Ternary logic and mass quantum numbers

M. D. Sheppeard
Ka Tiritiri O Te Moana, New Zealand

Abstract. Koide’s prediction of the tau mass may be formulated as a condition on the three
eigenvalues of a quantum Fourier series, using simple parameters, and similar triplets have been
found for neutrino and hadron masses [2]. Assuming these parameters arise from quantum gravity,
one would like to understand them from the more abstract context of category theory. In particular,
whereas the logic of lepton spin is a linear analogue of the ordinary Boolean logic of the category
of sets, mass triplets suggest an analogous ternary logic, requiring higher dimensional categorical
structures.
Keywords: category theory, particle masses, ternary logic, quantum information
PACS: 11.10.Cd, 11.10.Nx, 11.30.Na, 14.60.-z

INTRODUCTION
As quantum numbers, the rest masses of fundamental particles specify a tripling of spin
states to include generation number. In [2] it was shown that triplets of leptons and
hadrons yield phenomenologically viable mass matrices in the normalised Hermitian
form  
√ 1 reiθ re−iθ
1
M =  re−iθ 1 reiθ  , (1)
3 iθ −iθ
re re 1
whose eigenvectors are a quantum Fourier transform matrix and which involve two
parameters. When θ = 2π n/3, r determines a square root mass difference for two
distinct eigenvalues, whereas a more generic θ determines a nondegenerate mass triplet.
When measuring masses with√ respect
√ to the√ natural normalisation scale, the eigenvalues
obey a simple sum rule m1 + m2 + m3 = 1. The Koide rule [1] for the natural
masses of charged leptons then takes the form m1 + m2 + m3 = 2/3.
The quantum information path integral representation of the generations [2] uses the
concept of mutually unbiased bases, the canonical example of which is the set of three
Pauli matrices in dimension 2, representing spin measurements X, Y and Z. Mutually
unbiased bases in prime power dimensions have a categorical description and are closely
connected with representations of finite fields [3]. Roughly speaking, the dimension
of the information space determines the number of elements in the finite field, which
indexes the full set of mutually unbiased bases in a way that requires only a finite set of
complex numbers to define them. This suggests studying finite arithmetic structures in
category theory.
The complex number field has the property that any two distinct phases form a pair of
mutually unbiased bases for the field as a one dimensional space. As a set, the complex
numbers have uncountably many elements. Sets are considered as vector spaces over the
field with one element, suggesting that there exists an alternative description of complex
numbers involving uncountably many unbiased bases. The idea of associating complex
numbers with uncountable matrices also naturally appears in monoidal category theory.
A 3×3 complex mass matrix should therefore be viewed as an infinite reduction of some
higher dimensional operator.
Although it is easy to describe such mass operators in the language of quantum
information theory, we would prefer to understand their origins using the more abstract
mathematics of quantum gravity, namely higher dimensional category theory [5].
The question is, what are the natural categorical axioms associated to such operators?
Here we consider simple diagrams that express the finite geometry of mutually unbiased
bases and mass operators, without preselecting any known categorical axioms. Further
motivation for such an investigation comes from the rich categorical combinatorics of
twistor quantum field theory [4], which uses the recursion of the associahedra polytopes
for massless particle scattering amplitudes.

CATEGORICAL DIAGRAMS AND BINARY LOGIC

Modular arithmetic does not usually play a fundamental role in categorical axioms. It is
said that the categorification of the number 2 is a set with two elements. As an object
in the category of sets, the two element set plays the special role of representing the
two truth values, 0 and 1, of Boolean logic. On the other hand, the number 2, as an
ordinal, is the set with two elements. The lattice of subsets of the two element set takes
the diagrammatic form
! / 01
00
!
² ²
10 / 11

where 00 represents the empty set and 11 the whole set, according to occupancy. The
association of simple logical entities with occupancy extends beyond sets to other object
types, such as Hilbert spaces of dimension 2n , representing n qubits. This is the domain
of monoidal categories [5], with tensor products.
For three qubits, the occupancy of states corresponds to a directed binary cube. Each
face of this cube is a copy of the binary square, fixing one coordinate. The cube has a
source 000 and target 111. In a higher dimensional setting we allow directed 2-arrows
to decorate the faces of a cube, or general polytope law.
The Mac Lane pentagon axiom for monoidal categories may be raised in dimension
so that associator arrows become pseudonatural transformation squares. This pentagon
decorates five sides of a binary cube, but in a general categorical setting one must allow
the sixth side to represent a non trivial 2-arrow. This first occurs in the axioms for
tricategories [6], rarely considered in physics (except for the braided monoidal case)
due to their complexity.
The two element set is also pivotal in describing the fundamental Stone duality
between algebraic and spatial structures, in analogy with the special role played by the
circle U(1) in the theory of the Fourier transform. The arithmetic of mutually unbiased
bases suggests that the extrapolation from two points to the full circle is carried out in
increasing dimension. Note that the categorical limit of the sequence of finite fields F pn ,
for a fixed prime p, is the p-adic integers, and the Fourier dual of this set is the subgroup
of U(1) consisting of all pn -th roots of unity.
So in this game, the subject of binary logic has tentacles reaching into diverse areas of
mathematics. Ternary logic must now mean more than a basic logic of three outcomes.

TERNARY STRUCTURE
The three qutrit analogue of the binary cube for three qubits was first discussed in [7].
Note that 000 (and 111) are the source (and target) of a binary cube, and similarly for
the pairs (111, 222) and (222, 000). There is an arrow 2 → 0, but the reverse only occurs
as the composition of two paths. These three binary cubes form a thickened triangle
with 21 vertices. The 6 other vertices associated to the 27 possible paths are those with
maximally mixed coordinates, namely 012, 120, 201, 021, 102 and 210.
111
q•TTTTTT
Ä Äqq
q T
Ä Ä ÄÄ ÄÄ
P
'Ä'Ä ÄÄ PPP•bbbbbÄÄ ºÄÄ
'' ¼ º
'? ¼¼¼012 ºº
•? ¼ jjjj•º
000 ???? ¼Ä¼ Ä ÄÄ 222
?ÄÄ ÄÄ

Consider specifically the mixed vertex 012. It is the target of 1-arrows from 011, 212
and 002, each a vertex from one of the binary cubes. Similarly, the other mixed vertices
are targets for three 1-arrows from the binary cubes. The triangle of binary cubes, along
with this set of 1-arrows, forms a tetrahedron with source 000 and target 012. Between
maximally mixed paths, such as 012 and 210, there can never be a 1-arrow, since
more than one coordinate must change. However, it is natural to view triples such as
(012, 120, 201) as triangular analogues of the interval. In other words, ternary structure
involves inherently two dimensional objects, in contrast to a simple binary arrow 0 → 1.
The vertex 012 is itself the source of 1-arrows to 112, 022 and 010. The cyclic nature
of 1-arrow rules ensures that there are no true sources and targets in the full ternary
diagram on 27 vertices. Since these vertices are 6-valent, the real embedding dimension
of the full diagram is 6, double that of the binary cube. This is like the doubling of
triangles to squares in the dual polygons for tree diagrams [4], where binary tree vertices
are replaced by ternary tree vertices, which have four legs when a root is included.
The tripling of spin states for generation quantum number would correspond to a
tripling in the basic arrow structure, as in the difference between the two diagrams
00 212FF 002 011
FF xx
FF xxx
FF x
² " ² |xx
01 012F
xx FF
x FF
xxx FF
² |xx ² F"
11 112 022 010
where the arrows for tripled spin may be viewed as state rotation operators rather than
raising operators.
Thus for qutrit experiments, such as particle rest mass, the ternary labels 0, 1 and 2
stand for three possible outcomes. Now allow the categorical 1-arrows to correspond to
modular arithmetic on these indices, as in a basic triangle of successor arrows.

@1=
¢¢¢ ===
¢¢ ==
¢¢ =Á
0o 2

The full three qutrit diagram is unlike the structures usually considered in connection
with axioms for ordinary n-categories, where the associahedra naturally occur. However,
Batanin’s higher dimensional arithmetic accounts for a wider variety of special finite
graphs. Moreover, these generalised categorical diagrams for paths of mutually unbiased
bases are similar to those that occur in the diagrammatic realisation of mutually unbiased
bases for ordinary dagger monoidal categories [8], using Frobenius algebra objects.
The ternary diagram is also suggestive of the exceptional Jordan algebra of 3 × 3
Hermitian matrices over the octonions, for which the off diagonal octonion entries may
be written in terms of units lying on the vertices of a cube. The three real diagonal entries
of a Jordan algebra element may be labeled by the vertices 012, 120 and 201.
If the mass matrices are indeed reductions from a Hermitian Jordan operator, then
the off diagonal binary cubes collapse to a single complex index edge, and perhaps the
phase parameters are derivable from features of the qutrit logic, along with cosmological
considerations.

ACKNOWLEDGMENTS
The author thanks UWA for their hospitality, Carl Brannen for his participation in
FFP10, and also Michael Rios for many interesting comments on M theory and en-
tanglement.

REFERENCES
1. Y. Koide, Nuovo Cimento A 70, 411 (1982).
2. C. A. Brannen, Koide mass equations for hadrons, http://www.brannenworks.com/koidehadrons.pdf
3. M. D. Sheppeard, Arithmetic, MUBs and complementary observables, J. Math. Phys. to appear.
4. N. Arkani-Hamed, F. Cachazo, C. Cheung and J. Kaplan, The S-matrix in twistor space, preprint
hep-th/0903.2110.
5. S. MacLane, Categories for the Working Mathematician, Springer, 2000.
6. R. Gordon, A. J. Power and R. Street, Mem. Amer. Math. Soc. 117, no. 558 (1995).
7. M. Rios, http://uduality.blogspot.com/2007/03/ternary-logic.html
8. B. Coecke and R. Duncan, “Interacting Quantum Observables,” in ICALP (2) 2008, ref
DBLP:conf/icalp/2008-2, pp. 298–310; arxiv.org/abs/0906.4725