Neutrino flavors do not propagate cleanly through space. An electron neutrino produced in the Sun is a quantum-mechanical superposition of three different mass eigenstates, each propagating with a slightly different phase. As the superposition evolves, the probabilities of detecting the neutrino as each of the three flavors oscillate. This is the basic phenomenon of neutrino oscillation, and it requires that flavor eigenstates differ from mass eigenstates.
The mathematical object that describes the relationship between these two bases is the PMNS matrix — named after the four physicists most central to its development: Bruno Pontecorvo, Ziro Maki, Masami Nakagawa, and Shoichi Sakata. The PMNS matrix is the bookkeeping device of neutrino oscillation physics. Every measurable oscillation probability, every CP-violation effect, every prediction about how flavor evolves with distance, is encoded in its entries.
What the matrix actually represents
For three known flavors and three mass eigenstates, the PMNS matrix is 3 × 3. Its entries U_αi describe the probability amplitude for a neutrino of flavor α (electron, muon, or tau) to be in mass eigenstate i (ν₁, ν₂, or ν₃).
An electron neutrino, mathematically, is:
|ν_e⟩ = U_e1 |ν₁⟩ + U_e2 |ν₂⟩ + U_e3 |ν₃⟩
The probability of detecting an electron neutrino later as a muon neutrino — after some propagation time t over distance L — depends on the matrix entries U_ei and U_μi, plus the differences between the squared masses of ν₁, ν₂, and ν₃.
Each row of the matrix sums to one in probability — a neutrino of any flavor must be in some mass eigenstate. Each column sums to one in probability — a neutrino in any mass eigenstate must be detected as some flavor. The matrix is therefore unitary: U⁺U = I.
The parametrization
A general 3×3 unitary matrix has nine real parameters. Removing unphysical phases (those that can be absorbed into the definitions of the basis states) leaves four physical parameters: three mixing angles and one CP-violating phase. This is the same parameter count as the CKM matrix in the quark sector.
The conventional parametrization writes the PMNS matrix as a product of three rotation matrices, each parametrized by a single angle, plus a phase factor for CP violation:
U_PMNS = R₂₃ × R₁₃(δ_CP) × R₁₂
Where R_ij is a rotation in the i-j plane parametrized by the angle θ_ij, and δ_CP enters the R₁₃ block as a complex phase.
This decomposition is convenient because each rotation corresponds to a different physical context.
R₁₂ and θ₁₂. Dominate solar-neutrino oscillation, where the two lighter mass states ν₁ and ν₂ are most relevant. Measured primarily by SNO and KamLAND. Current value θ₁₂ ≈ 33.4°.
R₂₃ and θ₂₃. Dominate atmospheric-neutrino oscillation, where ν₂ and ν₃ mix. Measured primarily by Super-Kamiokande, T2K, NOvA, and IceCube. Current value θ₂₃ ≈ 49° — close to maximal mixing.
R₁₃ and θ₁₃. The smallest of the three angles. Important for CP-violation phenomena. Measured primarily by Daya Bay and RENO. Current value θ₁₃ ≈ 8.6°.
δ_CP. The CP-violating phase. Currently being measured by T2K and NOvA, with DUNE and Hyper-K coming online to nail it down.
The Majorana phases
If neutrinos are Majorana particles rather than Dirac, the PMNS matrix has two additional phases that do not appear in the Dirac case. These are typically called α₁ and α₂.
The Majorana phases do not affect ordinary oscillation measurements. They only enter into lepton-number-violating processes, primarily neutrinoless double-beta decay. The effective Majorana mass that 0νββ experiments measure depends on combinations of m₁, m₂, m₃, and the Majorana phases.
This is why a 0νββ discovery would constrain both the absolute neutrino mass scale and the otherwise-invisible Majorana phases — although extracting them cleanly is hard because the underlying nuclear matrix elements have large theoretical uncertainties.
What the measured values tell us
The pattern of measured PMNS parameters is striking and not predicted by any obvious theoretical framework.
θ₂₃ is close to 45° — maximal mixing. This means that the ν_μ and ν_τ flavors are almost equal superpositions of the heavier mass eigenstates. There is no consensus on why this is the case, but it is a clean experimental finding.
θ₁₂ is moderately large at 33° — well above the corresponding quark mixing angles (Cabibbo angle is about 13°) but not maximal. There are various proposed explanations from flavor-symmetry models, but no widely accepted one.
θ₁₃ is small but not extraordinarily small at about 9°. Before the Daya Bay/RENO measurement, many theorists had expected θ₁₃ to be much smaller — consistent with zero in some simple models. The “natural” value at 9° makes CP-violation studies more accessible than they would have been at, say, 1°.
The CP phase δ_CP is being measured. The current preferred value is loosely around -90°, but it has not been pinned down at high significance.
Compared to the quark sector, where the CKM matrix is close to diagonal with small off-diagonal entries, the PMNS matrix is strikingly anti-diagonal-like. The reason for this difference is one of the open questions of flavor physics.
Why the matrix is not the whole story
The PMNS matrix tells you how flavor eigenstates relate to mass eigenstates. But it doesn’t tell you several other essential things.
The absolute mass scale. The PMNS matrix tells you about mass differences, not absolute masses. The absolute mass scale is constrained by direct measurements (KATRIN) and cosmological measurements (CMB + DESI), but is not encoded in the PMNS matrix itself.
The mass ordering. The PMNS matrix tells you about mass-squared differences, but the sign of one of them — whether m₃ > m₂ or m₃ < m₂ — is still being measured. This is the normal versus inverted ordering question.
Whether neutrinos are Dirac or Majorana. The mixing matrix structure is the same in either case (apart from the two extra Majorana phases). The Dirac-versus-Majorana question is independent and must be answered through different experiments.
Whether there are sterile partners. A fourth or further neutrino, mixing with the three active ones, would extend the matrix to 4×4 or larger. The PMNS matrix as conventionally defined describes only the three active flavors.
The PMNS matrix in tabular form
The current best-fit values of the standard parametrization are roughly:
| Parameter | Value | Source |
|---|---|---|
| sin²(θ₁₂) | 0.307 | Solar + KamLAND |
| sin²(θ₂₃) | 0.572 | Atmospheric + accelerator |
| sin²(θ₁₃) | 0.0220 | Daya Bay + RENO |
| δ_CP | -90° (preliminary) | T2K + NOvA |
| Δm²₂₁ | 7.5 × 10⁻⁵ eV² | Solar + KamLAND |
| Δm²₃₁ |
The sign of Δm²₃₁ is currently unknown — that is the mass-ordering question — but its magnitude is solid. Combined fits including all neutrino data put each parameter on its own precision level. The most recent global analyses are quoted in the Particle Data Group’s review.
What the matrix means for the field
The PMNS matrix is the central data structure of three-flavor neutrino oscillation physics. Every long-baseline experiment, every reactor experiment, every solar or atmospheric measurement, contributes to constraining its parameters. Every theoretical proposal about why neutrinos have mass and mix the way they do must produce a PMNS matrix consistent with observation.
The next decade of neutrino physics is largely about completing the PMNS measurement. The remaining unknowns are:
- The CP-violating phase δ_CP (current 2σ-level hints, awaiting decisive results from DUNE and Hyper-K).
- The mass ordering (current 2-3σ-level preference for normal, awaiting JUNO, DUNE, Hyper-K).
- The absolute mass scale (awaiting KATRIN and successor experiments, plus cosmology).
- Whether the matrix is exactly 3×3 unitary, or whether sterile neutrinos extend it.
When these questions are resolved, the PMNS matrix will be one of the most thoroughly measured objects in particle physics — a precise empirical input that any theoretical framework for flavor physics has to reproduce. Whether nature provides a clean theoretical explanation for the observed pattern, or whether the structure is essentially “just the way it is,” is something the next generation of experiments will help decide.
A matrix that took half a century
The structure that became the PMNS matrix was first written down in the early 1960s — by Pontecorvo in 1957 (with a two-flavor version), and by Maki, Nakagawa, and Sakata in 1962 (extending it to three flavors). At the time, it was a theoretical possibility with no experimental support. The first measurement of any of its parameters did not happen until decades later.
Now, in 2026, we know three of its four physical parameters to good precision, are actively measuring the fourth, and have a clear roadmap for finishing the job within the next decade. It is one of the longest arcs from theoretical proposal to precision experimental verification in the history of particle physics.
The bookkeeping of how neutrino flavor mixes into mass eigenstates is, in the end, just six numbers — three angles, one phase, and two more if Majorana. But fitting those six numbers to the data has required a half-century of experimental effort and continues to be the central enterprise of the field.
For the underlying oscillation physics, see How neutrino oscillation works. For the experimental campaign measuring each parameter, see Daya Bay and RENO for θ₁₃, Super-Kamiokande for θ₂₃, and SNO for θ₁₂. For the CP phase being measured, see CP violation in the neutrino sector. For the mass-ordering question, see Normal or inverted.
Further reading
Primary sources
- Z. Maki, M. Nakagawa, and S. Sakata, “Remarks on the Unified Model of Elementary Particles”, Prog. Theor. Phys. 28:870 (1962) — the three-flavor mixing paper
- B. Pontecorvo, “Mesonium and antimesonium,” Sov. Phys. JETP 6:429 (1958) — the original two-flavor oscillation proposal
Background and context
- NuFIT global fit (nu-fit.org) — continuously updated global analysis of all PMNS parameters
- Particle Data Group — Neutrino Masses, Mixing, and Oscillations — the canonical review
- Wikipedia: Pontecorvo–Maki–Nakagawa–Sakata matrix
Frequently asked
What is the PMNS matrix?
The PMNS matrix — named after Pontecorvo, Maki, Nakagawa, and Sakata — is a 3-by-3 unitary matrix that describes how neutrino flavor eigenstates (electron, muon, tau) relate to neutrino mass eigenstates. It is the leptonic analog of the CKM matrix in the quark sector and is the central bookkeeping tool of neutrino oscillation physics.
How many parameters does the PMNS matrix have?
If neutrinos are Dirac particles, the matrix has four free physical parameters: three mixing angles (θ₁₂, θ₂₃, θ₁₃) and one CP-violating phase δ_CP. If neutrinos are Majorana particles, there are two additional Majorana phases (α₁, α₂), bringing the total to six. The two additional phases only affect lepton-number-violating processes like neutrinoless double-beta decay, not standard oscillation measurements.
What are the current best values of the mixing angles?
θ₁₂ ≈ 33.4° (solar mixing angle, measured primarily by SNO and KamLAND); θ₂₃ ≈ 49° (atmospheric mixing angle, measured by Super-Kamiokande, T2K, NOvA, IceCube); θ₁₃ ≈ 8.6° (the smallest, measured primarily by Daya Bay and RENO). All three have been measured to roughly 5-10% precision, far better than the corresponding CKM angles in the quark sector were known a decade ago.
How does the PMNS differ from the CKM matrix?
Both are 3-by-3 unitary mixing matrices. But the PMNS has much larger mixing angles — the leptonic sector is roughly maximal-mixing while the quark sector is near-diagonal. The PMNS may also have additional Majorana phases the CKM does not have. Theoretical models try to explain why these mixing patterns are so different, but no consensus picture has emerged.
Why is the PMNS structure considered surprising?
Before the discovery of large mixing in atmospheric and solar neutrinos, most theorists expected the leptonic mixing to be small, like the quark mixing. The observed pattern — with one angle near maximum (θ₂₃), one moderately large (θ₁₂), and one small (θ₁₃) — is unusual and is one of the active research questions in flavor physics: why does nature pick this pattern of mixings?
Cite this article 5 formats
APA
Neutrino Times Editorial Team. (2025, November 28). The PMNS matrix: the bookkeeping of neutrino mixing. Neutrino Times. https://neutrino-times.com/articles/pmns-matrix-bookkeeping-of-neutrino-mixing/
Chicago
Neutrino Times Editorial Team. "The PMNS matrix: the bookkeeping of neutrino mixing." Neutrino Times, November 28, 2025. https://neutrino-times.com/articles/pmns-matrix-bookkeeping-of-neutrino-mixing/.
MLA
Neutrino Times Editorial Team. "The PMNS matrix: the bookkeeping of neutrino mixing." Neutrino Times, 28 Nov. 2025, https://neutrino-times.com/articles/pmns-matrix-bookkeeping-of-neutrino-mixing/.
BibTeX
@misc{neutrino-times-pmns-matrix-bookkeeping-of-neutrino-mixing,
author = {Neutrino Times Editorial Team},
title = {The PMNS matrix: the bookkeeping of neutrino mixing},
howpublished = {Neutrino Times},
year = {2025},
month = {nov},
url = {https://neutrino-times.com/articles/pmns-matrix-bookkeeping-of-neutrino-mixing/},
note = {Accessed: 2025-11-28}
} RIS
TY - GEN TI - The PMNS matrix: the bookkeeping of neutrino mixing AU - Neutrino Times Editorial Team PY - 2025 DA - 2025-11-28 PB - Neutrino Times UR - https://neutrino-times.com/articles/pmns-matrix-bookkeeping-of-neutrino-mixing/ ER -