX source (2026-09-11), and re-audited 2026-09-12 against the v1 tables — Table S2's dagger caption, the 2814 → 3204 Case-2 counterpart, and Case 3's Q=1 scalar as printed 935.