Polyadic integer numbers and finite (m,n) fields

arXiv:1707.00719v1[math.RA]3Jul2017
POLYADIC INTEGER NUMBERS AND FINITE (m, n)-FIELDS
STEVEN DUPLIJ
ABSTRACT. The polyadic integer numbers, which form a polyadic ring, are representatives of a fixed
congruence class. The basics of polyadic arithmetic are presented: prime polyadic numbers, the
polyadic Euler function, polyadic division with a remainder, etc. are introduced. Secondary congruence
classes of polyadic integer numbers, which become ordinary residue classes in the binary limit, and the
corresponding finite polyadic rings are defined. Polyadic versions of (prime) finite fields are introduced.
These can be zeroless, zeroless and nonunital, or have several units; it is even possible for all of their
elements to be units. There exist non-isomorphic finite polyadic fields of the same arity shape and or-
der. None of the above situations is possible in the binary case. It is conjectured that any finite polyadic
field should contain a certain canonical prime polyadic field as a smallest finite subfield, which can be
considered a polyadic analogue of GF (p).
CONTENTS
INTRODUCTION 2
1. PRELIMINARIES 3
2. RING OF POLYADIC INTEGER NUMBERS 4
2.1. External and internal operations for congruence classes 4
2.2. Prime polyadic integer numbers 6
2.3. The parameters-to-arity mapping 10
3. FINITE POLYADIC RINGS 11
3.1. Secondary congruence classes 11
3.2. Finite polyadic rings of secondary classes 12
4. FINITE POLYADIC FIELDS 14
4.1. Abstract finite polyadic fields 14
4.2. Multiplicative structure 16
5. CONCLUSIONS 23
ACKNOWLEDGEMENTS 23
APPENDIX. Multiplicative properties of exotic finite polyadic fields 23
REFERENCES 25
LIST OF TABLES 25
Date: July 3, 2017.
2010 Mathematics Subject Classification. 16T05, 16T25, 17A42, 20N15, 20F29, 20G05, 20G42, 57T05.

2 STEVEN DUPLIJ
In Galois Fields, full of flowers
primitive elements dance for hours
climbing sequentially through the trees
and shouting occasional parities.
S.B. Weinstein, IEEE Trans. 1971
INTRODUCTION
The theory of finite fields LIDL AND NIEDERREITER [1997] plays a very important role. From
one side, it acts as a “gluing particle” connecting algebra, combinatorics and number theory (see,
e.g. MULLEN AND PANARIO [2013]), and from another it has numerous applications to “reality”: in
coding theory, cryptography and computer science MENEZES ET AL. [1997]. Therefore, any gener-
alization or variation of its initial statements can lead to interesting and useful consequences for both
of the above. There are two principal peculiarities of finite fields: 1) Uniqueness - they can have only
special numbers of elements (the order is any power of a prime integer pr
) and this fully determines
them, in that all finite fields of the same order are isomorphic; 2) Existence of their “smallest” (prime)
finite subfield of order p, which is isomorphic to the congruence class of integers Z pZ. Investiga-
tion of the latter is a bridge to the study of all finite fields, since they act as building blocks of the
extended (that is, all) finite fields.
We propose a special - polyadic - version of the (prime) finite fields in such a way that, instead of the
binary ring of integers Z, we consider a polyadic ring. The concept of the polyadic integer numbers
Z(m,n) as representatives of a fixed congruence class, which form the (m, n)-ring (with m-ary addition
and n-ary multiplication), was introduced in DUPLIJ [2017]. Here we analyze Z(m,n) in more detail,
by developing elements of a polyadic analog of binary arithmetic: polyadic prime numbers, polyadic
division with a remainder, the polyadic Euler totient function, etc. ... It is important to stress that the
polyadic integer numbers are special variables (we use superscripts for them) which in general have
no connection with ordinary integers (despite the similar notation used in computations), because
the former satisfy different relations, and coincide with the latter in the binary limit only. Next we
will define new secondary congruence classes and the corresponding finite (m, n)-rings Z(m,n) (q)
of polyadic integer numbers, which give Z qZ in the binary limit. The conditions under which
these rings become fields are given, and the corresponding “abstract” polyadic fields are defined
and classified using their idempotence polyadic order. They have unusual properties, and can be
zeroless, zeroless-nonunital or have several units, and it is even possible for all elements to be units.
The subgroup structure of their (cyclic) multiplicative finite n-ary group is analyzed in detail. For
some zeroless finite polyadic fields their multiplicative n-ary group is a non-intersecting union of
subgroups. It is shown that there exist non-isomorphic finite polyadic fields of the same arity shape
and order. None of the above situations is possible in the binary case.
Some general properties of polyadic rings and fields were given in CROMBEZ [1972],
LEESON AND BUTSON [1980], POP AND POP [2002], DUPLIJ AND WERNER [2015], but their
concrete examples using integers differ considerably from our construction here, and the latter leads
to so called nonderived (proper) versions which have not been considered before.
We conjecture that any (m, n)-field contains as a subfield one of the prime polyadic fields con-
structed here, which can be considered as a polyadic analog of GF (p).

POLYADIC INTEGER NUMBERS AND FINITE (m, n)-FIELDS 3
1. PRELIMINARIES
We use the notations and definitions from DUPLIJ [2012, 2017] (see, also, references therein). We
recall (only for self-consistency) some important elements and facts about polyadic rings, which will
be needed below.
Informally, a polyadic (m, n)-ring is Rm,n = R | νm, µn , where R is a set, equipped with m-ary
addition νm : Rm
→ R and n-ary multiplication µn : Rn
→ R which are connected by the polyadic
distributive law, such that R | νm is a commutative m-ary group and R | µn is a semigroup. A
commutative (cancellative) polyadic ring has a commutative (cancellative) n-ary multiplication µn. A
polyadic ring is called derived, if νm and µn are equivalent to a repetition of the binary addition and
multiplication, while R | + and R | · are commutative (binary) group and semigroup respectively.
If only one operation νm (or µn) has this property, we call such a Rm,n additively (or multiplicatively)
derived (half-derived).
In distinction to binary rings, an n-admissible “length of word (x)” should be congruent to
1 mod (n − 1), containing ℓµ (n − 1) + 1 elements (ℓµ is a “number of multiplications”) µ
(ℓµ)
n [x]
(x ∈ Rℓµ(n−1)+1
), so called (ℓµ (n − 1) + 1)-ads, or polyads. An m-admissible “quantity of words
(y)” in a polyadic “sum” has to be congruent to 1 mod (m − 1), i.e. consisting of ℓν (m − 1) + 1
summands (ℓν is a “number of additions”) ν
(ℓν)
m [y] (y ∈ Rℓν(m−1)+1
). Therefore, a straightforward
“polyadization” of any binary expression (m = n = 2) can be introduced as follows: substitute the
number of multipliers ℓµ + 1 → ℓµ (n − 1) + 1 and number of summands ℓν + 1 → ℓν (m − 1) + 1,
respectively.
An example of “trivial polyadization” is the simplest (m, n)-ring derived from the ring of integers
Z as the set of ℓν (m − 1)+1 “sums” of n-admissible (ℓµ (n − 1) + 1)-ads (x), where x ∈ Zℓµ(n−1)+1
LEESON AND BUTSON [1980].
The additive m-ary polyadic power and the multiplicative n-ary polyadic power are defined by (inside
polyadic products we denote repeated entries by
k
x, . . . , x as xk
)
x ℓν +m = ν(ℓν )
m xℓν (m−1)+1
, x ℓµ ×n = µ(ℓµ)
n xℓµ(n−1)+1
, x ∈ R, (1.1)
such that the polyadic powers and ordinary powers differ by one: x ℓν +2 = xℓν+1
, x ℓµ ×2 = xℓµ+1
.
The polyadic idempotents in Rm,n satisfy
x ℓν +m = x, x ℓµ ×n = x, (1.2)
and are called the additive ℓν-idempotent and the multiplicative ℓµ-idempotent, respectively.
The additive 1-idempotent, the zero z ∈ R, is (if it exists) defined by
νm [x, z] = z, ∀x ∈ Rm−1
. (1.3)
An element x ∈ R is called (polyadic) nilpotent, if x 1 +m = z, and all higher powers of a nilpotent
element are nilpotent, as follows from (1.3) and associativity.
The unit e of Rm,n is a multiplicative 1-idempotent which is defined (if it exists) as
µn en−1
, x = x, ∀x ∈ R, (1.4)
where (in case of a noncommutative polyadic ring) x can be on any place. An element x ∈ R is called
a (polyadic) ℓµ-reflection, if x ℓµ ×n = e (multiplicative analog of a nilpotent element).
Polyadic rings with zero or unit(s) are called additively or multiplicatively half-derived, and derived
rings have a zero and unit(s) simultaneously. There are polyadic rings which have no unit and no zero,
or with several units and no zero, or where all elements are units. But if a zero exists, it is unique. If
a polyadic ring contains no unit and no zero, we call it a zeroless nonunital polyadic ring. It is obvious
that zeroless nonunital rings can contain other idempotents of higher polyadic powers.

4 STEVEN DUPLIJ
So, in polyadic rings (including the zeroless nonunital ones) invertibility can be governed in a way
which is not connected with unit and zero elements. For a fixed element x ∈ R its additive querelement
˜x and multiplicative querelement ¯x are defined by
νm xm−1
, ˜x = x, µn xn−1
, ¯x = x, (1.5)
where in the second equation, if the n-ary multiplication µn is noncommutative, ¯x can be on any
place. Because R | νm is a commutative group, each x ∈ R has its additive querelement ˜x (and
is querable or “polyadically invertible”). The n-ary semigroup R | µn can have no multiplicatively
querable elements at all. However, if every x ∈ R has its unique querelement, then R | µn is an
n-ary group. Obviously, that n-ary group cannot have nilpotent elements, but can have ℓµ-reflections.
Denote R∗
= R {z}, if the zero z exists. If R∗
| µn is the n-ary group, then Rm,n is a (m, n)-
division ring.
Definition 1.1. A commutative (m, n)-division ring Rm,n is a (m, n)-field Fm,n.
The simplest example of a (m, n)-field derived from R is the set of ℓν (m − 1) + 1 “sums” of
admissible (ℓµ (n − 1) + 1)-ads (x), where x ∈ Rℓµ(n−1)+1
. Some nonderived (m, n)-fields are in
Example 1.2. a) The set iR with i2
= −1 is a (2, 3)-field with a zero and no unit (operations are made
in C), but the multiplicative querelement of ix is −i x (x = 0).
b) The set of fractions ix/y | x, y ∈ Zodd
, i2
= −1 is a (3, 3)-field with no zero and no unit
(operations are in C), while the additive and multiplicative querelements of ix/y are −ix/y and
−iy/x, respectively.
c) The set of antidiagonal 2 × 2 matrices over R is a (2, 3)-field with zero z =
0 0
0 0
and two
units e = ±
0 1
1 0
, but the unique querelement of
0 x
y 0
is
0 1/y
1/x 0
.
2. RING OF POLYADIC INTEGER NUMBERS
Recall the notion of the ring of polyadic integer numbers Z(m,n) which was introduced in DUPLIJ
[2017], where its difference from the (m, n)-ring of integers from LEESON AND BUTSON [1980]
was outlined.
Let us consider a congruence class (residue class) of an integer a modulo b
[[a]]b = {{a + bk} | k ∈ Z, a ∈ Z+, b ∈ N, 0 ≤ a ≤ b − 1} . (2.1)
We denote a representative element by xk = x
[a,b]
k = a + bk, where obviously {xk} is an infinite set.
2.1. External and internal operations for congruence classes. Informally, there are two ways to
equip (2.1) with operations:
1) The “External” way is to define (binary) operations between the congruence classes. Let us
define on the finite underlying set of b congruence classes {[[a]]b}, a = 0, 1, . . . , b − 1 the fol-
lowing new binary operations (here, if b is fixed, and we denote the binary class representative
by an integer with one prime [[a]]b ≡ a′
, as well as the corresponding binary operations +′
, ·′
between classes)
a′
1 +′
a′
2 = (a1 + a2)′
, (2.2)
a′
1 ·′
a′
2 = (a1a2)′
. (2.3)
Then, the binary residue class ring is defined by
Z bZ = {{a′
} | +′
, ·′
, 0′
, 1′
} . (2.4)
In the case of prime b = p, the ring Z pZ becomes a binary finite field having p elements.

2) The “Internal” way is to introduce (polyadic) operations inside a given class [[a]]b (with both
a and b fixed). We introduce the commutative m-ary addition and commutative n-ary multipli-
cation of representatives xki
of the fixed congruence class by
νm [xk1 , xk2 , . . . , xkm ] = xk1 + xk2 + . . . + xkm , (2.5)
µn [xk1 , xk2 , . . . , xkn ] = xk1 xk2 . . . xkn , xki
∈ [[a]]b , ki ∈ Z. (2.6)
In general, the binary sums xk1 + xk2 and products xk1 xk2 are not in [[a]]b.
Proposition 2.1 ( DUPLIJ [2017]). The polyadic operations νm and µn become closed in [[a]]b,
if the arities (m, n) have the minimal values satisfying
ma ≡ a (mod b) , (2.7)
an
≡ a (mod b) . (2.8)
Polyadic distributivity is inherited from that of Z, and therefore we have
Definition 2.2 ( DUPLIJ [2017]). The congruence class [[a]]b equipped with a structure of non-
derived infinite commutative polyadic ring is called a (m, n)-ring of polyadic integer numbers
Z(m,n) ≡ Z
[a,b]
(m,n) = {[[a]]b | νm, µn} . (2.9)
Obviously, Z(m,n) (as in the binary case) cannot become a polyadic field with any choice of
parameters.
Example 2.3. In the residue class
[[3]]4 = {. . . − 25, −21, −17, −13, −9, −5, −1, 3, 7, 11, 15, 19, 23, 27, 31, 35, 39 . . .} (2.10)
we can add only 4ℓν + 1 representatives and multiply 2ℓµ + 1 representatives (ℓν, ℓµ are “numbers”
of m-ary additions and n-ary multiplications respectively) to retain the same class, e.g., take ℓν = 2,
ℓµ = 3 to get (7 + 11 + 15 + 19 + 23)−5−9−13−1 = 47 ∈ [[3]]4, ((7 · 3 · 11) · 19 · 15)·31·27 =
55 103 895 ∈ [[3]]4. Obviously, we cannot add and multiply arbitrary quantities of numbers in [[3]]4,
only the admissible ones. This means that [[3]]4 is the polyadic (5, 3)-ring Z(5,3) = Z
[3,4]
(5,3).
Remark 2.4. After imposing the operations (2.5)–(2.6) the representatives x
[a,b]
k become abstract vari-
ables (polyadic integer numbers, elements of the corresponding (m, n)-ring) which are not ordinary
integers (forming a (2, 2)-ring), but have the latter as the binary limit. So in computations the inte-
gral numbers (denoting representatives) should carry their arity shape (m, n) as additional indices.
Indeed, the representative, e.g. 3 = 3(5,3) ∈ Z
[3,4]
(5,3) is different from 3 = 3(1,2) ∈ Z
[1,2]
(3,2), i.e. properly
speaking 3 = 3(5,3) = 3(1,2) = 3, since their multiplication and addition are different, as they belong
to different polyadic rings. For conciseness, here we omit the indices (m, n), if their value is clear
from the context.
Thus, at first sight it seems that one can obtain a polyadic field only in the “external” way, i.e.
using the “trivial polyadization” of the binary finite field Z pZ (just a repetition of the binary group
operations). However, this leads to the derived polyadic finite fields having a very simple struc-
ture, in which the admissible binary sums and binary products of the congruence classes are used
LEESON AND BUTSON [1980]. In the next section we propose a special approach to construct the
nonderived finite (m, n)-fields of polyadic integer numbers Z(m,n).
Remark 2.5. If n = b = p is prime, then (2.8) is valid for any a ∈ N, which is another formulation of
Fermat’s little theorem.

6 STEVEN DUPLIJ
2.2. Prime polyadic integer numbers. Let us introduce a polyadic analog of prime numbers in
Z(m,n). First we need
Definition 2.6. A polyadically composite (reducible) number is xk ∈ Z(m,n), such that the expansion
xk = µ(ℓ)
n xk1 , xk2 , , xkℓ(n−1)+1
, xki
∈ Z(m,n), (2.11)
is unique, where ℓ is a number of n-ary multiplications, and there exist at least one xki
= xk and
xki
= e (i.e. is not equal to unit of Z(m,n), if it exists). Denote the set of such numbers {xki
} = D (xk)
which is called the composition set of xk.
Definition 2.7. An irreducible polyadic number is xk ∈ Z(m,n) cannot be expressed as any (long)
polyadic product (2.11).
Proposition 2.8. In the polyadic ring Z
[a,b]
(m,n) without the unit the elements satisfying
− |a − b|n
< xk < |a − b|n
(2.12)
are irreducible.
Proof. Since 0 ≤ a ≤ b − 1, the minimal absolute value of an element xk ∈ Z
[a,b]
(m,n) is |a − b|. The
minimum of its n-ary product is |a − b|n
, and therefore smaller elements cannot be decomposed.
Example 2.9. In the (6, 5)-ring Z
[8,10]
(6,5) all polyadic integer numbers are even, and there is no unit, and
so they are binary composite
Z
[8,10]
(6,5) = {. . . − 72, −62, −52, −42, −32, −22, −12, −2, 8, 18, 28, 38, 48, 58, . . .} (2.13)
Nevertheless, the lowest elements, e.g. {−22, −12, 8, 18, 28}, are irreducible, while the smallest (by
absolute value) polyadically composite element is (−32) = µ5 (−2)5
.
Definition 2.10. A range in which all elements are indecomposable is called a polyadic irreducible gap.
Remark 2.11. We do not demand positivity, as in the binary case, because polyadic integer numbers
Z
[a,b]
(m,n) (2.9) are “symmetric” not with respect to x = 0, but under x = xk=0 = a.
The polyadic analog of binary prime numbers plays an intermediate role between composite and
irreducible elements.
Definition 2.12. A polyadic prime number is xkp ∈ Z(m,n), such that it obeys only the unique expansion
xkp = µ(ℓ)
n xkp , eℓ(n−1)
, (2.14)
where e a polyadic unit of Z(m,n) (if exists).
So, the polyadic prime numbers can appear only in those polyadic rings Z
[a,b]
(m,n) which contain
units. In DUPLIJ [2017] (Proposition 6.15) it was shown that such rings correspond to the limiting
congruence classes [[1]]b and [[b − 1]]b, and indeed only for them can a + bk = 1 mod b, and eℓ(n−1)
can be a neutral sequence (for e = 1 always, while for e = −1 only when ℓ (n − 1) is even).
Proposition 2.13. The prime polyadic numbers can exist only in the limiting polyadic rings Z
[1,b]
(b+1,2)
and Z
[b−1,b]
(b+1,3).
Proof. The equation a + bk = 1 mod b (for 0 ≤ a ≤ b − 1) has two solutions: a = 1 and a = b − 1
corresponding for two limiting congruence classes [[1]]b and [[b − 1]]b, which correspond to
x+
k = bk + 1, (2.15)
x−
k = b (k + 1) − 1, k ∈ Z. (2.16)

The parameters-to-arity mapping (2.41) ﬁxes their multiplication arity to n = 2 and n = 3 respec-
tively, which gives manifestly
µ2 x+
k1
, x+
k2
= x+
k , k = bk1k2 + k1 + k2, (2.17)
µ3 x−
k1
, x−
k2
, x−
k3
= x−
k , k, ki ∈ Z, i = 1, 2, 3, b ∈ N,
k = b2
k1k2k3 + (b − 1) [b (k1k2 + k2k3 + k1k3) + (b − 1) (k1 + k2 + k3) + (b − 2)] . (2.18)
Therefore, for Z
[1,b]
(b+1,2) we have the unit e = x+
k=0 = 1 (which is obvious for the binary multiplication),
while in Z
[b−1,b]
(b+1,3) the unit is e = x−
k=−1 = −1 (b ≥ 3), and the sequence is (e2
) is evidently neutral.
Denote the set of ordinary binary prime numbers in the interval 1 ≤ k ≤ kmax by P (kmax), kmax ∈
N. The set of prime polyadic numbers for the polyadic ring Z
[a,b]
(m,n) in the interval x−kmax ≤ xk ≤ xkmax
is denoted by P
[a,b]
(m,n) (kmax). Obviously, in the binary limit P
[0,1]
(2,2) (kmax) = P (kmax) ∪ {−P (kmax)}.
Nevertheless, prime polyadic numbers can be composite as binary numbers.
Assertion 2.14. The set of prime polyadic numbers in the interval x−kmax ≤ xk ≤ xkmax for Z
[a,b]
(m,n)
can contain composite binary numbers, i.e.
∆P
[a,b]
(m,n) (kmax) = P
[a,b]
(m,n) (kmax) {{P (xkmax ) ∪ {P (−x−kmax )}} ∩ [[a]]b} = ∅. (2.19)
Deﬁnition 2.15. 1) The cardinality of the set of ordinary binary prime numbers P (kmax) is called a
prime-counting function and denoted by π (kmax) = |P (kmax)|.
2) The cardinality of the set of prime polyadic numbers P
[a,b]
(m,n) (kmax) is called a polyadic prime-
counting function and denoted by
π
[a,b]
(m,n) (kmax) = P
[a,b]
(m,n) (kmax) . (2.20)
Example 2.16. 1) Consider Z
[43,44]
(45,3) and kmax = 2, then
P
[43,44]
(45,3) (2) = {−45, −1, 43, 87, 131}, (2.21)
∆P
[43,44]
(45,3) (2) = {−45} , π
[43,44]
(45,3) (2) = 5. (2.22)
2) For Z
[50,51]
(52,3) and kmax = 5 we have
P
[50,51]
(52,3) (5) = {−205, −154, −103, −52, −1, 50, 101, 152, 203, 254, 305}, (2.23)
∆P
[50,51]
(52,3) (5) = {−205, −154, −52, 50, 152, 203, 254, 305}, π
[50,51]
(52,3) (5) = 11. (2.24)
Remark 2.17. This happens because in Z
[a,b]
(m,n) the role of “building blocks” (prime polyadic numbers)
is played by those xk which cannot be presented as a (long) ternary product of other polyadic integer
numbers from the same Z
[a,b]
(m,n) as in (2.11), but which satisfy (2.14) only. Nevertheless, such prime
polyadic numbers can be composite binary prime numbers.
In general, for the limiting cases, in which polyadic prime numbers exist, we have
Proposition 2.18. 1) In Z
[1,b]
(b+1,2) the “smallest” polyadic integer numbers satisfying
−b < kp < b + 2, (2.25)
1 − b2
< xkp < (b + 1)2
, (2.26)
are not decomposable, and therefore such xkp ∈ Z
[1,b]
(b+1,2) are all polyadic prime numbers.

8 STEVEN DUPLIJ
2) For another limiting case Z
[b−1,b]
(b+1,3) the polyadic integer numbers satisfying
1 − b < kp < b − 1, (2.27)
− (b − 1)2
< xkp < b2
− 1, (2.28)
are not ternary decomposable and so all such xkp ∈ Z
[b−1,b]
(b+1,2) are polyadic prime numbers.
Proof. This follows from determining the maximum of the negative values and the minimum of the
positive values of the functions x+
k and x−
k in (2.17)–(2.18).
Definition 2.19. The range in which all elements are polyadically prime numbers is called the polyadic
primes gap, and for the two limiting cases it is given by (2.26) and (2.28), respectively.
For instance, in Z
[50,51]
(52,3) for the polyadic primes gap we have −2500 < xkp < 2600: all such
polyadic integer numbers are polyadically prime, but there are many composite binary numbers
among them.
In the same way we can introduce a polyadic analog of the Euler (totient) function which in the
binary case counts the number of coprimes to a given natural number. Denote the set of ordinary
binary numbers k > 1 which are coprime to kmax ∈ N by S (kmax) (named totatives of kmax). Then,
the cardinality of S (kmax) is defined as Euler function ϕ (kmax) = |S (kmax)|. Obviously, if kmax = p
is prime, then ϕ (p) = p − 1. The notion of coprime numbers is based on the divisors: the coprime
numbers k1 and k2 have the greatest common divisor gcd (k1, k2) = 1. In the polyadic case it is not
so straightforward, and we need to start from the basic definitions.
First, we observe that in a (commutative) polyadic ring Rm,n = {R | νm, µn} the analog of the
division operation is usually not defined uniquely, which makes it useless for real applications. In-
deed, y divides x, where x, y ∈ R, if there exists a sequence z ∈ Rn−1
of length (n − 1), such that
x = µn [y, z]. To be consistent with the ordinary integer numbers Z, we demand in the polyadic
number ring Z(m,n): 1) Uniqueness of the result; 2) i.e. only one polyadic number (not a sequence)
as the result. This naturally leads to
Definition 2.20. A polyadic number (quotient) xk2 polyadically divides a polyadic number (dividend)
xk1 , if there exists xkq := xk1 ÷p xk2 , called the (unique) result of division, such that
xk1 = µn xk2 , xkq
n−1
, xk1 , xk2 , xkq ∈ Z(m,n). (2.29)
Remark 2.21. For polyadic prime numbers (2.14) the only possibility for the quotient is xk2 = xk1
such that xk1 = µn xk1 , (e)n−1
or xk1 ÷p xk1 = e, where e is the unit of Z(m,n).
Assertion 2.22. Polyadic division is distributive from the left
νm [xk1 , xk2 , . . . , xkm ] ÷p xk = νm [(xk1 ÷p xk) , (xk2 ÷p xk) , . . . , (xkm ÷p xk)] , (2.30)
but not distributive from the right.
Proof. This follows from the polyadic distributivity in the (m, n)-ring Z(m,n).
Example 2.23. 1) In the polyadic ring Z
[4,9]
(10,4) we have uniquely x28 ÷p x4 = x4 or 256 ÷p 4 = 4.
2) For the limiting ring Z
[3,4]
(5,3), we find x43 ÷p x1 = x−2 or 175 ÷p 7 = −5.
In the same way we define a polyadic analog of division with a remainder.
Definition 2.24. A polyadic division with a remainder is defined, if for a polyadic dividend xk1 and
divisor xk2 there exists a polyadic remainder xkr such that
xk1 = νm µn xk2 , xkq
n−1
, (xkr )m−1
, xk1 , xk2 , xkq , xkr ∈ Z(m,n), (2.31)

which is denoted by xkr = xk1 modp xk2 , and (2.31) can be presented in the following binary form
xk1 = xk2 ⊡p xkq ⊞p xkr . (2.32)
The distributivity of these operations is governed by distributivity in the polyadic ring Z(m,n).
Example 2.25. In the polyadic ring Z
[8,10]
(6,5) we can have different divisions for the same dividend as
38 = ((−22) ⊡p (−2)) ⊞p 78 = ((−92) ⊡p (−2)) ⊞p 238.
Secondly, because divisibility in the polyadic case is not symmetric with respect to dividend and
divisor (2.29), we define polyadically coprime numbers using the definition of compositeness (2.11).
Definition 2.26. The s polyadic integer numbers xk1 , . . . , xks ∈ Z(m,n) are polyadically coprime, if
their composition sets do not intersect D (xk1 ) ∩ D (xk1 ) ∩ . . . ∩ D (xks ) = ∅.
It is important that this definition does not imply the existence of a unit in Z(m,n), as opposed to the
definition of the polyadic prime numbers (2.14) in which the availability of a unit is crucial.
Assertion 2.27. Polyadically coprime numbers can exist in any polyadic ring Z(m,n), and not only in
the limiting cases with unit (see Proposition 2.13).
Corollary 2.28. All elements in the polyadic irreducible gap are polyadically coprime.
Example 2.29. In Z
[8,10]
(6,5) (2.13) the polyadic integer numbers (−32) = µ5 (−2)5
and 32768 =
µ5 (8)5
are both composed, but polyadically coprime, because the composition sets D (−32) =
{−2} and D (32768) = {8} do not intersect. Alternatively, not polyadically coprime numbers
here are, e.g., (−3072) = µ5 (−12) , (−2)2
, (8)2
and (−64512) = µ5 −2, (8)2
, 18, 28 , because
D (−3072) ∩ D (−64512) = {−2, 8} = ∅. We cannot multiply these two numbers, because the arity
of multiplication is 5.
Remark 2.30. For polyadic integer numbers Z(m,n) we cannot measure the property “to be coprime”
in terms of a single element, as in binary case, by their gcd, because the n-ary multiplication is only
allowed for admissible sequences. Therefore, we need to consider the intersection of the composition
sets.
In the polyadic ring Z
[a,b]
(m,n) for a given kmax ∈ Z+ we denote by S
[a,b]
(m,n) (kmax) the set of polyadic
prime numbers xk ∈ Z
[a,b]
(m,n) which are polyadically coprime to xkmax and x−kmax in the open interval
x−kmax < xk < xkmax (we assume that, if two numbers are coprime, then their opposite numbers are
also coprime). For the binary limit, obviously, S
[0,1]
(2,2) (kmax) = S (kmax) ∪ {−S (kmax)}.
Definition 2.31. The cardinality of S
[a,b]
(m,n) (kmax) is called a polyadic Euler function denoted by
ϕ
[a,b]
(m,n) (kmax) = S
[a,b]
(m,n) (kmax) . (2.33)
In the binary case ϕ
[0,1]
(2,2) (kmax) = 2ϕ (kmax). Because of Remark 2.30, the computation of the
polyadic Euler function requires for each element in the range x−kmax < xk < xkmax a thorough
consideration of composition sets.
Example 2.32. 1) For the limiting polyadic ring Z
[1,29]
(30,2) and kmax = 10 we have x−10 = −289,
x10 = 291 and
S
[1,29]
(30,2) (10) = {−260, −202, −173, −115, −86, −28, 1, 59, 88, 146, 175, 233, 262}, (2.34)
ϕ
[1,29]
(30,2) (10) = 13. (2.35)

10 STEVEN DUPLIJ
2) In another limiting case with ternary multiplication Z
[31,32]
(33,3) we get x−5 = −129, x5 = 191 and
S
[31,32]
(33,3) (5) = {−97, −65, −1, 31, 95, 127}, (2.36)
ϕ
[1,29]
(30,2) (5) = 6. (2.37)
3) In the non-limiting case Z
[7,10]
(11,5) and kmax = 10 we have x−10 = −93, x10 = 107 with
S
[7,10]
(11,5) (10) = {−83, −73, −53, −43, −23, −13, 7, 17, 37, 47, 67, 77, 97}, (2.38)
ϕ
[7,10]
(11,5) (10) = 13. (2.39)
4) For the polyadic Euler function in some other non-limiting cases we have
ϕ
[27,49]
(50,15) (7) = 6, ϕ
[17,38]
(39,10) (20) = 21, ϕ
[16,28]
(8,4) (30) = ϕ
[46,50]
(26,6) (15) = 0. (2.40)
2.3. The parameters-to-arity mapping. Let us consider the connection between congruence classes
and arities in more detail.
Remark 2.33. a) Solutions to (2.7) and (2.8) do not exist simultaneously for all a and b; b) The pair a,
b determines m, n uniquely; c) It can occur that for several different pairs a, b there can be the same
arities m, n.
Therefore, we have
Assertion 2.34. The parameters-to-arity mapping
ψ
[a,b]
(m,n) : (a, b) → (m, n) (2.41)
is a partial surjection.
Here we list the lowest arities which can be obtained with different choices of (a, b).
m = 3
n = 2
:
a = 1
b = 2
,
a = 3
b = 6
,
a = 5
b = 10
,
a = 7
b = 14
,
a = 9
b = 18
,
a = 11
b = 22
,
a = 13
b = 26
,
a = 15
b = 30
;
m = 4
n = 2
:
a = 1
b = 3
,
a = 4
b = 6
,
a = 4
b = 12
,
a = 10
b = 15
,
a = 7
b = 21
,
a = 16
b = 24
,
a = 10
b = 30
,
a = 22
b = 33
;
m = 4
n = 3
:
a = 2
b = 3
,
a = 2
b = 6
,
a = 8
b = 12
,
a = 14
b = 21
,
a = 8
b = 24
,
a = 20
b = 30
,
a = 11
b = 33
,
a = 14
b = 42
;
m = 5
n = 2
:
a = 1
b = 4
,
a = 9
b = 12
,
a = 5
b = 20
,
a = 21
b = 28
,
a = 9
b = 36
,
a = 33
b = 44
,
a = 13
b = 52
,
a = 45
b = 60
;
m = 5
n = 3
:
a = 3
b = 4
,
a = 3
b = 12
,
a = 15
b = 20
,
a = 7
b = 28
,
a = 27
b = 36
,
a = 39
b = 52
,
a = 15
b = 60
,
a = 51
b = 68
.
Also, not all pairs (a, b) are allowed due to (2.7)–(2.8). We list the forbidden (a, b) for b ≤ 20
b = 4 : a = 2;
b = 8 : a = 2, 4, 6;
b = 9 : a = 3, 6;
b = 12 : a = 2, 6, 10;
b = 16 : a = 2, 4, 6, 8, 12, 14;
b = 18 : a = 3, 6, 12, 15;
b = 20 : a = 2, 6, 10, 14, 18. (2.42)

The characterization of the fixed congruence class [[a]]b and the corresponding (m, n)-ring of
polyadic integer numbers Z
[a,b]
(m,n) can be done in terms of the shape invariants I, J ∈ Z+ defined
uniquely by (TABLE 3 in DUPLIJ [2017])
I = I[a,b]
m = (m − 1)
a
b
, J = J[a,b]
n =
an
− a
b
. (2.43)
Obviously, in the binary case, when m = n = 2 (a = 0, b = 1) both shape invariants vanish,
I = J = 0. Nevertheless, there exist “partially” binary cases, when only n = 2 and m = 2, while J is
nonzero, for instance in Z
[6,10]
(6,2) we have I = J = 3. In Example 2.3 for Z
[3,4]
(5,3) we have I = 3, J = 6.
In the limiting cases (2.15)–(2.16) we have, in general,
I
[1,b]
b+1 = 1, J
[1,b]
2 = 0, (2.44)
I
[b−1,b]
b+1 = b − 1, J
[b−1,b]
3 = (b − 1) (b − 2) . (2.45)
Thus, one can classify and distinguish the limiting cases of the congruence classes in terms of the
invariants and their manifest form (2.44)–(2.45).
3. FINITE POLYADIC RINGS
Now we present a special method of constructing a finite nonderived polyadic ring by combining
the “external” and “internal” methods. Let us “apply” 1) to 2), such that instead of (2.4), we introduce
the finite polyadic ring Z(m,n) cZ, where Z(m,n) is defined in (2.9). However, if we directly consider
the “double” class {a + bk + cl} and fix a and b, then the factorization by cZ will not give closed
operations for arbitrary c.
Assertion 3.1. If the finite polyadic ring Z
[a,b]
(m,n) cZ has q elements, then
c = bq. (3.1)
Proof. It follows from (3.1), that the “double” class remains in [[a]]b.
Remark 3.2. The representatives x
[a,b]
k = a + bk belong to a (m, n)-ring with the polyadic operations
(2.5)–(2.6), while the notions of subtraction, division, modulo and remainder are defined for binary
operations. Therefore, we cannot apply the standard binary modular arithmetic to x
[a,b]
k directly, but
we can define the equivalence relations and corresponding class operations in terms of k.
3.1. Secondary congruence classes. On the set of the “double” classes {a + bk + bql}, k, l ∈ Z and
fixed b ∈ N and a = 0, . . . , b − 1 we define the equivalence relation
k
∼ by
{a + bk1 + bql1}
k
∼ {a + bk2 + bql2} =⇒ k1 − k2 = ql, l, l1, l2 ∈ Z. (3.2)
Proposition 3.3. The equivalence relation
k
∼ is a congruence.
Proof. It follows from the binary congruency of k’s such that k1 ≡ k2 (mod q) which is just a rewrit-
ten form of the last condition of (3.2).
So now we can factorize Z
[a,b]
(m,n) by the congruence
k
∼ and obtain
Definition 3.4. A secondary (equivalence) class of a polyadic integer x
[a,b]
k = a + bk ∈ Z
[a,b]
(m,n) “mod-
ulo” bq (with q being the number of representatives x
[a,b]
k , for fixed b ∈ N and 0 ≤ a ≤ b − 1)
is
x
[a,b]
k
bq
= {{(a + bk) + bql} | l ∈ Z, q ∈ N, 0 ≤ k ≤ q − 1} . (3.3)

12 STEVEN DUPLIJ
Remark 3.5. If the binary limit is given by a = 0, b = 1 and Z
[0,1]
(2,2) = Z, then the secondary class
becomes the ordinary class (2.1).
If the values of the parameters a, b, q are clear from the context, we denote the secondary class
representatives by an integer with two primes, as follows x
[a,b]
k
bq
≡ x′′
k ≡ x′′
.
Example 3.6. a) Let a = 3, b = 5, then for q = 4 elements we have the secondary classes with
k = 0, 1, 2, 3 (the corresponding binary limits are in brackets)
x
[3,5]
k
20
= 3′′
, 8′′
, 13′′
, 18′′
=



3′′
= {. . . − 17, 3, 23, 43, 63, . . .} ,
8′′
= {. . . − 12, 8, 28, 48, 68, . . .} ,
13′′
= {. . . − 7, 13, 33, 53, 73, . . .} ,
18′′
= {. . . − 2, 18, 38, 58, 78, . . .} ,
(3.4)



[[k]]4 = 0′
, 1′
, 2′
, 3′
=



0′
= {. . . − 4, 0, 4, 8, 12, . . .} ,
1′
= {. . . − 3, 1, 5, 9, 13, . . .} ,
2′
= {. . . − 2, 2, 6, 10, 14, . . .} ,
3′
= {. . . − 1, 3, 7, 11, 15, . . .} .



 (3.5)
b) For a = 3, b = 6 and for 4 elements and k = 0, 1, 2, 3
x
[3,6]
k
24
= 3′′
, 9′′
, 15′′
, 21′′
, ([[k]]4 = 0′
, 1′
, 2′
, 3′
) . (3.6)
c) If a = 4, b = 5, for 3 elements and k = 0, 1, 2 we get
x
[4,5]
k
15
= 4′′
, 9′′
, 14′′
, ([[k]]3 = 0′
, 1′
, 2′
) . (3.7)
The crucial difference between these sets of classes are: 1) they are described by rings of different
arities determined by (2.7) and (2.8); 2) some of them are fields.
3.2. Finite polyadic rings of secondary classes. Now we determine the operations between sec-
ondary classes. The most significant difference with the binary class operations (2.2)–(2.3) is the fact
that secondary classes obey nonderived polyadic operations.
Proposition 3.7. The set {x′′
k} of q secondary classes k = 0, . . . , q − 1 (with the fixed a, b) can be
endowed with the following commutative m-ary addition
x′′
kadd
= ν′′
m x′′
k1
, x′′
k1
, . . . , x′′
km
, (3.8)
kadd ≡ (k1 + k2 + . . . + km) + I[a,b]
m (mod q) (3.9)
and commutative n-ary multiplication
x′′
kmult
= µ′′
n x′′
k1
, x′′
k1
, . . . , x′′
kn
, (3.10)
kmult ≡ an−1
(k1 + k2 + . . . + kn) + an−2
b (k1k2 + k2k3 + . . . + kn−1kn) + . . .
+bn−1
k1 . . . kn + J[a,b]
n (mod q) , (3.11)
which satisfy the polyadic distributivity, and the shape invariants I
[a,b]
m , J
[a,b]
n are defined in (2.43).
Proof. This follows from the definition of the secondary class (3.3) and manifest form of the “under-
lying” polyadic operations (2.5)–(2.6), which are commutative and distributive.
Remark 3.8. The binary limit is given by a = 0, b = 1 and m = n = 2, I
[a,b]
m = J
[a,b]
n = 0, such that the
secondary class becomes the ordinary congruence class x′′
k → k′
, obeying the standard binary class
operations (2.2)–(2.3), which in terms of k are kadd ≡ (k1 + k2) (mod q), kmult ≡ (k1k2) (mod q).

Definition 3.9. The set of secondary classes (3.3) equipped with operations (3.8), (3.10) is denoted
by
Z(m,n) (q) ≡ Z
[a,b]
(m,n) (q) = Z
[a,b]
(m,n) (bq) Z = {{x′′
k} | ν′′
m, µ′′
n} , (3.12)
and is a finite secondary class (m, n)-ring of polyadic integer numbers Z(m,n) ≡ Z
[a,b]
(m,n). The value q
(the number of elements) is called its order.
Informally, Z(m,n) = Z(m,n) (∞). First, note that the constructed finite (m, n)-rings (3.12) have a
much richer structure and exotic properties which do not exist in the binary finite residue class rings
(2.4), and, in general, they give many concrete examples for possible different kinds polyadic rings.
One of such “non-binary” properties is the availability of several units (for odd multiplicative arity n),
and moreover sometimes all ring elements are units (such rings are “automatically” fields, see below).
Example 3.10. a) In (5, 3)-ring Z
[3,4]
(4,3) (2) with 2 secondary classes both elements are units (we mark
units by subscript e) e1 = 3′′
e = 3′′
, e2 = 7′′
e = 7′′
, because they are both multiplicative idempotents
and satisfy the following ternary multiplication (cf. (1.4))
µ3 [3′′
, 3′′
, 3′′
] = 3′′
, µ3 [3′′
, 3′′
, 7′′
] = 7′′
, µ3 [3′′
, 7′′
, 7′′
] = 3′′
, µ3 [7′′
, 7′′
, 7′′
] = 7′′
. (3.13)
b) In the same way the ring Z
[5,6]
(7,3) (4) consists of only 4 units e1 = 5′′
e, e2 = 11′′
e, e3 = 17′′
e, e4 = 23′′
e,
and no zero.
c) Equal arity rings of the same order may be not isomorphic. For instance, Z
[1,3]
(4,2) (2) consists of
unit e = 1′′
e = 1′′
and zero z = 4′′
z = 4′′
only, satisfying
µ2 [1′′
, 1′′
] = 1′′
, µ2 [1′′
, 4′′
] = 4′′
, µ2 [4′′
, 4′′
] = 4′′
, (3.14)
and therefore Z
[1,3]
(4,2) (2) is a field, because {1′′
, 4′′
z} 4′′
z is a (trivial) binary group, consisting of one
element 1′′
e. However, Z
[4,6]
(4,2) (2) has the zero z = 4′′
z = 4′′
, 10′′
and has no unit, because
µ2 [4′′
, 4′′
] = 4′′
, µ2 [4′′
, 10′′
] = 4′′
, µ2 [10′′
, 10′′
] = 4′′
, (3.15)
so that Z
[4,6]
(4,2) (2) is not a field, because of the last relation (nilpotency of 10′′
). Their additive 4-ary
groups are also not isomorphic (which is easy to show). However, Z
[1,3]
(4,2) (2) and Z
[4,6]
(4,2) (2) have the
same arity and order.
Recalling Assertion 2.34, we conclude more concretely:
Assertion 3.11. For a fixed arity shape (m, n), there can be non-isomorphic secondary class polyadic
rings Z(m,n) (q) of the same order q, which describe different binary residue classes [[a]]b.
A polyadic analog of the characteristic can be introduced, when there exist both a unit and zero in
a finite ring. Recall, that if R is a finite binary ring with unit1 and zero 0, then its characteristic is
defined as a smallest integer χ, such that
χ
1 + 1+, . . . , +1 = χ · 1 = 0. (3.16)
This means that the “number” of unit additions being χ − 1 produces zero. The same is evident for
any other element x ∈ R, because x = x · 1.
Definition 3.12. For the finite polyadic ring Z(m,n) (q) which contains both the unit e and the zero z,
a polyadic characteristic χp is defined as a smallest additive polyadic power (1.2) of e which is equal to
zero
e χp +m = z. (3.17)

14 STEVEN DUPLIJ
In the binary limit, obviously, χp = χ − 1. A polyadic analog of the middle term in (3.16) can be
obtained by using the polyadic distributivity and (2.5) as
e χp +m = e χp(m−1)+1 ×n . (3.18)
In TABLE 1 we present the parameters-to-arity mapping ψ
[a,b]
(m,n) (2.41) together with the polyadic
characteristics of those finite secondary class rings Z
[a,b]
(m,n) (q) which contain both unit(s) and zero, and
which have order less or equal than 10 for b ≤ 6.
Now we turn to the question of which secondary classes can be described by polyadic finite fields.
4. FINITE POLYADIC FIELDS
Let us consider the structure of the finite secondary class rings Z
[a,b]
(m,n) (q) in more detail and deter-
mine which of them are polyadic fields.
Proposition 4.1. A finite polyadic ring Z
[a,b]
(m,n) (q) is a secondary class finite (m, n)-field F
′′[a,b]
(m,n) (q)
if all its elements except z (if it exists) are polyadically multiplicative invertible having a unique
querelement.
Proof. In both cases {{x′′
k} | µ′′
n} and {{x′′
k z} | µ′′
n} are commutative and cancellative n-ary groups,
which follows from the concrete form of multiplication (3.10). Therefore, according to Definition 1.1,
in such a case Z
[a,b]
(m,n) (q) becomes a polyadic field F
′′[a,b]
(m,n) (q).
4.1. Abstract finite polyadic fields. In the binary case LIDL AND NIEDERREITER [1997] the
residue (congruence) class ring (2.4) with q elements Z qZ is a congruence class (non-extended)
field, if its order q = p is a prime number, such that F′
(p) = [[a]]p | +′
, ·′
, 0′
, 1′
, a =
0, 1, . . . , p − 1. Because all non-extended binary fields of a fixed prime order p are isomorphic each
other and, in tern, isomorphic to the congruence class field F′
(p), it is natural to study them in a more
“abstract” way, i.e. without connection to a specific congruence class structure. This can be achieved
by consideration of the one-to-one onto mapping from the congruence class to its representative which
preserves the field (ring) structure and provides operations (binary multiplication and addition with
ordinary 0 and 1) by modulo p. In other words, the mapping Φp [[a]]p = a is an isomorphism of
binary fields Φp : F′
(p) → F (p), where F (p) = {{a} | +, ·, 0, 1}mod p is an “abstract” non-extended
(prime) finite field of order p (or Galois field GF (p)).
In a similar way, we introduce a polyadic analog of the “abstract” binary non-extended (prime)
finite fields. Let us consider the set of polyadic integer numbers {xk} ≡ x
[a,b]
k = {a + bk} ∈
Z
[a,b]
(m,n), b ∈ N and 0 ≤ a ≤ b − 1, 0 ≤ k ≤ q − 1, q ∈ N, which obey the operations (2.5)–(2.6). The
polyadic version of the prime finite field F (p) of order p (or Galois field GF (p)) is given by
Definition 4.2. The “abstract” non-extended (prime) finite (m, n)-field of order q is
F(m,n) (q) ≡ F
[a,b]
(m,n) (q) = {{a + bk} | νm, µn}mod bq , (4.1)
if {{xk} | νm}mod bq is an additive m-ary group, and {{xk} | µn}mod bq (or, when zero z exists,
{{xk z} | µn}mod bq) is a multiplicative n-ary group.
Then we define a one-to-one onto mapping from the secondary congruence class to its representa-
tive by Φ
[a,b]
q x
[a,b]
k
bq
= x
[a,b]
k and arrive at the following

TABLE 1. Polyadic characteristics χp for the ﬁnite secondary class (m, n)-rings
Z
[a,b]
m,n (q) of order 2 ≤ q ≤ 10 for 2 ≤ b ≤ 6. The orders q which do not give
ﬁelds are slanted.
a b 2 3 4 5 6
1
m = 3
n = 2
q=3, χp = 1
q=5, χp = 2
q=7, χp = 3
q=9, χp = 4
m = 4
n = 2
q=2, χp = 1
q=4, χp = 1
q=5, χp = 3
q=7, χp = 2
q=8, χp = 5
q=10, χp = 3
m = 5
n = 2
q=3, χp = 2
q=5, χp = 1
q=7, χp = 5
q=9, χp = 2
m = 6
n = 2
q=2, χp = 1
q=3, χp = 1
q=4, χp = 3
q=6, χp = 1
q=7, χp = 4
q=8, χp = 3
q=9, χp = 7
m = 7
n = 2
q=5, χp = 4
q=7, χp = 1
2
m = 4
n = 3
q=2, χp = 1
q=4, χp = 1
q=5, χp = 3
q=7, χp = 2
q=8, χp = 5
q=10, χp = 3
m = 6
n = 5
q=2, χp = 1
q=3, χp = 1
q=4, χp = 3
q=6, χp = 1
q=7, χp = 4
q=8, χp = 3
q=9, χp = 7
m = 4
n = 3
q=5, χp = 3
q=7, χp = 2
q=10, χp = 3
3
m = 5
n = 3
q=3, χp = 2
q=5, χp = 1
q=7, χp = 5
q=9, χp = 2
m = 6
n = 5
q=2, χp = 1
q=3, χp = 1
q=4, χp = 3
q=6, χp = 1
q=7, χp = 4
q=8, χp = 3
m = 3
n = 2
q=5, χp = 2
q=7, χp = 3
4
m = 6
n = 3
q=2, χp = 1
q=3, χp = 1
q=4, χp = 3
q=6, χp = 1
q=7, χp = 4
q=8, χp = 3
q=9, χp = 7
m = 4
n = 2
q=5, χp = 3
q=7, χp = 2
q=10, χp = 3
5
m = 7
n = 3
q=5, χp = 4
q=7, χp = 1

16 STEVEN DUPLIJ
Proposition 4.3. The mapping Φ
[a,b]
q : F
′′[a,b]
(m,n) (q) → F
[a,b]
(m,n) (q) is a polyadic ring homomorphism
(being, in fact, an isomorphism) and satisfies (here we use the “prime” notations)
Φ[a,b]
q (ν′′
m [x′′
1, x′′
2, . . . , x′′
m]) = νm Φ[a,b]
q (x1) , Φ[a,b]
q (x2) , . . . , Φ[a,b]
q (xm) , (4.2)
Φ[a,b]
q (µ′′
n [x′′
1, x′′
2, . . . , x′′
n]) = µn Φ[a,b]
q (x1) , Φ[a,b]
q (x2) , . . . , Φ[a,b]
q (xn) . (4.3)
Proof. This follows directly from (2.5)–(2.6) and (3.8)–(3.10). Obviously, a mapping defined in this
way governs the polyadic distributivity, and therefore Φ
[a,b]
q is a ring homomorphism, or, more exactly,
a 1-place heteromorphism for m-ary addition together with n-ary multiplication (see DUPLIJ [2017]).
Because x′′
k → xk is one-to-one for any fixed 0 ≤ k ≤ q − 1, Φ
[a,b]
q is an isomorphism.
In TABLE 2 we present the “abstract” non-extended polyadic finite fields F
[a,b]
(m,n) (q) of lowest arity
shape (m, n) and orders q. The forbidden pairs (a, b) for F
[a,b]
(m,n) (q) coincide with ones for polyadic
rings Z
[a,b]
(m,n) listed in (2.42).
4.2. Multiplicative structure. In the multiplicative structure the following crucial differences be-
tween the binary finite fields and F(m,n) (q) can be outlined.
Remark 4.4. The order of a non-extended finite polyadic field may not be prime (e.g., F
[1,2]
(3,2) (4),
F
[3,4]
(5,3) (8), F
[2,6]
(4,3) (9)), and may not even be a power of a prime binary number (e.g. F
[5,6]
(7,3) (6),
F
[3,10]
(11,5) (10)), see TABLE 3.
Remark 4.5. The polyadic characteristic χp of a non-extended finite polyadic field can have values
such that χp + 1 (corresponding in the binary case to the ordinary characteristic χ) can be nonprime
(TABLE 1).
Assertion 4.6. If a secondary class x
[a,b]
k
bq
contains no zero, it can be isomorphic to the “ab-
stract” zeroless finite polyadic field F
[a,b]
(m,n) (q).
Zeroless fields are marked by one frame in TABLE 2. There exist finite polyadic fields with more
than one unit, and also all elements can be units. Such cases are marked in TABLE 3 by subscripts
which indicate the number of units.
Denote the Abelian finite multiplicative n-ary group of F
[a,b]
(m,n) (q) by G
[a,b]
n (q).
Example 4.7. 1) The finite (6, 3)-field F
[4,5]
(6,3) (3) of order 3 has two units {4, 14} ≡ {4e, 14e} and zero
9 ≡ 9z, and its multiplicative 3-ary group G
[4,5]
3 (2) is
µ3 [4, 4, 4] = 4, µ3 [4, 4, 14] = 14, µ3 [4, 14, 14] = 4, µ3 [14, 14, 14] = 14.
2) In F
[5,6]
(7,3) (4) of order 4 all the elements {5, 11, 17, 23} ≡ {5e, 11e, 17e, 23e} are units (see (1.4)),
because for its 3-ary group G
[5,6]
3 (4) we have
µ3 [5, 5, 5] = 5, µ3 [11, 11, 11] = 11, µ3 [17, 17, 17] = 17, µ3 [23, 23, 23] = 23,
µ3 [5, 5, 11] = 11, µ3 [5, 5, 17] = 17, µ3 [5, 5, 23] = 23,
µ3 [11, 11, 5] = 5, µ3 [11, 11, 17] = 17, µ3 [11, 11, 23] = 23,
µ3 [17, 17, 5] = 5, µ3 [17, 17, 11] = 11, µ3 [17, 17, 23] = 23,
µ3 [23, 23, 5] = 5, µ3 [23, 23, 11] = 11, µ3 [23, 23, 17] = 17.

TABLE 2. Content and arities of the secondary class finite polyadic rings Z
[a,b]
(m,n) (q)
and the corresponding simple finite polyadic (m, n)-fields F
[a,b]
m,n (q) of order 2 ≤ q ≤ 4
(framed) for 2 ≤ b ≤ 6. The subscripts e and z mark those secondary classes (two
primes are omitted) which play the roles of polyadic unit and polyadic zero respec-
tively. The double frames denote the finite polyadic fields containing both a unit and
zero. The last line in a cell (corresponding to a fixed congruence class [[a]]b) gives the
allowed orders of finite polyadic fields for 5 ≤ q ≤ 10, and bold numbers mark the
orders of such fields which contain both unit(s) and zero.
a b 2 3 4 5 6
1
m = 3
n = 2
1e,3
1e,3z,5
1e,3,5,7
q=5,7,8
m = 4
n = 2
1e,4z
1e,4,7
1e,4z,7,10
q=5,7,9
m = 5
n = 2
1e,5
1e,5,9z
1e,5,9,13
q=5,7,8
m = 6
n = 2
1e,6z
1e,6z,11
1e,6,11,16z
q=5,7
m = 7
n = 2
1e,7
1e,7,13
1e,7,13,19
q=5,6,7,8,9
2
m = 4
n = 3
2z,5e
2,5,8e
2,5e,8z,11e
q=5,7,9
m = 6
n = 5
2z,7e
2e,7,12z
2,7e,12z,17e
q=5,7
m = 4
n = 3
2,8z
2,8e,14
2,8z,14,20
q=5,7,9
3
m = 5
n = 3
3e,7e
3z,7e,11e
3,7e,11,15e
q=5,6,7,8
m = 6
n = 5
3e,8z
3z,8e,13e
3e,8z,13e,18
q=5,7
m = 3
n = 2
3,9e
3,9z,15
3,9e,15,21
q=5,7,8
4
m = 6
n = 3
4z,9e
4e,9z,14e
4z,9e,14,19e
q=5,7
m = 4
n = 2
4z,10
4,10e,16
4,10,16z,22
q=5,7,9
5
m = 7
n = 3
5e,11e
5,11,17e
5e,11e,17e,23e
q=5,6,7,8,9

18 STEVEN DUPLIJ
In general, n-ary groups may contain no units (and multiplicative idempotents) at all, and invert-
ibility is controlled in another way, by querelements: each element of any n-ary group should be
(uniquely) “quereable” (1.5). In case of (m, n)-fields both m-ary additive group and n-ary multiplica-
tive group G
[a,b]
n (q) can be of this kind. By analogy with zeroless-nonunital rings we have
Definition 4.8. A polyadic field F(m,n) is called zeroless-nonunital, if it contains no zero and no unit.
Assertion 4.9. The zeroless-nonunital polyadic fields are totally (additively and multiplicatively) non-
derived.
Proposition 4.10. 1) If gcd (bq) = 1, then F
[a,b]
(m,n) (q) is zeroless. 2) The zero can exist only, if
gcd (bq) = 1 and the field order q = p is prime.
Proof. It follows directly from the definition of the polyadic zero (1.3) and (3.9).
Let us consider examples of zeroless-nonunital finite fields of polyadic integer numbers F
[a,b]
(m,n) (q).
Example 4.11. 1) The zeroless-nonunital polyadic finite fields having lowest |a + b| are, e.g.,
F
[3,8]
(9,3) (2), F
[3,8]
(9,3) (4), F
[5,8]
(9,3) (4), F
[5,8]
(9,3) (8), also F
[4,9]
(10,4) (3), F
[4,9]
(10,4) (9), and F
[7,9]
(10,4) (3), F
[7,9]
(10,4) (9).
2) The multiplication of the zeroless-nonunital (9, 3)-field F
[5,8]
(9,3) (2) is
µ3 [5, 5, 5] = 13, µ3 [5, 5, 13] = 5, µ3 [5, 13, 13] = 13, µ3 [13, 13, 13] = 5.
Using (1.2) we find the (unique) multiplicative querelements ¯5 = 13, 13 = 5. The addition of
F
[5,8]
(9,3) (2) is
ν9 59
= 13, ν9 58
, 13 = 5, ν9 57
, 132
= 13, ν9 56
, 133
= 5, ν9 55
, 134
= 13,
ν9 54
, 135
= 5, ν9 53
, 136
= 13, ν9 52
, 137
= 5, ν9 5, 138
= 13, ν9 139
= 5.
The additive (unique) querelements are ˜5 = 13, 13 = 5. So all elements are additively and multi-
plicatively querable (polyadically invertible), and therefore ν9 is 9-ary additive group operation and
µ3 is 3-ary multiplicative group operation, as it should be for a field. Because it contains no unit and
no zero, F
[5,8]
(9,3) (2) is actually a zeroless-nonunital finite (9, 3)-field of order 2.
Other zeroless-nonunital finite polyadic fields are marked by frames in TABLE 3.
Remark 4.12. The absence of zero does not guarantee that a (m, n)-ring R
[a,b]
(m,n) (q) is a field. For that,
both [[a]]b | νm and [[a]]b | µn have to be polyadic groups.
Example 4.13. The (4, 3)-ring R
[2,3]
(4,3) (6) is zeroless, and [[3]]4 | ν4 is its 4-ary additive group (each
element has a unique additive querelement). Despite each element of [[2]]3 | µ3 having a querele-
ment, it is not a multiplicative 3-ary group, because for the two elements 2 and 14 we have nonunique
querelements
µ3 [2, 2, 5] = 2, µ3 [2, 2, 14] = 2, µ3 [14, 14, 2] = 14, µ3 [14, 14, 11] = 14. (4.4)
The conditions on the congruence classes [[a]]b and the invariants I, J (2.43), which give the same
arity structure are given in DUPLIJ [2017]. Note, that there exist polyadic fields of the same arities
(m, n) and the same order q which are not isomorphic (in contrast with what is possible in the binary
case).
Example 4.14. The polyadic (9, 3)-fields corresponding to the congruence classes [[5]]8 and [[7]]8 are
not isomorphic for orders q = 2, 4, 8 (see TABLE 3). Despite both being zeroless, the first F
[5,8]
(9,3) (q)
are nonunital, while the second F
[7,8]
(9,3) (q) has two units, which makes an isomorphism impossible.

Recall LIDL AND NIEDERREITER [1997], that in a (non-extended, prime) finite binary field F (p),
the order of an element x ∈ F (p) is defined as a smallest integer λ such that xλ
= 1. Obviously, the
set of fixed order elements forms a cyclic subgroup Gλ of the multiplicative binary group of F (p),
and λ | (p − 1). If λ = p − 1, such an element is called a primitive (root), it generates all elements,
and these exist in any finite binary field. Moreover, any element of F (p) is idempotent xp
= x, while
all its nonzero elements satisfy xp−1
= 1 (Fermat’s little theorem). A non-extended (prime) finite
field is fully determined by its order p (up to isomorphism), and, moreover, any F (p) is isomorphic
to Z pZ.
In the polyadic case, the situation is more complicated. Because the related secondary class struc-
ture (4.1) contains parameters in addition to the number of elements q, the order of (non-extended)
polyadic fields may not be prime, or nor even a power of a prime integer (e.g. F
[5,6]
(7,3) (6) or F
[3,10]
(11,5) (10)).
Also, as was shown above, finite polyadic fields can be zeroless, nonunital and have many (or even
all) units (see TABLE 3). Therefore, we cannot use units in the definition of the element order. Instead,
we propose an alternative:
Definition 4.15. If an element of the finite polyadic field x ∈ F(m,n) (q) satisfies
x λp ×n = x, (4.5)
then the smallest such λp is called the idempotence polyadic order and denoted ord x = λp.
Obviously, λp = λ (see (1.1)).
Definition 4.16. The idempotence polyadic order λ
[a,b]
p of a finite polyadic field F
[a,b]
(m,n) (q) is the maxi-
mum λp of all its elements, and we call such field λ
[a,b]
p -idempotent and denote ord F
[a,b]
(m,n) (q) = λ
[a,b]
p .
In TABLE 3 we present the idempotence polyadic order λ
[a,b]
p for the (allowed) finite polyadic fields
F
[a,b]
(m,n) (q) (4.1) with 2 ≤ b ≤ 10 and order q ≤ 10.
Definition 4.17. Denote by q∗ the number of nonzero distinct elements in F(m,n) (q)
q∗ =
q − 1, if ∃z ∈ F(m,n) (q)
q, if ∄z ∈ F(m,n) (q) ,
(4.6)
which is called a reduced (field) order.
The second choice of (4.6) in the binary case is absent, because any commutative binary group (as
the additive group of a field) contains a zero (the identity of this group), and therefore any binary field
has a zero, which does not always hold for the m-ary additive group of F(m,n) (see Example 4.11).
Theorem 4.18. If a finite polyadic field F(m,n) (q) has an order q, such that q∗ = qadm
∗ = ℓ (n − 1)+1
is n-admissible, then (for n ≥ 3 and one unit):
1) A sequence of the length q∗ (n − 1) built from any fixed element y ∈ F(m,n) (q) is neutral
µ(q∗)
n x, yq∗(n−1)
= x, ∀x ∈ F(m,n) (q) . (4.7)
2) Any element y satisfies the polyadic idempotency condition
y q∗ ×n = y, ∀y ∈ F(m,n) (q) . (4.8)
Proof. 1) Take a long n-ary product of the q∗ distinct nonzero elements x0 = µ
(ℓ)
n x1, x2, . . . , xq∗
,
such that q∗ can take only multiplicatively n-admissible values qadm
∗ , where ℓ ∈ N is a “number” of
n-ary multiplications. Then polyadically multiply each xi by a fixed element y ∈ F(m,n) (q) such that

20 STEVEN DUPLIJ
TABLE 3. Idempotence polyadic orders λ
[a,b]
p for (allowed) finite polyadic fields
F
[a,b]
(m,n) (q), with 2 ≤ b ≤ 10 and field orders 2 ≤ q ≤ 10. If q∗ is n-admissible,
they are underlined. The zeroless-nonunital cases are framed. Subscripts correspond
to the number of units κe ≥ 2, while nonframed entries have exactly one unit.
[[a]]b Arities Finite polyadic field order q
b a (m, n) 2 3 4 5 6 7 8 9 10
2 1 (3, 2) 2 2 2 4 ∅ 6 4 ∅ ∅
3 1 (4, 2) 1 3 ∅ 4 ∅ 6 ∅ 9 ∅
2 (4, 3) 1 3 ∅ 22e ∅ 32e ∅ 9 ∅
4 1 (5, 2) 2 2 4 4 ∅ 6 8 ∅ ∅
3 (5, 3) 12e 12e 22e 22e ∅ 32e 42e ∅ ∅
5 1 (6, 2) 1 2 ∅ 5 ∅ 6 ∅ ∅ ∅
2 (6, 5) 1 12e ∅ 5 ∅ 32e ∅ ∅ ∅
3 (6, 5) 1 12e ∅ 5 ∅ 32e ∅ ∅ ∅
4 (6, 3) 1 12e ∅ 5 ∅ 32e ∅ ∅ ∅
6 1 (7, 2) 2 3 2 4 6 6 4 9 ∅
2 (4, 3) ∅ 3 ∅ 22e ∅ 32e ∅ 9 ∅
3 (3, 2) 2 ∅ 2 4 ∅ 6 4 ∅ ∅
4 (4, 2) ∅ 3 ∅ 4 ∅ 6 ∅ 9 ∅
5 (7, 3) 12e 3 14e 22e 32e 32e 2 9 ∅
7 1 (8, 2) 1 2 ∅ 4 ∅ 7 ∅ ∅ ∅
2 (8, 4) 1 2 ∅ 4 ∅ 7 ∅ ∅ ∅
3 (8, 7) 1 12e ∅ 22e ∅ 7 ∅ ∅ ∅
4 (8, 4) 1 2 ∅ 4 ∅ 7 ∅ ∅ ∅
5 (8, 7) 1 12e ∅ 22e ∅ 7 ∅ ∅ ∅
6 (8, 3) 1 12e ∅ 22e ∅ 7 ∅ ∅ ∅
8 1 (9, 2) 2 2 4 4 ∅ 6 8 ∅ ∅
3 (9, 3) 2 12e 4 22e ∅ 32e 8 ∅ ∅
5 (9, 3) 2 12e 4 22e ∅ 32e 8 ∅ ∅
7 (9, 3) 12e 12e 22e 22e ∅ 32e 42e ∅ ∅
9 1 (10, 2) 1 3 ∅ 4 ∅ 6 ∅ 9 ∅
2 (10, 7) 1 3 ∅ 22e ∅ 16e ∅ 9 ∅
4 (10, 4) 1 3 ∅ 4 ∅ 23e ∅ 9 ∅
5 (10, 7) 1 3 ∅ 22e ∅ 16e ∅ 9 ∅
7 (10, 4) 1 3 ∅ 4 ∅ 23e ∅ 9 ∅
8 (10, 3) 1 3 ∅ 22e ∅ 32e ∅ 9 ∅
10 1 (11, 2) 2 2 2 5 ∅ 6 8 ∅ 10
2 (6, 5) ∅ 12e ∅ 5 ∅ 32e ∅ ∅ ∅
3 (11, 5) 12e 12e 14e 5 ∅ 32e 18e ∅ 52e
4 (6, 3) ∅ 12e ∅ 5 ∅ 32e ∅ ∅ ∅
5 (3, 2) 2 2 2 ∅ ∅ 6 ∅ ∅ ∅
6 (6, 2) ∅ 2 ∅ 5 ∅ 6 ∅ ∅ ∅
7 (11, 5) 12e 12e 14e 5 ∅ 32e 18e ∅ 52e
8 (6, 5) ∅ 12e ∅ 5 ∅ 32e ∅ ∅ ∅
9 (11, 3) 12e 12e 14e 5 ∅ 32e 24e ∅ 52e

all q∗ elements µn [xi, yn−1
] will be distinct as well. Therefore, their product should be the same x0.
Using commutativity and associativity, we obtain
x0 = µ(ℓ)
n x1, x2, . . . , xq∗
= µ(ℓ)
n µn x1, yn−1
, µn x2, yn−1
, . . . , µn xq∗ , yn−1
= µ(q∗)
n µ(ℓ)
n x1, x2, . . . , xq∗
, yq∗(n−1)
= µ(q∗)
n x0, yq∗(n−1)
. (4.9)
2) Insert into the formula obtained above x0 = y, then use (1.2) to get (4.8).
Finite polyadic fields F
[a,b]
(m,n) (q) having n-admissible reduced order q∗ = qadm
∗ = ℓ (n − 1) + 1
(ℓ ∈ N) (underlined in TABLE 3) are closest to the binary finite fields F (p) in their general properties:
they are half-derived, while if they contain a zero, they are fully derived. If q∗ = qadm
∗ , then F
[a,b]
(m,n) (q)
can be nonunital or contain more than one unit (subscripts in TABLE 3).
Assertion 4.19. The finite fields F
[a,b]
(m,n) (q) of n-admissible reduced order q∗ = qadm
∗ cannot have
more than one unit and cannot be zeroless-nonunital.
Assertion 4.20. If q∗ = qadm
∗ , and F
[a,b]
(m,n) (q) is unital zeroless, then the reduced order q∗ is the product
of the idempotence polyadic field order λ
[a,b]
p = ord F
[a,b]
(m,n) (q) and the number of units κe (if a ∤ b and
n ≥ 3)
q∗ = λ[a,b]
p κe. (4.10)
Let us consider the structure of the multiplicative group G
[a,b]
n (q∗) of F
[a,b]
(m,n) (q) in more detail.
Some properties of commutative cyclic n-ary groups were considered for particular relations between
orders and arity. Here we have: 1) more parameters and different relations between these, the arity
and order; 2) the (m, n)-field under consideration, which leads to additional restrictions. In such a
way exotic polyadic groups and fields arise which have unusual properties that have not been studied
before.
Definition 4.21. An element xprim ∈ G
[a,b]
n (q∗) is called n-ary primitive, if its idempotence order is
λp = ord xprim = q∗. (4.11)
Then, all λp polyadic powers x
1 ×n
prim , x
2 ×n
prim , . . . , x
q∗ ×n
prim ≡ xprim generate other distinct elements,
and so G
[a,b]
n (q∗) is a finite cyclic n-ary group generated by xprim, i.e. G
[a,b]
n (q∗) = x
i ×n
prim | µn .
We denote a number primitive elements in F
[a,b]
(m,n) (q) by κprim.
Assertion 4.22. For zeroless F
[a,b]
(m,n) (q) and prime order q = p, we have λ
[a,b]
p = q, and G
[a,b]
n (q) is
indecomposable (n ≥ 3).
Example 4.23. The smallest 3-admissible zeroless polyadic field is F
[2,3]
(4,3) (3) with the unit e = 8 and
two 3-ary primitive elements 2, 5 having 3-idempotence order ord 2 = ord 5 = 3, so κprim = 2 ,
because
2 1 ×3 = 8, 2 2 ×3 = 5, 2 3 ×3 = 2, 5 1 ×3 = 8, 5 2 ×3 = 2, 5 3 ×3 = 5, (4.12)
and therefore G
[2,3]
3 (3) is a cyclic indecomposable 3-ary group.
Assertion 4.24. If F
[a,b]
(m,n) (q) is zeroless-nonunital, then every element is n-ary primitive, κprim = q,
also λ
[a,b]
p = q (the order q can be not prime), and G
[a,b]
n (q) is a indecomposable commutative cyclic
n-ary group without identity (n ≥ 3).

22 STEVEN DUPLIJ
Example 4.25. The (10, 7)-field F
[5,9]
(10,7) (9) is zeroless-nonunital, each element (has λp = 9) is prim-
itive and generates the whole field, and therefore κprim = 9, thus the 7-ary multiplicative group
G
[5,9]
7 (9) is indecomposable and without identity.
The structure of G
[a,b]
n (q∗) can be extremely nontrivial and may have no analogs in the binary case.
Assertion 4.26. If there exists more than one unit, then:
1) If G
[a,b]
n (q∗) can be decomposed on its n-ary subroups, the number of units κe coincides with
the number of its cyclic n-ary subgroups G
[a,b]
n (q∗) = G1 ∪ G2 . . . ∪ Gke which do not intersect
Gi ∩ Gj = ∅, i, j = i = 1, . . . , κe, i = j.
2) If a zero exists, then each Gi has its own unit ei, i = 1, . . . , κe.
3) In the zeroless case G
[a,b]
n (q) = G1 ∪ G2 . . . ∪ Gke ∪ E (G), where E (G) = {ei} is the split-off
subgroup of units.
Example 4.27. 1) In the (9, 3)-field F
[5,8]
(9,3) (7) there is a single zero z = 21 and two units e1 =
13 ≡ 13e, e2 = 29 ≡ 29e, and so its multiplicative 3-ary group G
[5,8]
3 (6) = {5, 13, 29, 37, 45, 53}
consists of two nonintersecting (which is not possible in the binary case) 3-ary cyclic subgroups
G1 = {5, 13e, 45} and G2 = {29e, 37, 53} (for both λp = 3)
G1 = 5 1 ×3 = 13e, 5 2 ×3 = 45, 5 3 ×3 = 5 ,
G2 = 37 1 ×3 = 29e, 37 2 ×3 = 53, 37 3 ×3 = 37 .
Also, all nonunital elements in G
[5,8]
3 (6) are (polyadic) 1-reflections, because 5 1 ×3 = 45 1 ×3 = 13e
and 37 1 ×3 = 53 1 ×3 = 29e. Therefore, the subgroup of units E (G) = {13e, 29e} is unsplit: e1 =
13e ∈ G1, e2 = 29e ∈ G2.
2) For the zeroless F
[7,8]
(9,3) (8), its multiplicative 3-group G
[5,8]
3 (6) = {7, 15, 23, 31, 39, 47, 55, 63}
has two units e1 = 31, e2 = 63, and it splits into two nonintersecting nonunital cyclic 3-subgroups
(λp = 4 and λp = 2) and the subgroup of units
G1 = 7 1 ×3 = 23, 7 2 ×3 = 39, 7 3 ×3 = 55, 7 4 ×3 = 4 ,
G2 = 15 1 ×3 = 47, 15 2 ×3 = 15 , E (G) = {31, 63} .
There are no ℓµ-reflections, and so E (G) splits out E (G) ∩ G1,2 = ∅.
If all elements are units E (G) = G
[a,b]
n (q), then, obviously, this group is 1-idempotent, and λp = 1.
Assertion 4.28. If F
[a,b]
(m,n) (q) is zeroless-nonunital, then there no n-ary cyclic subgroups in G
[a,b]
n (q).
The subfield structure of F
[a,b]
(m,n) (q) can coincide with the corresponding subgroup structure of the
multiplicative n-ary group G
[a,b]
n (q∗), only if its additive m-ary group has the same subgroup structure.
However, additive m-ary groups of all polyadic fields F
[a,b]
(m,n) (q) have the same structure: they are
cyclic and have no proper m-ary subgroups, each element generates all other elements, i.e. it is a
primitive root. Therefore, we arrive at
Theorem 4.29. The polyadic field F
[a,b]
(m,n) (q), being isomorphic to the (m, n)-field of polyadic integer
numbers Z
[a,b]
(m,n) (q), has no any proper subfield.
In this sense, F
[a,b]
(m,n) (q) can be named a prime polyadic field.

5. CONCLUSIONS
Recall that any binary finite field has an order which is a power of a prime number q = pr
(its
characteristic), and all such fields are isomorphic and contain a prime subfield GF (p) of order p which
is isomorphic to the congruence (residue) class field Z pZ LIDL AND NIEDERREITER [1997].
Conjecture 5.1. Any finite (m, n)-field should contain a smallest subfield which is isomorphic to
one of the prime polyadic fields constructed above, and therefore F
[a,b]
(m,n) (q) can be interpreted as a
polyadic analog of GF (p).
ACKNOWLEDGEMENTS. The author would like to express his sincere thankfulness and gratitude to
Joachim Cuntz, Christopher Deninger, Mike Hewitt, Grigorij Kurinnoj, Daniel Lenz, Jim Stasheff,
Alexander Voronov, and Wend Werner for fruitful discussions.
APPENDIX. MULTIPLICATIVE PROPERTIES OF EXOTIC FINITE POLYADIC FIELDS
Here we list concrete examples of finite polyadic fields which have properties that are not possible
in the binary case (see TABLE 3). Only the multiplication of fields will be shown, because their addi-
tive part is huge (many pages) for higher arities, and does not carry so much distinctive information.
1) The first exotic finite polyadic field which has a number of elements which is not a prime number,
or prime power (as it should be for a finite binary field) is F
[5,6]
(7,3) (6), which consists of 6 elements
{5, 11, 17, 23, 29, 35}, q = 6, It is zeroless and contains two units {17, 35} ≡ {17e, 35e}, κe = 2,
and each element has the idempotence polyadic order λp = 3, i.e. µ3 [x7
] = x, ∀x ∈ F
[5,6]
(7,3) (6). The
multiplication is
µ3 [5, 5, 5] = 17, µ3 [5, 5, 11] = 23, µ3 [5, 5, 17] = 29, µ3 [5, 5, 23] = 35, µ3 [5, 5, 29] = 5,
µ3 [5, 5, 35] = 11, µ3 [5, 11, 11] = 29, µ3 [5, 11, 17] = 35, µ3 [5, 11, 23] = 5, µ3 [5, 11, 29] = 11,
µ3 [5, 11, 35] = 17, µ3 [5, 17, 17] = 5, µ3 [5, 17, 23] = 11, µ3 [5, 17, 29] = 17, µ3 [5, 17, 35] = 23,
µ3 [5, 23, 23] = 17, µ3 [5, 23, 29] = 23, µ3 [5, 23, 35] = 29, µ3 [5, 29, 29] = 29, µ3 [5, 29, 35] = 35,
µ3 [5, 35, 35] = 5, µ3 [11, 11, 11] = 35, µ3 [11, 11, 17] = 5, µ3 [11, 11, 23] = 11, µ3 [11, 11, 29] = 17,
µ3 [11, 11, 35] = 23, µ3 [11, 11, 17] = 5, µ3 [11, 17, 17] = 11, µ3 [11, 17, 23] = 17, µ3 [11, 17, 29] = 23,
µ3 [11, 17, 35] = 29, µ3 [11, 17, 23] = 17, µ3 [11, 17, 29] = 23, µ3 [11, 17, 35] = 29, µ3 [11, 23, 23] = 23,
µ3 [11, 23, 29] = 29, µ3 [11, 23, 23] = 23, µ3 [11, 23, 35] = 35, µ3 [11, 29, 29] = 35, µ3 [11, 29, 35] = 5,
µ3 [11, 35, 35] = 11, µ3 [17, 17, 17] = 17, µ3 [17, 17, 23] = 23, µ3 [17, 17, 29] = 29, µ3 [17, 17, 35] = 35,
µ3 [17, 23, 23] = 29, µ3 [17, 23, 29] = 35, µ3 [17, 29, 29] = 5, µ3 [17, 29, 35] = 11, µ3 [17, 35, 35] = 17,
µ3 [23, 23, 23] = 35, µ3 [23, 23, 29] = 5, µ3 [23, 23, 35] = 11, µ3 [23, 29, 29] = 11, µ3 [23, 29, 35] = 17,
µ3 [23, 35, 35] = 23, µ3 [29, 29, 29] = 17, µ3 [29, 29, 35] = 23, µ3 [29, 35, 35] = 29, µ3 [35, 35, 35] = 35.
The multiplicative querelements are ¯5 = 29, 29 = 5, 11 = 23, 23 = 11. Because
5 1 ×3 = 17e, 5 2 ×3 = 29, 5 3 ×3 = 5, 29 1 ×3 = 17e, 29 2 ×3 = 5, 29 3 ×3 = 29, (A.1)
11 1 ×3 = 35e, 11 2 ×3 = 23, 11 3 ×3 = 11, 23 1 ×3 = 35e, 23 2 ×3 = 11, 23 3 ×3 = 23, (A.2)
the multiplicative 3-ary group G
[5,6]
(7,3) (6) consists of two nonintersecting cyclic 3-ary subgroups
G
[5,6]
(7,3) (6) = G1 ∪ G2, G1 ∩ G2 = ∅, (A.3)
G1 = {5, 17e, 29} , (A.4)
G2 = {11, 23, 35e} , (A.5)

24 STEVEN DUPLIJ
which is impossible for binary subgroups, as these always intersect in the identity of the binary group.
2) The finite polyadic field F
[5,6]
(7,3) (4) = {{5, 11, 17, 23} | ν7, µ3} which has the same arity shape
as above, but with order 4, has the exotic property that all elements are units, which follows from its
multiplication
µ3 [5, 5, 5] = 5, µ3 [5, 5, 11] = 11, µ3 [5, 5, 17] = 17, µ3 [5, 5, 23] = 23, µ3 [5, 11, 11] = 5,
µ3 [5, 11, 17] = 23, µ3 [5, 11, 23] = 17, µ3 [5, 17, 17] = 5, µ3 [5, 17, 23] = 11, µ3 [5, 23, 23] = 5,
µ3 [11, 11, 11] = 11, µ3 [11, 11, 17] = 17, µ3 [11, 11, 23] = 23, µ3 [11, 17, 17] = 11, µ3 [11, 17, 23] = 5,
µ3 [11, 23, 23] = 11, µ3 [17, 17, 17] = 17, µ3 [17, 17, 23] = 23, µ3 [17, 23, 23] = 17, µ3 [23, 23, 23] = 23.
3) Next we show by construction, that (as opposed to the case of binary finite fields) there exist
non-isomorphic finite polyadic fields of the same order and arity shape. Indeed, consider these two
(9, 3)-fields of order 2, that are F
[3,8]
(9,3) (2) and F
[7,8]
(9,3) (2). The first is zeroless-nonunital, while the
second is zeroless with two units, i.e. all elements are units. The multiplication of F
[3,8]
(9,3) (2) is
µ3 [3, 3, 3] = 11, µ3 [3, 3, 11] = 3, µ3 [3, 11, 11] = 11, µ3 [11, 11, 11] = 3,
having the multiplicative querelements ¯3 = 11, 11 = 3. For F
[7,8]
(9,3) (2) we get the 3-group of units
µ3 [7, 7, 7] = 7, µ3 [7, 7, 15] = 15, µ3 [7, 15, 15] = 7, µ3 [15, 15, 15] = 15.
They have different idempotence polyadic orders ord F
[3,8]
(9,3) (2) = 2 and ord F
[7,8]
(9,3) (2) = 1. Despite
their additive m-ary groups being isomorphic, it follows from the above multiplicative structure, that
it is not possible to construct an isomorphism between the fields themselves.
4) The smallest exotic finite polyadic field with more than one unit is F
[2,3]
(4,3) (5) =
{{2, 5, 8, 11, 14} | ν4, µ3} of order 5 with two units {11, 14} ≡ {11e, 14e} and the zero 5 ≡ 5z.
The presence of zero allows us to define the polyadic characteristic (3.17) which is 3 (see TABLE 1),
because the 3rd additive power of all elements is equal to zero
2 3 +4 = 8 3 +4 = 11
3 +4
e = 14
3 +4
e = 5z. (A.6)
The additive querelements are
˜2 = 11e, ˜8 = 14e, 11e = 8, 14e = 2. (A.7)
The idempotence polyadic order is ord F
[2,3]
(4,3) (5) = 2, because for nonunit and nonzero elements
2 2 ×3 = 2, 8 2 ×3 = 8, (A.8)
and their multiplicative querelements are ¯2 = 8, ¯8 = 2. The multiplication is given by the cyclic 3-ary
group G
[2,3]
3 (4) = {{2, 8, 11, 14} | µ3} as:
µ3 [2, 2, 2] = 8, µ3 [2, 2, 8] = 2, µ3 [2, 2, 11] = 14, µ3 [2, 2, 14] = 11, µ3 [2, 8, 8] = 8,
µ3 [2, 8, 11] = 11, µ3 [2, 8, 14] = 14, µ3 [2, 11, 11] = 2, µ3 [2, 11, 14] = 8, µ3 [2, 14, 14] = 2,
µ3 [8, 8, 8] = 2, µ3 [8, 8, 11] = 14, µ3 [8, 8, 14] = 11, µ3 [8, 11, 11] = 8, µ3 [8, 11, 14] = 2,
µ3 [8, 14, 14] = 8, µ3 [11, 11, 11] = 11, µ3 [11, 11, 14] = 14, µ3 [11, 14, 14] = 11, µ3 [14, 14, 14] = 14.
We observe that, despite having two units, the cyclic 3-ary group G
[2,3]
3 (4) has no decomposition
into nonintersecting cyclic 3-ary subgroups, as in (A.3). One of the reasons is that the polyadic field
F
[5,6]
(7,3) (6) is zeroless, while F
[2,3]
(4,3) (5) has a zero (see Assertion 4.26).

REFERENCES
CROMBEZ, G. (1972). On (n, m)-rings. Abh. Math. Semin. Univ. Hamb. 37, 180–199.
DUPLIJ, S. (2012). Polyadic systems, representations and quantum groups. J. Kharkov
National Univ., ser. Nuclei, Particles and Fields 1017 (3(55)), 28–59. Expanded version
arXiv:math.RT/1308.4060v2.
DUPLIJ, S. (2017). Arity shape of polyadic algebraic structures, preprint Math. Inst., Münster, 43 p.,
arXiv:1703.10132.
DUPLIJ, S. AND W. WERNER (2015). Structure of unital 3-fields, preprint Math. Inst., Münster,
17 p., arXiv:math.RA/1505.04393.
LEESON, J. J. AND A. T. BUTSON (1980). On the general theory of (m, n) rings. Algebra
Univers. 11, 42–76.
LIDL, R. AND H. NIEDERREITER (1997). Finite Fields. Cambridge: Cambridge Univ. Press.
MENEZES, A. J., P. C. VAN OORSCHOT, AND S. A. VANSTONE (1997). Handbook of Applied
Cryptography. Boca Raton: CRC Press.
MULLEN, G. L. AND D. PANARIO (2013). Handbook of finite fields. Boca Raton: CRC Press.
POP, A. AND M. S. POP (2002). Some embeddings theorems for (n, 2)-rings. Bul. S¸tiint¸. Univ. Baia
Mare, Ser. B, Fasc. Mat.-Inform. 18 (2), 311–316.
LIST OF TABLES
1 Polyadic characteristics of finite (m, n)-rings 15
2 Content and arities of finite polyadic rings and fields 17
3 Idempotence polyadic orders for the (allowed) finite polyadic fields 20
MATHEMATISCHES INSTITUTE, UNIVERSIT ÄT M ÜNSTER
EINSTEINSTR. 62, D-48149 M ÜNSTER, DEUTSCHLAND
E-mail address: duplijs@math.uni-muenster.de, sduplij@gmail.com
URL: http://homepages.spa.umn.edu/˜duplij

Polyadic integer numbers and finite (m,n) fields

More Related Content

What's hot (20)

Similar to Polyadic integer numbers and finite (m,n) fields (20)

More from Steven Duplij (Stepan Douplii) (20)

Recently uploaded (20)

Polyadic integer numbers and finite (m,n) fields