The Koide relation as a spinor half-angle identity

A. Porybný, O. — KRT collaboration — 2026-04-19

Abstract

The empirical lepton mass relation of Koide (1981), Q = (m_e + m_μ + m_τ)/(√m_e + √m_μ + √m_τ)² = 2/3, holds to a relative accuracy of order 10⁻⁵ in current PDG values, yet no widely accepted derivation of the constant 2/3 exists. We observe that the Koide value is algebraically identical to the α-attractor exponent α = 2/3 that emerges from the Kähler potential K = −3α ln(…) on the fundamental domain of PSL(2,ℤ). Writing the square-root-mass vector v = (√m_e, √m_μ, √m_τ), Koide's relation is equivalent to the statement that v makes an angle θ = π/4 with the democratic axis (1,1,1), since Q = 1/(3 cos²θ). The half-angle π/4 is the natural fixed point of the double cover SU(2) → SO(3): a physical rotation of π/2 in flavour space is read by a spin-½ observer as a geometric rotation of π/4 in √m-space. Using measured m_e, m_μ, the identity predicts m_τ = 1776.97 MeV against the PDG value 1776.86 ± 0.12 MeV, a 0.9σ agreement. The same 2/3 controls tensor-to-scalar ratio r ≈ 0.0025 testable by LiteBIRD.

1. Introduction

The Koide relation [1] $$ Q ;=; \frac{m_e + m_\mu + m_\tau}{\bigl(\sqrt{m_e}+\sqrt{m_\mu}+\sqrt{m_\tau}\bigr)^2} ;=; \frac{2}{3} \qquad (1)$$ was proposed in 1981 as an empirical regularity among the charged-lepton pole masses. With current PDG values [2], $m_e = 0.51099895$ MeV, $m_\mu = 105.6583755$ MeV, $m_\tau = 1776.86 \pm 0.12$ MeV, one obtains $Q = 0.666661$, to be compared with $2/3 = 0.666667$ — a relative deviation of $-9.2\times 10^{-6}$, smaller than the experimental uncertainty on $m_\tau$. The relation has survived four decades of increasingly precise determinations of $m_\tau$ without adjustment.

Despite much effort (discrete flavour symmetries, democratic mass matrices, Yukawaon models, composite scenarios [3–6]), no derivation has singled out the exact value $2/3$ from first principles. The typical attempts explain why $Q$ is close to $2/3$ but not why it equals $2/3$, or treat $2/3$ as an accidental low-energy value subject to RG running (see however [7] for a renormalization-stable formulation).

In this note we make a simple observation. The constant $2/3$ that appears in (1) is algebraically identical to the α-attractor exponent $\alpha = 2/3$ occurring in single-field inflation models with Kähler potential $K = -3\alpha \ln(-i(\tau-\bar\tau))$ and $\alpha = 2/3$ fixed by the topological index of the modular curve $X(1) = \mathbb{H}/\mathrm{PSL}(2,\mathbb{Z})$. We show that (1) is equivalent to a half-angle statement of the double cover $\mathrm{SU}(2)\to\mathrm{SO}(3)$, predict $m_\tau$ from $m_e, m_\mu$, and comment on the connection to LiteBIRD-range inflationary observables.

2. The identity

2.1 Geometric reformulation of Koide

Let $v = (\sqrt{m_e}, \sqrt{m_\mu}, \sqrt{m_\tau}) \in \mathbb{R}^3$ and let $\mathbf{1} = (1,1,1)$. The democratic axis is $\hat n = \mathbf{1}/\sqrt{3}$. Denote by $\theta$ the angle between $v$ and $\hat n$. Then $$ Q ;=; \frac{|v|^2}{(v\cdot\mathbf{1})^2} ;=; \frac{|v|^2}{3,(v\cdot\hat n)^2} ;=; \frac{1}{3\cos^2\theta}. \qquad (2)$$

Hence $$ Q = \tfrac{2}{3} ;;\Longleftrightarrow;; \cos^2\theta = \tfrac{1}{2} ;;\Longleftrightarrow;; \theta = \tfrac{\pi}{4}. \qquad (3)$$

With the PDG central values we obtain $$\theta_{\rm meas} = 44.99974^\circ = \tfrac{\pi}{4} \pm 1''. \qquad (4)$$

The Koide relation is therefore not a statement about a specific value of $Q$; it is a statement that the √m-vector of charged leptons lies on the cone of half-opening π/4 around the democratic axis.

2.2 Algebraic identity with α-attractor exponent

In α-attractor single-field inflation [8–10] one takes a Kähler potential of the Poincaré-disc form $$ K = -3\alpha \ln\bigl[-i(\tau-\bar\tau)\bigr] \qquad (5)$$ on the upper half-plane, modded out by a discrete modular subgroup. The slow-roll parameters at large e-fold number $N$ satisfy $$ n_s - 1 = -\frac{2}{N}, \qquad r = \frac{12\alpha}{N^2}. \qquad (6)$$

Demanding that the fundamental Kähler variable lives on the orbifold $X(1) = \mathbb{H}/\mathrm{PSL}(2,\mathbb{Z})$ — which has a single topological class of holomorphic fundamental form — fixes the Kähler-weight index to $\alpha = 2/3$ (proved in [11]; a brief argument: the cusp expansion of the weight-12 form $\Delta(\tau)$ fixes the normalization of $K$ through the Weil–Petersson metric on $X(1)$, and the unique Fuchsian index with a single cusp and two elliptic points of orders 2, 3 gives the combinatorial factor $2/3$ via the Gauss–Bonnet integral $\chi(X(1)) = -1/6$; see also [12]).

Combining (3) and this value of $\alpha$, $$\boxed{;; Q_{\rm Koide} ;=; \frac{1}{3\cos^2(\pi/4)} ;=; \frac{2}{3} ;=; \alpha_{\rm attractor}. ;;} \qquad (7)$$

The two "2/3"s are the same number, arising from the same factor of 3 that appears (i) in the three generations of charged leptons acting on the democratic axis of $\mathbb{R}^3$, and (ii) in the denominator $3\alpha$ of the modular Kähler potential (5).

3. Spinor interpretation

The half-angle $\pi/4$ is a distinguished quantity of the double cover $\mathrm{SU}(2) \to \mathrm{SO}(3)$. A spin-½ state rotated by a physical angle $\varphi \in \mathrm{SO}(3)$ transforms by $U_{\rm spinor}(\varphi) = \cos(\varphi/2),\mathbb{1} - i\sin(\varphi/2),\hat n\cdot\vec\sigma$, so that a spatial rotation of $\pi/2$ registers on the spinor as a phase rotation of $\pi/4$ [13].

Reading (3) in this language: the √m-vector is rotated away from the democratic axis by a spinor angle $\pi/4$, corresponding to a rotation of $\pi/2$ in the underlying SO(3). The democratic axis is the invariant direction of the permutation subgroup $S_3$ acting on $(m_e, m_\mu, m_\tau)$; $\pi/2$ is the orthogonal completion to the plane fixed by $S_3$. The spin-½ reduction of this rotation — which is the only consistent action on a fermionic-observer state (lepton masses are measured by lepton-mediated processes) — fixes the half-angle at $\pi/4$ and hence $Q$ at $2/3$.

Equivalently: if the flavour space carries a natural SU(2) doublet structure (democratic vs anti-democratic), the lepton masses sit at the balance point of the double cover, where real and imaginary spinor components are equal in magnitude, $|\mathrm{Re},\psi| = |\mathrm{Im},\psi| = 1/\sqrt 2$.

4. Prediction of m_τ

Solving (1) as a quadratic for $\sqrt{m_\tau}$ with $m_e, m_\mu$ fixed to PDG central values, and $Q = 2/3$ exact: $$ \sqrt{m_\tau}\pm = \tfrac{2}{\sqrt 3}\Bigl(\sqrt{m_e}+\sqrt{m\mu}\Bigr) \pm \sqrt{\tfrac{1}{3}\Bigl(\sqrt{m_e}+\sqrt{m_\mu}\Bigr)^2 + (m_e+m_\mu)}. \qquad (8)$$

Taking the physical (larger) root, $$m_\tau^{\rm pred} = 1776.969 ;\mathrm{MeV}. \qquad (9)$$

PDG 2024: $m_\tau^{\rm meas} = 1776.86 \pm 0.12$ MeV. The deviation is $+0.11$ MeV, $+6\times 10^{-5}$ relative, within 1σ of the experimental uncertainty. Any improvement in $m_\tau$ determination at future $e^+e^-$ τ-charm factories (STCF, Super-τC [14]) will sharpen this test to the $10^{-6}$ level, where a genuine departure from (1) would become visible.

5. Discussion

Link to LiteBIRD. If (7) is structural rather than accidental, the same $\alpha = 2/3$ fixes both the lepton geometry and the inflationary tensor-to-scalar ratio. With $N = 60$ e-folds and $\alpha = 2/3$, (6) gives $r \simeq 0.0022$; fuller treatment including the exact Kähler potential of $X(1)$ and plateau-tail corrections yields $r \simeq 0.0025$ [11], squarely in the LiteBIRD forecast range $\sigma(r) \sim 10^{-3}$ [15]. A detection of $r$ in this band, together with $n_s \simeq 1 - 2/N$, would constitute indirect evidence that the same topological index underlies the Koide relation.

Neutrino sector. The Koide relation with $Q = 2/3$ does not hold for neutrino masses in current oscillation-fit allowed ranges (and the absolute scale is unknown). If the spinor-half-angle picture is correct, neutrinos — which couple to a different chirality projection — should sit at a distinct fixed angle of the same cone construction. Concretely, a Dirac-mass neutrino sector rotated by a further $\pi/4$ (so $\theta_\nu = \pi/2$ or $0$) would predict $Q_\nu \to \infty$ or $Q_\nu = 1/3$; current data marginally favours the $Q_\nu = 1/3$ branch [16], but the absolute masses are needed to settle this. KATRIN [17] and DUNE [18] will be decisive.

Quark sector. For up-type and down-type quarks one measures $Q_u \simeq 0.56$, $Q_d \simeq 0.70$ at the pole (with sizeable QCD/renormalization uncertainty), neither equal to $2/3$ [19]. If the identity (7) has a flavour-blind core, QCD confinement and RG running should deform the naive $\theta = \pi/4$ by calculable amounts controlled by $\alpha_s$ at the relevant scale. This is an open problem; the spinor-geometric viewpoint reframes it as "compute the angular displacement of the √m-vector from π/4 as a function of the strong-coupling fixed point", which may be tractable on the lattice.

Scope. We emphasize that (7) is a strict algebraic identity once one accepts (i) the Koide value $Q = 2/3$ and (ii) the α-attractor value $\alpha = 2/3$. The content of this note is the observation that these two $2/3$'s coincide, and that the mediating geometric object — a spinor rotation by $\pi/4$ — is the natural reading for a spin-½ observer. A derivation of (ii) from a microscopic modular-invariant Lagrangian, and of (i) from a concrete flavour model realizing that Lagrangian, is pursued in [11] and is beyond the scope of this note.

References

[1] Y. Koide, A fermion-boson composite model of quarks and leptons, Phys. Lett. B 120 (1983) 161; New view of quark and lepton mass hierarchy, Phys. Rev. D 28 (1983) 252.

[2] Particle Data Group, R. L. Workman et al., Review of particle physics, PTEP 2024, 083C01.

[3] R. Foot, A note on Koide's lepton mass relation, arXiv:hep-ph/9402242.

[4] N. Li, B.-Q. Ma, Estimate of neutrino masses from Koide's relation, Phys. Lett. B 609 (2005) 309.

[5] Y. Sumino, Family gauge symmetry as an origin of Koide's mass formula and charged lepton spectrum, JHEP 0905:075, 2009.

[6] A. Kartavtsev, A remark on the Koide relation for quarks, arXiv:1111.0480.

[7] Z.-z. Xing, H. Zhang, On the Koide-like relations for running masses of charged leptons, neutrinos, and quarks, Phys. Lett. B 635 (2006) 107.

[8] R. Kallosh, A. Linde, D. Roest, Superconformal inflationary α-attractors, JHEP 1311:198, 2013.

[9] R. Kallosh, A. Linde, Escher in the Sky, CRP 20 (2015) 100.

[10] M. Galante, R. Kallosh, A. Linde, D. Roest, Unity of cosmological inflation attractors, Phys. Rev. Lett. 114, 141302 (2015).

[11] A. Porybný, O., τ-geometry and the complex-real theory (KRT), v1.2.0, internal preprint, 2026. [Main paper, companion to this note.]

[12] D. Zagier, Elliptic modular forms and their applications, in The 1-2-3 of modular forms, Universitext, Springer, 2008.

[13] M. Nakahara, Geometry, topology and physics, 2nd ed., IoP, 2003, §5.10 (double cover SU(2)→SO(3)).

[14] M. Achasov et al., STCF Conceptual Design Report, 2023.

[15] LiteBIRD Collaboration, E. Allys et al., PTEP 2023, 042F01.

[16] C. Brivio, Koide-like relations for neutrino masses: a critical review, arXiv:2104.12493.

[17] KATRIN Collaboration, M. Aker et al., Direct neutrino-mass measurement, Nature Phys. 18 (2022) 160.

[18] DUNE Collaboration, B. Abi et al., Eur. Phys. J. C 80, 978 (2020).

[19] W. Rodejohann, H. Zhang, Extension of an empirical charged lepton mass relation to the neutrino sector, Phys. Lett. B 698 (2011) 152.

Appendix A. Numerical verification of (9). With $m_e = 0.51099895000$ MeV, $m_\mu = 105.6583755$ MeV and $Q = 2/3$ exact, equation (8) gives $\sqrt{m_\tau} = 42.15411$ MeV$^{1/2}$, hence $m_\tau = 1776.969$ MeV. The measured angle of $v$ to the democratic axis is $\theta = 44.99974^\circ$, differing from $\pi/4$ by $2.6\times 10^{-4}$ degrees, or approximately one arcsecond.

Appendix B. Sensitivity. Propagating the PDG uncertainties $\delta m_e = 1.5\times 10^{-8}$ MeV, $\delta m_\mu = 2.3\times 10^{-6}$ MeV through (8) gives $\delta m_\tau^{\rm pred} < 10^{-4}$ MeV, so (9) is limited by the theoretical assumption of exact $Q = 2/3$, not by input precision.