Steven Duplij, "Polyadic analog of Grothendieck group", arxiv: 2206.14840.pdf

arXiv:2206.14840v1
[math.RA]
15
Jun
2022
POLYADIC ANALOG OF GROTHENDIECK GROUP
STEVEN DUPLIJ
Center for Information Technology (WWU IT), Universität Münster, Röntgenstrasse 7-13
D-48149 Münster, Deutschland
ABSTRACT. We generalize the Grothendieck construction of the completion group for a monoid (being the
starting point of the algebraic K-theory) to the polyadic case, when an initial semigroup is m-ary and the
corresponding final class group K0 can be n-ary. As opposed to the binary case: 1) there can be different
polyadic direct products which can be built from one polyadic semigroup; 2) the final arity n of the class
groups can be different from the arity m of initial semigroup; 3) commutative initial m-ary semigroups
can lead to noncommutative class n-ary groups; 4) the identity is not necessary for initial m-ary semigroup
to obtain the class n-ary group, which in its turn can contain no identity at all. The presented numerical
examples show that the properties of the polyadic completion groups are considerably nontrivial and have
more complicated structure than in the binary case.
CONTENTS
1. INTRODUCTION 2
2. PRELIMINARIES 2
3. GROTHENDIECK GROUP OF COMMUTATIVE MONOID 4
4. n-ARY GROUP COMPLETION OF m-ARY SEMIGROUP 8
4.1. Polyadic direct power construction 8
4.2. Equivalence relations for m-ary semigroups 11
4.3. Polyadic group completion 13
REFERENCES 20
E-mail address: douplii@uni-muenster.de; sduplij@gmail.com; https://ivv5hpp.uni-muenster.de/u/douplii.
Date: of start February 19, 2022. Date: of completion May 8, 2022. Corrections: June 14, 2022.
Total: 25 references.
2010 Mathematics Subject Classification. 16E20, 16T25, 17A42, 18F30, 19A99, 20B30, 20F36, 20M17, 20N15.
Key words and phrases. K-theory, completion group, class group, direct product, polyadic semigroup, arity, polyadic group.

1. INTRODUCTION
The Grothendieck construction of the completion group for a monoid is a ground and starting point
of the algebraic K-theory (see, e.g. KAROUBI [1978], ROSENBERG [1994]). Here we generalize this
construction to the polyadic case, when an initial monoid (or semigroup) and a final class group are m-ary
and n-ary, correspondingly, and we denote such a class polyadic group K
pm,nq
0 . As concrete examples, we
provide several computations for K
pm,nq
0 , including the case, when the arities do not coincide m ‰ n.
2. PRELIMINARIES
We introduce here briefly the usual notation, for details see DUPLIJ [2018]. For a non-empty (un-
derlying) set G the n-tuple (or polyad POST [1940]) of elements is denoted by pg1, . . . , gnq, gi P G,
i “ 1, . . . , n, and the Cartesian product is denoted by Gˆn
”
n
hkkkkkkikkkkkkj
G ˆ . . . ˆ G and consists of all such n-
tuples. For all elements equal to g P G, we denote n-tuple (polyad) by a power pgn
q. To avoid unneeded
indices we denote with one bold letter pgq a polyad for which the number of elements in the n-tuple is
clear from the context, and sometimes we will write
`
gpnq
˘
. On the Cartesian product Gˆn
we define a
polyadic (or n-ary) operation µpnq
: Gˆn
Ñ G such that µpnq
rgs ÞÑ h, where h P G. The operations with
n “ 1, 2, 3 are called unary, binary and ternary.
Recall the definitions of some algebraic structures and their special elements (in the notation of DUPLIJ
[2018]). A (one-set) polyadic algebraic structure G is a set G closed with respect to polyadic operations.
In the case of one n-ary operation µpnq
: Gˆn
Ñ G, it is called polyadic multiplication (or n-ary multi-
plication). A one-set n-ary algebraic structure Mpnq
“
@
G | µpnq
D
or polyadic magma (n-ary magma)
is a set G closed with respect to one n-ary operation µpnq
and without any other additional structure. In
the binary case Mp2q
was also called a groupoid by Hausmann and Ore HAUSMANN AND ORE [1937]
(and CLIFFORD AND PRESTON [1961]). Since the term “groupoid” was widely used in category theory
for a different construction, the so-called Brandt groupoid BRANDT [1927], BRUCK [1966], Bourbaki
BOURBAKI [1998] later introduced the term “magma”.
Denote the number of iterating multiplications by ℓµ, and call the resulting composition an iterated
product
`
µpnq
˘˝ℓµ
, such that
µ1pn1q
“
`
µpnq
˘˝ℓµ def
“
ℓµ
hkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkikkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkj
µpnq
˝
´
µpnq
˝ . . .
´
µpnq
ˆ idˆpn´1q
¯
. . . ˆ idˆpn´1q
¯
, (2.1)
where the arities are connected by
n1
“ niter “ ℓµ pn ´ 1q ` 1, (2.2)
which gives the length of a iterated polyad pgq in our notation
`
µpnq
˘˝ℓµ
rgs.
– 2 – July 1, 2022 at 00:39

PRELIMINARIES
A polyadic zero of a polyadic algebraic structure Gpnq
@
G | µpnq
D
is a distinguished element z P G (and
the corresponding 0-ary operation µ
p0q
z ) such that for any pn ´ 1q-tuple (polyad) gpn´1q
P Gˆpn´1q
we have
µpnq
“
gpn´1q
, z
‰
“ z, (2.3)
where z can be on any place in the l.h.s. of (2.3). If its place is not fixed it can be a single zero. As in the
binary case, an analog of positive powers of an element POST [1940] should coincide with the number of
multiplications ℓµ in the iteration (2.1).
A (positive) polyadic power of an element is
gxℓµy
“
`
µpnq
˘˝ℓµ
“
gℓµpn´1q`1
‰
. (2.4)
We define associativity as the invariance of the composition of two n-ary multiplications. An element of
a polyadic algebraic structure g is called ℓµ-nilpotent (or simply nilpotent for ℓµ “ 1), if there exist ℓµ
such that
gxℓµy
“ z. (2.5)
A polyadic (n-ary) identity (or neutral element) of a polyadic algebraic structure is a distinguished element
e (and the corresponding 0-ary operation µ
p0q
e ) such that for any element g P G we have
µpnq
“
g, en´1
‰
“ g, (2.6)
where g can be on any place in the l.h.s. of (2.6).
In polyadic algebraic structures, there exist neutral polyads n P Gˆpn´1q
satisfying
µpnq
rg, ns “ g, (2.7)
where g can be on any of n places in the l.h.s. of (2.7). Obviously, the sequence of polyadic identities
en´1
is a neutral polyad (2.6).
A one-set polyadic algebraic structure
@
G | µpnq
D
is called totally associative, if
`
µpnq
˘˝2
rg, h, us “ µpnq
“
g, µpnq
rhs , u
‰
“ invariant, (2.8)
with respect to placement of the internal multiplication µpnq
rhs in r.h.s. on any of n places, with a fixed
order of elements in the any fixed polyad of p2n ´ 1q elements tp2n´1q
“ pg, h, uq P Gˆp2n´1q
.
A polyadic semigroup Spnq
is a one-set S one-operation µpnq
algebraic structure in which the n-
ary multiplication is associative, Spnq
“
@
S | µpnq
| associativity (2.8)
D
. A polyadic algebraic structure
Gpnq
“
@
G | µpnq
D
is σ-commutative, if µpnq
“ µpnq
˝ σ, or
µpnq
rgs “ µpnq
rσ ˝ gs , g P Gˆn
, (2.9)
where σ ˝ g “
`
gσp1q, . . . , gσpnq
˘
is a permutated polyad and σ is a fixed element of Sn, the permutation
group on n elements. If (2.9) holds for all σ P Sn, then a polyadic algebraic structure is commutative. A
special type of the σ-commutativity
µpnq
“
g, tpn´2q
, h
‰
“ µpnq
“
h, tpn´2q
, g
‰
, (2.10)
where tpn´2q
P Gˆpn´2q
is any fixed pn ´ 2q-polyad, is called semicommutativity. If an n-ary semigroup
Spnq
is iterated from a commutative binary semigroup with identity, then Spnq
is semicommutative. A
polyadic algebraic structure is called (uniquely) i-solvable, if for all polyads t, u and element h, one can
(uniquely) resolve the equation (with respect to h) for the fundamental operation
µpnq
ru, h, ts “ g (2.11)
where h can be on any place, and u, t are polyads of the needed length.
– 3 – July 1, 2022 at 00:39

PRELIMINARIES
A polyadic algebraic structure which is uniquely i-solvable for all places i “ 1, . . . , n is called a n-ary
(or polyadic) quasigroup Qpnq
“
@
Q | µpnq
| solvability
D
. An associative polyadic quasigroup is called a
n-ary (or polyadic) group. In an n-ary group Gpnq
“
@
G | µpnq
D
the only solution of (2.11) is called a
querelement of g and denoted by ḡ DÖRNTE [1929], such that
µpnq
rh, ḡs “ g, g, ḡ P G, (2.12)
where ḡ can be on any place. Any idempotent g coincides with its querelement ḡ “ g. The unique
solvability relation (2.12) in a n-ary group can be treated as a definition of the unary (multiplicative)
queroperation
µ̄p1q
rgs “ ḡ. (2.13)
We observe from (2.12) and (2.7) that the polyad
ng “
`
gn´2
ḡ
˘
(2.14)
is neutral for any element of a polyadic group, where ḡ can be on any place. If this i-th place is important,
then we write ng;i. In a polyadic group the Dörnte relations DÖRNTE [1929]
µpnq
rg, nh;is “ µpnq
rnh;j, gs “ g (2.15)
hold true for any allowable i, j. In the case of a binary group the relations (2.15) become g ¨ h ¨ h´1
“
h ¨ h´1
¨ g “ g.
Using the queroperation (2.13) one can give a diagrammatic definition of a polyadic group
GLEICHGEWICHT AND GŁAZEK [1967]: an n-ary group is a one-set algebraic structure (universal al-
gebra)
Gpnq
“
@
G | µpnq
, µ̄p1q
| associativity (2.8), Dörnte relations (2.15)
D
, (2.16)
where µpnq
is a n-ary associative multiplication and µ̄p1q
is the queroperation (2.13).
3. GROTHENDIECK GROUP OF COMMUTATIVE MONOID
First, we describe the standard Grothendieck construction (see, e.g. KAROUBI [1978], ROSENBERG
[1994], WEIBEL [1986]) of a commutative group from a commutative semigroup with identity (monoid).
We will use multiplicative notation, which will allow us to provide a straightforward “polyadization”
according to the arity invariance principle DUPLIJ [2021].
Let us have a (binary, arity m “ 2) commutative monoid S ” Sp2q
“
@
S | µ “ µp2q
– p¨q | assoc
D
,
where S is the underlying set, and µ “ µp2q
: S ˆ S Ñ S is the (associative) multiplication in S.
The Cartesian product of two underlying sets S1
“ S ˆ S can be endowed with the componentwise
multiplication pa1, b1q ‚1
pa2, b2q “ pa1 ¨ a2, b1 ¨ b2q, ai, bi P S, to define the binary direct product S1
“
S1p2q
“
@
S1
| µ1
“ µ1p2q
” p‚1
q
D
(which coincides with the direct sum, because of the finite number of
factors in the product).
For convenience and conciseness, we introduce the doubles
S “
ˆ
a
b
˙
P S ˆ S (3.1)
and use vector-like notation for the multiplication (being the Kronecker product of the doubles)
S1 ‚1
S2 “
ˆ
a1
b1
˙
‚1
ˆ
a2
b2
˙
“
ˆ
a1 ¨ a2
b1 ¨ b2
˙
, Si P S1
, ai, bi P S, (3.2)
– 4 – July 1, 2022 at 00:39

GROTHENDIECK GROUP OF COMMUTATIVE MONOID
or in “polyadic” notation
µ1
rS1, S2s “
˜
µ ra1, a2s
µ rb1, b2s
¸
. (3.3)
The associativity of the direct product µ1
follows immediately from that of µ, because of the compo-
nentwise multiplication in (3.3). Since S is a monoid with the neutral element (identity) e P S, satisfying
µ re, as “ µ ra, es “ a, a P S, then the identity of the direct product S1
is the double
E “
ˆ
e
e
˙
, (3.4)
such that
µ1
rE, Gs “ µ1
rG, Es “ G P S ˆ S. (3.5)
Therefore, S1
“ S ˆ S is a commutative monoid, as is S.
Another associative direct product S2
“ S2p2q
“ xS1
| µ2
“ µ2
2 ” p‚2
qy can be obtained using the
“twisted” multiplication of the doubles µ2
defined by
S1 ‚2
S2 “
ˆ
a1
b1
˙
‚2
ˆ
a2
b2
˙
“
ˆ
a1 ¨ b2
a2 ¨ b1
˙
, (3.6)
or
µ2
rS1, S2s “
˜
µ2 ra1, b2s
µ2 ra2, b1s
¸
. (3.7)
The neutral element (identity) E in the “twisted” direct product S2
coincides with (3.4), and therefore
S2
“ S ˆ S is a commutative monoid as well.
The question arises: how to construct a (binary) group corresponding to the binary monoid S “ S2
which would reflect its substantial and important properties? The answer was provided by Grothendieck:
to consider the equivalence relations and corresponding classes in the direct product S ˆ S. Because,
indeed on classes one can define the inverse elements which are needed to build a group (in addition to as-
sociativity and existence of neutral elements which are sufficient for monoids). Here we briefly reproduce
the construction of the Grothendieck group corresponding to the commutative monoid S (sometimes
this is called the symmetrization of S KAROUBI [1978] or the group completion of S ROSENBERG
[1994], WEIBEL [1986]) in multiplicative notation, which will allow us to provide its “polyadization” in
a straightforward way.
Let us consider two kinds of equivalence relations on the direct product S1
“ S ˆ S. The first one
p„1q is reminiscent of “gauge invariance” (in physical language), because it identifies the doubles with
equal “shifts”, such that
ˆ
a1
b1
˙
„1
ˆ
a2
b2
˙
ðñ Dx,yPS
˜
µ ra1, xs
µ rb1, xs
¸
“
˜
µ ra2, ys
µ rb2, ys
¸
, ai, bi P S, (3.8)
and we call it the “gauge” shifts. The second equivalence relation p„2q uses only one “shift” as follows
ˆ
a1
b1
˙
„2
ˆ
a2
b2
˙
ðñ DzPS pµq˝2
ra1, b2, zs “ pµq˝2
ra2, b1, zs , ai, bi P S, (3.9)
so we call this the “twisted” shift (which was used originally by Grothendieck).
Assertion 3.1. Two equivalence relations above coincide „1“„2”„.
– 5 – July 1, 2022 at 00:39

Proof. Let (3.9) holds, then putting x “ µ rz, b2s, y “ µ ra1, zs, we obtain (3.8). Conversely, from (3.8)
with
z “ µ rx, ys (3.10)
it follows that
pµq˝2
ra1, b2, zs “ µ rµ ra1, xs , µ ry, b2ss “ µ rµ ry, b1s , µ ra2, xss “ pµq˝2
ra2, b1, zs . (3.11)

The group completion of the commutative monoid S is defined as the (binary n “ 2) group r
Gp2q
(of
isomorphism classes) being the factorization of S ˆ S by the equivalence relation p„q given in (3.8) or
(3.9). It is called the Grothendieck group, and is usually denoted by K0 pSq “ r
Gp2q
(for this concrete case
of monoid ROSENBERG [1994], WEIBEL [1986])
K0 pSq “ S ˆ S ä„ . (3.12)
The representatives, “observables” (in physical language), are “gauge invariant” doubles (3.8)
r
G “
„
a
b

. (3.13)
The multiplication of the representatives r
G in K0 pSq inherits the product of doubles in S ˆ S (3.3)
r
µ
”
r
G1, r
G2
ı
“
«
µ ra1, a2s
µ rb1, b2s
ff
, (3.14)
where the elements of the resulting binary factor group K0 pSq “ r
Gp2q
“ r
G “
A!
r
G
)
| r
µ
E
(the repre-
sentative doubles r
G (3.13) and the binary operation r
µ “ r
µp2q
on the classes are marked by waves). The
structure of the Grothendieck group for monoids K0 pSq was considered as the staring example in, e.g.,
KAROUBI [1978], ROSENBERG [1994], WEIBEL [1986].
The first equivalence relation p„1q gives the form of the neutral element in K0 pSq
r
E “
„
e
e

„
„
a
a

, @a P S, (3.15)
where e P S is the identity of the monoid S. Indeed the shape (3.15) of the neutral element and commu-
tativity of the initial monoid S allows us to obtain the inverse element in K0 pSq using the multiplication
(3.14) in the following way
r
µ
„„
a
b

,
„
b
a

“
«
µ ra, bs
µ rb, as
ff
“
«
µ ra, bs
µ ra, bs
ff
„
„
e
e

, (3.16)
or
r
µ
”
r
G, r
G
´1
ı
“ r
E, (3.17)
where the unique inverse is
r
G
´1
“
„
a
b
´1
“
„
b
a

. (3.18)
Thus, r
G “ K0 pSq is indeed a (binary) commutative group (of classes) corresponding to the (binary)
monoid S “ S2. Using (3.8) and (3.15), the homomorphism (of monoids) ΦSG : S Ñ K0 pSq can be
written as
ΦSG paq “
„
µ ra, as
a

, @a P S, (3.19)
– 6 – July 1, 2022 at 00:39

where we did not use the identity e P S (this can be important in the “polyadization” below). It follows
from (3.15), that ΦSG can be written with the identity in the form
ΦSG paq “
„
a
e

, @a P S, e P S. (3.20)
Using (3.18) we observe that the image of ΦSG actually generates a group (being K0 pSq), because
„
a
b

“ r
µ
“
ΦSG paq , pΦSG pbqq´1
‰
, @a, b P S. (3.21)
The universal property can be shown in the following way (see, e.g., ROSENBERG [1994]). Let us
consider any commutative (binary) group r
G1
“ xG1
| r
µ1
y and the group homomorphism ΦGG1 : r
G Ñ r
G1
,
then there exists the unique homomorphism ΦSG1 : S Ñ r
G1
, such that
ΦSG1 “ ΦGG1 ˝ ΦSG. (3.22)
Indeed, we derive, using (3.20) and (3.21)
ΦGG1
ˆ„
a
b
˙
“ ΦGG1
˜
r
µ
«„
a
e

,
ˆ„
b
e
˙´1
ﬀ¸
“ r
µ1
«
ΦGG1
ˆ„
a
e
˙
,
ˆ
ΦGG1
ˆ„
b
e
˙˙´1
ﬀ
“ r
µ
“
ΦGG1 pΦSG paqq , pΦGG1 pΦSG pbqqq´1‰
“ r
µ1
“
ΦSG1 paq , pΦSG1 pbqq´1‰
. (3.23)
Conversely, for a given homomorphism ΦSG1 : S Ñ r
G1
the group homomorphism ΦGG1 : r
G Ñ r
G1
is
uniquely defined by (3.23) with ΦSG1 “ ΦGG1 ˝ ΦSG (3.22).
Example 3.2. The simplest example is the commutative monoid of nonnegative integers (natural numbers
with the zero N0) under addition S “ xN0 | µ “ p`qy, and e “ 0. The elements of the K0 pSq “
@
N0 ˆ N0 | r
µ “
`
r̀
˘D
are doubles of natural numbers
ˆ
n
m
˙
P N0 ˆ N0, and the neutral double is
ˆ
0
0
˙
. The equivalence relation (3.9) becomes
ˆ
n1
m1
˙
„
ˆ
n2
m2
˙
ðñ n1 ` m2 “ n2 ` m1, ni, mi P N0, (3.24)
because the monoid S is cancellative. It follows from (3.24) by scaling, that in K0 pSq there exist two
minimal representatives
„
n
0

and
„
0
m

, and from (3.16) we have
„
n
0

r̀
„
0
n

“
„
n
n

“
„
0
0

. (3.25)
Therefore, the representatives
„
n
0

and
„
0
m

can be treated as positive and negative integers in Z,
such that there exists the homomorphism K0 pN0q Ñ Z defined by
„
n
m

ÞÑ n ´ m P Z, (3.26)
– 7 – July 1, 2022 at 00:39

GROTHENDIECK GROUP OF COMMUTATIVE MONOID Polyadic direct power construction
which is a bijection due to (3.24). The universal property (3.23) becomes
ΦGG1
ˆ„
n
m
˙
“ ΦGG1
ˆ„
n
0

´
„
m
0
˙
“ ΦGG1
ˆ„
n
0
˙
´ ΦGG1
ˆ„
m
0
˙
“ ΦGG1 pΦSG pnqq ´ ΦGG1 pΦSG pmqq “ ΦSG1 pnq ´ ΦSG1 pmq . (3.27)
Thus, the Grothendieck group of the monoid N0 of natural numbers with zero is the group of integers Z
under addition K0 pN0q “ Z.
4. n-ARY GROUP COMPLETION OF m-ARY SEMIGROUP
Here we propose the “polyadization” (along the ideas of DUPLIJ [2019]) of the group completion con-
cept using the arity invariance principle DUPLIJ [2021]. By considering the polyadic algebraic structures,
the main differences with the binary case will be the following DUPLIJ [2022] (and refs therein):
1) There can exist several associative polyadic direct products (powers) which can be built from one
polyadic semigroup.
2) If initial m-ary semigroup is commutative, the polyadic direct product can be noncommutative.
3) The arity n of the direct product of m-ary semigroups can be not equal to m.
4) The neutral element (identity) is not necessary for the m-ary semigroup and its power, because
the polyadic analog of the binary inverse element is the querelement of an n-ary group, which is
defined without the usage of a neutral element, and moreover some n-ary groups do not contain an
identity at all.
4.1. Polyadic direct power construction. Let us consider the m-ary semigroup Spmq
“
@
S | µpmq
| total assoc
D
, where S is the underlying set, and µpmq
: Sˆm
Ñ S is the (totally associa-
tive) m-ary multiplication in S. The Cartesian product of two underlying sets S1
“ Sˆm
can be endowed
with an associative n-ary multiplication in various different ways, and also it can be that n ‰ m DUPLIJ
[2022]. Again, we introduce the doubles S “
ˆ
a
b
˙
P S ˆ S, and the polyadic direct product which is
the n-ary semigroup of doubles S1pnq
“
@
S1
| µ1pnq
| total assoc
D
.
There are two possibilities to build the polyadic associative direct product DUPLIJ [2022]:
1) The componentwise construction which corresponds to the standard binary direct product (3.3).
2) The noncomponentwise one, which corresponds to the twisted direct product (3.7).
In the first case, the n-ary semigroup of doubles S1pnq
corresponds to the full polyadic external product
of DUPLIJ [2022].
Definition 4.1. An n-ary componentwise direct product (power) semigroup of doubles consists of two
n-ary semigroups S1pnq
“ Spnq
ˆ Spnq
(of the same arity)
µ1pnq
rS1, S2, . . . , Sns “
˜
µpnq
ra1, a2, . . . , ans
µpnq
rb1, b2, . . . , bns
¸
, Si “
ˆ
ai
bi
˙
, ai, bi P S, (4.1)
where the (total) polyadic associativity (2.8) of µ1pnq
is governed by the associativity of the constituent
semigroup Spnq
.
The simplest case is give by
Definition 4.2. A full polyadic direct product (power) S1pnq
“ Spnq
ˆ Spnq
is called derived, if the
constituent n-ary semigroup Spnq
is derived, such that its operation µpnq
is composition of the binary
operations µp2q
.
– 8 – July 1, 2022 at 00:39

Polyadic direct power construction n-ARY GROUP COMPLETION OF m-ARY SEMIGROUP
In the derived case all the operations in (4.1) have the form (see (2.1)–(2.2))
µpnq
“
`
µp2q
˘˝pn´1q
, µp2q
“ p¨q , µ1pnq
“
`
µ1p2q
˘˝pn´1q
, µp2q
“ p‚1
q . (4.2)
The operations of the derived polyadic direct product can be written as (cf. the binary case (3.3))
µ
1pnq
der rS1, S2, . . . , Sns “ S1 ‚1
S2 ‚1
. . . ‚1
Sn “
˜
a1 ¨ a2 ¨ . . . ¨ an
b1 ¨ b2 ¨ . . . ¨ bn
¸
, (4.3)
and so it is simply a repitition of the binary products (3.3). Therefore, it would be more interesting to
consider nonderived polyadic analogs of the direct product which do not come down to the binary ones.
The nonderived version of the polyadic direct product, the hetero (“entangled”) product, was introduced
in DUPLIJ [2022] for an arbitrary number of constituents. Here we apply it for two m-ary semigroups to
get the nonderived associative direct product, which can have a different arity n ‰ m.
The general structure of the hetero product formally coincides (“reversely”) with the main heteromor-
phism equation DUPLIJ [2018]. The additional parameter which determines the arity n of the hetero
power of the initial m-ary semigroup is the number of intact elements ℓid and number of constituents k.
In our case of k “ 2 multipliers, we have only two possibilities ℓid “ 0, 1. Thus, we arrive at
Definition 4.3. The hetero (“entangled”) power b (square) of the m-ary semigroup Spmq
“
@
S | µpmq
D
is the n-ary semigroup defined on the Cartesian power S1
“ S ˆ S, such that S1pnq
“
@
S1
| µ1pnq
D
,
S1pnq
“ Spmq
b Spmq
, (4.4)
and the n-ary multiplication of doubles Sij “
ˆ
ai
aj
˙
P S ˆ S, ai, aj P S, is given (informally) by
µ1pnq
„ˆ
a1
a2
˙
, . . . ,
ˆ
a2n´1
a2n
˙
“
$
’
’

’
’
%
ˆ
µpmq
ra1, . . . , ams ,
µpmq
ram`1, . . . , a2ms
˙
, ℓid “ 0, n “ m,
ˆ
µpmq
ra1, . . . , ams ,
am`1
˙
, ℓid “ 1, n “ m`1
2
,
, (4.5)
where ℓid “ 0, 1 is the number of intact elements in the r.h.s. of the polyadic direct product. The hetero
power arities are connected by the arity changing formula (with k “ 2) DUPLIJ [2018]
n “ m ´
m ´ 1
2
ℓid, (4.6)
with the integer
n ´ 1
2
ℓid ě 1.
In the case ℓid “ 1, the initial and final arities m and n are not arbitrary, but “quantized” such that the
fraction in (4.6) has to be an integer (see the first row in TABLE 1 of DUPLIJ [2022])
m “ 3, 5, 7, . . .
n “ 2, 3, 4, . . .
(4.7)
The concrete placement of elements and multiplications in (4.5) to obtain the associative µ1pn1q
is gov-
erned by the associativity quiver technique DUPLIJ [2018].
Thus, the classification of the hetero powers for a nonbinary initial semigroup m ě 3 consists of two
limiting cases:
1) Intactless power: there are no intact elements ℓid “ 0. The arity of the hetero power reaches its
maximum and coincides with the arity of the initial semigroup n “ m. Since m ě 3, there is no
binary hetero product.
– 9 – July 1, 2022 at 00:39

n-ARY GROUP COMPLETION OF m-ARY SEMIGROUP Polyadic direct power construction
2) Binary power: the final semigroup is of lowest arity, i.e. binary n “ 2, and it follows from (4.7),
that the only possibility is m “ 3.
Let us consider some concrete examples of the hetero powers.
Example 4.4. Let Sp3q
“
@
S | µp3q
D
be a commutative ternary semigroup, then we can construct its square
of the doubles S “
ˆ
a
b
˙
P S ˆ S in two ways to obtain the associative hetero power
µ1p2q
rS1, S2s “ µ1p2q
„ˆ
a1
b1
˙
,
ˆ
a2
b2
˙
$
’
’

’
’
%
ˆ
µp3q
ra1, b1, a2s
b2
˙
,
ˆ
µp3q
ra1, b2, a2s
b1
˙
,
ai, bi P S. (4.8)
This means that the Cartesian square S ˆ S can be endowed with an associative multiplication µ1p2q
in
two ways, and therefore S1p2q
“
@
S1
| µ1p2q
D
is a noncommutative semigroup, being the hetero power
S1p2q
“ Sp3q
b Sp3q
. If the ternary semigroup Sp3q
has a ternary identity e P S, then S1p2q
has only the
left (right) identity E “
ˆ
e
e
˙
P S ˆ S, since µ1p2q
rE, Ss “ S (µ1p2q
rS, Es “ S), but not the right
(left) identity. Thus, S1p2q
can be a semigroup only, even in the case when Sp3q
is a ternary group.
Example 4.5. Take Sp3q
“
@
S | µp3q
D
to be a commutative ternary semigroup, then the multiplication of
the doubles is ternary and is noncomponentwise (as opposed to the componentwise product (4.1))
µ1p3q
rS1, S2, S3s “ µ1p3q
„ˆ
a1
b1
˙
,
ˆ
a2
b2
˙
,
ˆ
a3
b3
˙
“
ˆ
µp3q
ra1, b2, a3s
µp3q
rb1, a2, b3s
˙
, ai, bi P S. (4.9)
Nevertheless µ1p3q
is associative (and is described by the Post-like associative quiver DUPLIJ [2018]),
therefore the hetero power S1p3q
“ Sp3q
b Sp3q
is indeed the noncommutative ternary semigroup S1p3q
“
@
G ˆ G | µ1p3q
D
. In this case, as opposed to the previous example, the existence of the ternary identity e
in Sp3q
implies the ternary identity in the direct product S1p3q
by E “
ˆ
e
e
˙
, such that
µ1p3q
rS, E, Es “ µ1p3q
rE, S, Es “ µ1p3q
rE, E, Ss “ S. (4.10)
Proposition 4.6. If the initial m-ary semigroup Spmq
contains an identity, then the hetero square n-ary
semigroup S1pnq
“ Spmq
bSpmq
can contain an identity only in the intactless case and the Post-like quiver
DUPLIJ [2018]. For the binary power n “ 2 only the one-sided identity is possible.
Next we consider more complicated hetero power (“entangled”) constructions with and without intact
elements, which is possible for m ě 4 only DUPLIJ [2018].
“
@
S | µp5q
D
be a 5-ary semigroup, then we construct its 5-ary totally associative
hetero square G1p5q
“
@
G1
| µ1p5q
D
using the Post-like associative quiver without intact elements. We
define the 5-ary multiplication of the doubles by
µ1p5q
rS1, S2, S3, S4, S5s “
˜
µp5q
ra1, b2, a3, b4, a5s
µp5q
rb1, a2, b3, a4, b5s
¸
, ai, bi P S. (4.11)
It can be shown that µ1p5q
is totally associative, and therefore S1p5q
“
@
S1
| µ1p5q
D
is a 5-ary commutative
semigroup. If Sp5q
has the 5-ary identity e satisfying
µp5q
re, e, e, e, as “ a, @a P S, e P S, (4.12)
– 10 – July 1, 2022 at 00:39

Equivalence relations for m-ary semigroups n-ARY GROUP COMPLETION OF m-ARY SEMIGROUP
then the hetero power S1p5q
has the 5-ary identity
E “
ˆ
e
e
˙
, e P S. (4.13)
A more nontrivial example is a hetero product (power) which has a different arity than that of the initial
semigroup.
“
@
S | µp5q
D
be a commutative 5-ary semigroup, then we can construct its ternary
associative hetero power S1p3q
“
@
S1
| µ1p3q
D
using the associative quivers with one intact element (see
(4.7) for allowed “quantized” arities). We propose for the triples Si the following ternary multiplication
µ1p3q
rS1, S2, S3s “ µ1p3q
„ˆ
a1
b1
˙
,
ˆ
a2
b2
˙
,
ˆ
a3
b3
˙
“
˜
µp5q
ra1, b2, a3, b1, a2s
b3
¸
, ai, bi P S.
(4.14)
It can be seen that µ1p3q
is totally associative, and therefore the hetero power of 5-ary semigroup Sp5q
“
@
S | µp5q
D
is a noncommutative ternary semigroup S1p3q
“
@
S1
| µ1p3q
D
, such that S1p3q
“ Sp5q
b Sp5q
. If
the initial 5-ary semigroup Sp5q
has the identity satisfying
µp5q
re, e, e, e, as “ a, @a P S, e P S, (4.15)
then the ternary hetero power S1p3q
has only the left ternary identity (4.13) satisfying one relation
µ1p3q
rE, E, Ss “ S, @S P S ˆ S, (4.16)
and therefore S1p3q
is a ternary semigroup with the left identity.
4.2. Equivalence relations for m-ary semigroups. Consider of the extension of the equivalence rela-
tions (3.8) and (3.9) for m-ary semigroups. On the polyadic direct product (square) Spmq
ˆSpmq
we define
two kinds of corresponding binary relations.
Definition 4.9. The first relation p„gaugeq is described by the polyadic “gauge” shifts of the doubles
S “
ˆ
a
b
˙
P S ˆ S in the form
ˆ
a1
b1
˙
„gauge
ˆ
a2
b2
˙
ðñ Dx,yPS
˜
µpmq
“
am´1
1 , x
‰
µpmq
“
bm´1
1 , x
‰
¸
“
˜
µpmq
“
am´1
2 , y
‰
µpmq
“
bm´1
2 , y
‰
¸
, @ai, bi P S. (4.17)
The second relation p„2q is given by the polyadic “twisted” shift
ˆ
a1
b1
˙
„twist
ˆ
a2
b2
˙
ðñ DzPS
`
µpmq
˘˝2 “
am´1
1 , bm´1
2 , z
‰
“
`
µpmq
˘˝2 “
am´1
2 , bm´1
1 , z
‰
, @ai, bi P S.
(4.18)
Proposition 4.10. The relations described by the polyadic “gauge” shifts and “twisted” shift coincide
„gauge“„twist”„m.
Proof. „twistñ„gauge) Let (4.18) hold, then putting x “ µpmq
“
bm´1
2 , z
‰
, y “ µpmq
“
am´1
1 , z
‰
, we get
(4.17).
„gaugeñ„twist) Conversely, if (4.17) takes place, then we can always find t1, . . . , tm´2 P S, such that
z “ µpmq
rx, y, t1, . . . tm´2s and
`
µpmq
˘˝2 “
am´1
1 , bm´1
2 , z
‰
“ µpmq
“
µpmq
“
am´1
1 , x
‰
, µpmq
“
bm´1
2 , y
‰
, t1, . . . tm´2
‰
“ µpmq
“
µpmq
“
am´1
2 , x
‰
, µpmq
“
bm´1
1 , y
‰
, t1, . . . tm´2
‰
“
`
µpmq
˘˝2 “
am´1
2 , bm´1
1 , z
‰
. (4.19)
– 11 – July 1, 2022 at 00:39

n-ARY GROUP COMPLETION OF m-ARY SEMIGROUP Equivalence relations for m-ary semigroups

Proposition 4.11. The relation „m is the equivalence relation on the set of doubles tSu, S P S ˆ S.
Proof. 1) Reflexivity (1 “ 2) of (4.17) is obvious, and for „twist it follows from (4.18), becoming the
identity
`
µpmq
˘˝2
ram´1
, bm´1
, zs “
`
µpmq
˘˝2
ram´1
, bm´1
, zs.
2) Symmetry (1 ÐÑ 2) is evident for „gauge, while (4.18) is symmetric with respect to (1 ÐÑ 2).
3) Transitivity. Let S1 „m S2 and S2 „m S3, and we will prove that S1 „m S3, Si P S ˆ S.
We start from the “gauge”-like relations S1 „gauge S2 and S2 „gauge S3 for the first components of the
doubles
µpmq
“
am´1
1 , x1
‰
“ µpmq
“
am´1
2 , y1
‰
, (4.20)
µpmq
“
am´1
2 , x2
‰
“ µpmq
“
am´1
3 , y2
‰
, Dxi, yi P S. (4.21)
Then we multiply separately the left hand sides and right hand sides of (4.20), (4.21) together with
pm ´ 2q identities t1 “ t1, t2 “ t2, . . . , tm´2 “ tm´2, t1, . . . , tm´2 P S, and derive
`
µpmq
˘˝3 “
am´1
1 , x1, am´1
2 , x2, t1, . . . , tm´2
‰
“
`
µpmq
˘˝3 “
am´1
2 , y1, am´1
3 , y2, t1, . . . , tm´2
‰
. (4.22)
Denoting x3 “
`
µpmq
˘˝2 “
am´1
2 , x1, x2, t1, . . . , tm´2
‰
and y3 “
`
µpmq
˘˝2 “
am´1
2 , y1, y2, t1, . . . , tm´2
‰
,
we obtain in the form
µpmq
“
am´1
1 , x3
‰
“ µpmq
“
am´1
3 , y3
‰
. (4.23)
The second components of the doubles can be treated similarly
µpmq
“
bm´1
1 , x3
‰
“ µpmq
“
bm´1
3 , y3
‰
, (4.24)
and so we have S1 „gauge S3.
The “twisted”-like relations S1 „twist S2 and S2 „twist S3 have the form
`
µpmq
˘˝2 “
am´1
1 , bm´1
2 , z1
‰
“
`
µpmq
˘˝2 “
am´1
2 , bm´1
1 , z1
‰
, (4.25)
`
µpmq
˘˝2 “
am´1
2 , bm´1
3 , z2
‰
“
`
µpmq
˘˝2 “
am´1
3 , bm´1
2 , z2
‰
. (4.26)
We multiply separately the left hand sides and right hand sides of (4.25), (4.26) together with pm ´ 2q
identities t1 “ t1, t2 “ t2, . . . , tm´2 “ tm´2, t1, . . . , tm´2 P S, to get
`
µpmq
˘˝5 “
am´1
1 , bm´1
2 , z1, am´1
2 , bm´1
3 , z2, t1, . . . , tm´2
‰
“
`
µpmq
˘˝5 “
am´1
2 , bm´1
1 , z1, am´1
3 , bm´1
2 , z2, t1, . . . , tm´2
‰
, (4.27)
which can be written as
`
µpmq
˘˝2 “
am´1
1 , bm´1
3 , z3
‰
“
`
µpmq
˘˝2 “
am´1
3 , bm´1
1 , z3
‰
, (4.28)
where z3 “
`
µpmq
˘˝3 “
am´1
2 , bm´1
2 , z1, z2, t1, . . . , tm´2
‰
, z3 P S. Therefore, it follows from (4.28), that
S1 „twist S3.
We next introduce the corresponding equivalence classes.
Definition 4.12. The equivalence class of doubles corresponding to the equivalence relation „m on the
m-ary semigroup Spmq
is the factor
r
S “
„
a
b

m
“ tSu { „m, S “
ˆ
a
b
˙
P S ˆ S, (4.29)
where „m is defined in (4.17) or (4.18).
– 12 – July 1, 2022 at 00:39

Polyadic group completion n-ARY GROUP COMPLETION OF m-ARY SEMIGROUP
Next we consider the structure of the equivalence classes (4.29) in some examples.
Example 4.13 (Negative numbers). Let S “ N´ Ă Z be the set of negative (integer) numbers (without
zero), which do not form a binary semigroup, obviously. However, we can introduce the ternary multi-
plication µp3q
ra, b, cs “ abc (ordinary product in Z), such that Sp3q
neg “
@
N´ | µp3q
D
becomes the ternary
semigroup (of negative numbers). The doubles have the form
Sneg “
ˆ
´p
´q
˙
P N´ ˆ N´, p, q P N. (4.30)
The equivalence relations (4.17) and (4.18) become (m “ 3)
ˆ
´p1
´q1
˙
„gauge
ˆ
´p2
´q2
˙
ðñ Dx,yPN
ˆ
´p2
1x
´q2
1x
˙
“
ˆ
´p2
2y
´q2
2y
˙
, pi, qi P N, (4.31)
and
ˆ
´p1
´q1
˙
„twist
ˆ
´p2
´q2
˙
ðñ Dx,yPN p2
1q2
2z “ p2
2q2
1z. (4.32)
For instance, we compute
ˆ
´1
´2
˙
„
ˆ
´2
´4
˙
„
ˆ
´3
´6
˙
. . . , (4.33)
ˆ
´4
´3
˙
„
ˆ
´8
´6
˙
„
ˆ
´12
´9
˙
. . . . (4.34)
In general, the two elements
ˆ
´p
´q
˙
,
ˆ
´kp
´kq
˙
(4.35)
are in the same class, k P N. Each equivalence class r
Sneg contains the maximal representative
„
´pmin
´qmin

3
corresponding to mininal p “ pmin and q “ qmin, together with other elements (4.35) of the class. Thus,
for the ternary semigroup of negative integers the structure of the binary equivalence classes in Sp3q
neg ˆ
Sp3q
neg ä„3 is given by the following maximal representatives
r
Sneg,pąq “
„
´pmin
´qmin

3
*
, pmin ą qmin, (4.36)
r
Sneg,păq “
„
´pmin
´qmin

3
*
, pmin ă qmin, pmin, qmin P N, (4.37)
where pmin ‰ qmin, and they are mutually prime gcd ppmin, qminq “ 1.
4.3. Polyadic group completion. In general, the procedure of group completion consists of two steps:
1) Endow the set of equivalence classes
!
r
S
)
with the multiplication r
µ
pñq
S (in the binary case (3.14)).
2) Find (a polyadic analog of) an inverse and construct a group (as in (3.18)).
– 13 – July 1, 2022 at 00:39

n-ARY GROUP COMPLETION OF m-ARY SEMIGROUP Polyadic group completion
It is commonly accepted that the multiplication of classes inherits the product of representatives. So
the informal formula of class multiplication coincides with the hetero (“entangled”) power (4.5), indeed
r
µ
pñq
S
„„
a1
a2

ñ
, . . . ,
„
a2n´1
a2n

ñ

“
$
’
’
’

’
’
’
%
„
µpmq
ra1, . . . , ams ,
µpmq
ram`1, . . . , a2ms

m
, ℓid “ 0, ñ “ m,
„
µpmq
ra1, . . . , ams ,
am`1

m
, ℓid “ 1, ñ “ m`1
2
,
, (4.38)
where ℓid “ 0, 1 is the number of intact elements in the r.h.s. of (4.38).
Definition 4.14. The n-ary semigroup of the binary equivalence classes is defined as
r
Spñq
“
A!
r
S
)
| r
µ
pñq
S
E
. (4.39)
Remark 4.15. In general, for a fixed initial semigroup Spmq
, the formulas of the doubles multiplication
(4.5) and classes multiplication (4.38) can be different (as two choices of (4.8)), because in one case we
multiply concrete elements, while in another case–their representatives. Moreover, the final arities n and ñ
need not coincide, if we use different choices in r.h.s. of (4.5) and (4.38). For instance, compare the 5-ary
(4.11) and ternary (4.14) multiplications of doubles obtained from the same 5-ary semigroup. This could
lead to diverse possible polyadic analogs of the group completions of the same semigroup. Nevertheless,
in what follows we will assume that
ñ “ n (4.40)
and choose the same product for elements and their representatives. Therefore, all examples of polyadic
direct products in Subsection 4.1 can be applied for the equivalence class multiplication (4.38) as well.
The polyadic analog of the inverse is the querelement (2.12) DÖRNTE [1929], which does not need an
identity at all. Therefore, the binary formulas (3.16)–(3.18) do not work for m ě 3. So we have
Assertion 4.16. The presence of an identity is not necessary for constructing polyadic groups, it can
be optional and plays no role, and therefore we consider polyadic group completion not of a polyadic
monoid, but of a polyadic semigroup Spmq
.
Let us suppose that we can construct the querelement for any double r
S in the n-ary semigroup r
Spnq
of
equivalence classes, which means that we define the unary queroperation r
µp1q
(see (2.13))
r
S “ r
µp1q
”
r
S
ı
“ r
µp1q
„„
a
b

n

“
„
ā
b̄

n
, (4.41)
which satisfies (2.12)
r
µpnq
”
r
S
n´1
, r
S
ı
“ r
S. (4.42)
Then the n-ary semigroup r
Spnq
becomes a polyadic group.
Definition 4.17. The n-ary group completion of the m-ary semigroup Spmq
is
K
pm,nq
0
`
Spmq
˘
“
A!
r
S
)
| r
µpnq
, r
µp1q
E
, (4.43)
where r
µpnq
and r
µp1q
are the n-ary multiplication of the doubles (4.38) and the queroperation (4.41).
In this notation the standard binary case (3.12) is
K0 pSq “ K
p2,2q
0
`
Sp2q
˘
. (4.44)
– 14 – July 1, 2022 at 00:39

Remark 4.18. The computation of the group completion K
pm,nq
0 with abitrary m and n is possible, if we
know both multiplications: of the initial m-ary semigroup µpmq
and of the n-ary group of equivalent
classes r
µpnq
together with its queroperation r
µp1q
.
Remark 4.19. As opposed to the binary case (4.44), for a given m-ary semigroup there can exist several
associative products of doubles µ1pnq
and corresponding products of classes r
µpnq
(see, e.g. (4.1) and
(4.9) or (4.11) and (4.14)), and therefore the polyadic group completion K
pm,nq
0 is not unique, in general.
The number of different K
pm,nq
0 coincides with the number of distinct associative quivers in the doubles
multiplication (4.5) and class products (4.38) having the same arity shape DUPLIJ [2018].
Remark 4.20. If we consider polyadic groups with the same arity shape of the multiplication of classes
r
µpnq
and the same queroperation r
µp1q
, then we can have a polyadic analog of universality (3.22) for this
fixed completion group arity n.
Example 4.21. Let us compute K
pm,nq
0 for the n-ary componentwise direct power (4.1) of the m-ary semi-
group Spmq
“
@
S | µpmq
D
. Taking to account Remark 4.15 and (4.40) we obtain for the equivalence
classes doubles r
Si the following m-ary multiplication
r
µpmq
compw
”
r
S1, r
S2, . . . , r
Sm
ı
“
˜
µpmq
ra1, a2, . . . , ams
µpmq
rb1, b2, . . . , bms
¸
, r
Si “
„
ai
bi

m
, ai, bi P S, (4.45)
because n “ m for the componentwise multiplication (4.1). We resolve the equation for the querelement
(4.42) as
r
Scompw “
„
ā
b̄

m
„
„
µpmq
raα
, bm´α
s
µpmq
ram´2`α
, b2´α
s

m
, (4.46)
where α “ 0, 1, 2. Because all three solutions are in the same equivalence class, we choose as the
representative the symmetric choice α “ 1
r
Scompw “ r
µp1q
compw
”
r
S
ı
“
„
ā
b̄

m
“
„
µpmq
ra, bm´1
s
µpmq
ram´1
, bs

m
. (4.47)
Therefore, for any m-ary semigroup with m-ary componentwise direct power we obtain
K
pm,mq
0
`
Spmq
˘
“
A!
r
S
)
| r
µpmq
compw, r
µp1q
compw
E
. (4.48)
Example 4.22 (Negative numbers (continued)). Let us introduce for the equivalence classes r
Sneg “
„
´p
´q

3
, p, q P N, gcd pp, qq “ 1 (see Example 4.13), their ternary componentwise multiplication (4.45)
r
µp3q
neg
„„
´p1
´q1

3
„
´p2
´q2

3
„
´p3
´q3

3

“
„
´p1p2p3
´q1q2q3

3
, pi, qi P N. (4.49)
We derive from (4.42) the querelement
r
Sneg “ r
µp1q
neg
”
r
S
ı
“ r
µp1q
neg
„„
´p
´q

3

“
„
´pq2
´p2
q

3
, @p, q P N, gcd pp, qq “ 1. (4.50)
Thus, we obtain the group completion of the negative numbers (with the componentwise multiplication of
classes r
µ
p3q
neg)
K
p3,3q
0 pN´qcomp “
A!
r
Sneg
)
| r
µp3q
neg, r
µp1q
neg
E
. (4.51)
– 15 – July 1, 2022 at 00:39

If we choose the ternary noncomponentwise multiplication (4.9), then
r
µ
p3q
neg2
„„
´p1
´q1

3
„
´p2
´q2

3
„
´p3
´q3

3

“
„
´p1q2p3
´q1p2q3

3
, pi, qi P N, (4.52)
and the querelement will be different from (4.50)
r
Sneg2 “ r
µ
p1q
neg2
”
r
S
ı
“ r
µ
p1q
neg2
„„
´p
´q

3

“
„
´pq2
´q3

3
, @p, q P N, gcd pp, qq “ 1. (4.53)
Therefore, the second polyadic group completion of the negative numbers (with the noncomponentwise
multiplication of classes r
µ
p3q
neg2 (4.52) being different from r
µ
p3q
neg (4.49)) becomes
K
p3,3q
0 pN´qnoncomp “
A!
r
Sneg
)
| r
µ
p3q
neg2, r
µ
p1q
neg2
E
. (4.54)
Another simple example is the set of odd positive numbers.
Example 4.23. Let Nodd “ t2k ` 1u, k P N0, then it is the ternary semigroup Sp3q
odd “
A
Nodd | µ
p3q
odd
E
,
where the multiplication is the ordinary addition
µ
p3q
odd ra, b, cs “ a ` b ` c, a, b, c P Nodd. (4.55)
The doubles (3.1) have the form
Sodd “
ˆ
2k1 ` 1
2k2 ` 1
˙
P Nodd ˆ Nodd, ki P N0. (4.56)
The equivalence relations (4.17) and (4.18) become (m “ 3)
ˆ
2k1 ` 1
2k2 ` 1
˙
„gauge
ˆ
2l1 ` 1
2l2 ` 1
˙
ô Dx,yPNodd
˜
µ
p3q
odd
“
p2k1 ` 1q2
, x
‰
µ
p3q
odd
“
p2k2 ` 1q2
, x
‰
¸
“
˜
µ
p3q
odd
“
p2l1 ` 1q2
, y
‰
µ
p3q
odd
“
p2l2 ` 1q2
, y
‰
¸
, ki, li P N0, (4.57)
and
ˆ
2k1 ` 1
2k2 ` 1
˙
„twist
ˆ
2l1 ` 1
2l2 ` 1
˙
ô Dx,yPNodd
´
µ
p3q
odd
¯˝2 “
p2k1 ` 1q2
, p2l2 ` 1q2
, z
‰
“
´
µ
p3q
odd
¯˝2 “
p2k2 ` 1q2
, p2l1 ` 1q2
, z
‰
.
(4.58)
For instance,
ˆ
3
1
˙
„
ˆ
5
3
˙
„
ˆ
7
5
˙
. . . ,
ˆ
5
1
˙
„
ˆ
7
3
˙
„
ˆ
9
5
˙
. . . ,
ˆ
7
1
˙
„
ˆ
9
3
˙
„
ˆ
11
5
˙
. . . ,
ˆ
1
3
˙
„
ˆ
3
5
˙
„
ˆ
5
7
˙
. . . ,
ˆ
1
5
˙
„
ˆ
3
7
˙
„
ˆ
5
9
˙
. . . ,
ˆ
1
7
˙
„
ˆ
3
9
˙
„
ˆ
5
11
˙
. . . .
(4.59)
In general, two elements ˆ
a
b
˙
,
ˆ
a ` 2
b ` 2
˙
, a, b P Nodd, (4.60)
– 16 – July 1, 2022 at 00:39

are in the same equivalence class
„
a
b

P Nodd ˆ Nodd ä„. The minimal representatives are the doubles
r
Sup “
„
2k ` 1
1

, r
Sdown “
„
1
2k ` 1

, k P N. (4.61)
Thus, the structure of the classes is the following
!
r
S
)
“
!
r
Sup
)
Y
!
r
Sdown
)
,
!
r
Sup
)
X
!
r
Sdown
)
“ ∅. (4.62)
We do not have an identity in the ternary semigroup Sp3q
odd, and therefore the representatives (4.61) cannot
be mapped to negative and positive numbers, as in the binary case (3.26), and moreover their multiplication
is ternary (4.55).
To find the polyadic group structure (i.e. querelements (2.12)), for the ternary multiplication of the
classes r
S we consider two cases:
1) The componentwise multiplication (4.1). We use the ternary multiplication of the classes as ordi-
nary addition
r
µ
p3q
odd1
„„
2k1 ` 1
1

,
„
2k2 ` 1
1

,
„
2k3 ` 1
1

“
„
2 pk1 ` k2 ` k3q ` 3
3

„
„
2 pk1 ` k2 ` k3q ` 1
1

, ki P N, (4.63)
and
r
µ
p3q
odd1
„„
1
2k1 ` 1

,
„
1
2k2 ` 1

,
„
1
2k3 ` 1

“
„
3
2 pk1 ` k2 ` k3q ` 3

„
„
1
2 pk1 ` k2 ` k3q ` 1

, ki P N. (4.64)
Then, apply the general formula for the querelement (4.47) to our ternary case
r
Scompw “
„
ā
b̄

3
“
„
µp3q
ra, b2
s
µp3q
ra2
, bs

3
, (4.65)
which gives using the equivalence relations (4.60)
r
Sup “ r
µ
p1q
odd1
”
r
Sup
ı
“
„
2k ` 1
1

„
„
1
2k ` 1

“ r
Sdown, (4.66)
r
Sdown “ r
µ
p1q
odd1
”
r
Sdown
ı
“
„
1
2k ` 1

„
„
2k ` 1
1

“ r
Sup, k P N. (4.67)
Thus, for the componentwise ternary group completion of the ternary semigroup of odd numbers
Nodd we obtain
K
p3,3q
0 pNoddqcompw “
A!
r
Sup
)
Y
!
r
Sdown
)
| r
µ
p3q
odd1, r
µ
p1q
odd1
E
. (4.68)
2) The noncomponentwise Post-like multiplication (4.9). In this case, we denote the multiplication
of classes corresponding to the double product (4.9) by r
µ
p3q
odd2, then the general formula for the
querelement will be different from (4.65) and have the form
r
Sncompw “
„
ā
b̄

3
“
„
µp3q
ra2
, bs
µp3q
ra, b2
s

3
, (4.69)
– 17 – July 1, 2022 at 00:39

which gives (using the equivalence relations (4.60))
r
Sup “ r
µ
p1q
odd2
”
r
Sup
ı
“
„
2k ` 1
1

„
„
2k ` 1
1

“ r
Sup, (4.70)
r
Sdown “ r
µ
p1q
odd2
”
r
Sdown
ı
“
„
1
2k ` 1

„
„
1
2k ` 1

“ r
Sdown, k P N. (4.71)
Now, each element is a kind of polyadic reflection obeying r
S “ r
S (an analog of g´1
“ g in
binary groups). Thus, the noncomponentwise ternary group completion of the ternary semigroup of
odd numbers Nodd is
K
p3,3q
0 pNoddqncompw “
A!
r
Sup
)
Y
!
r
Sdown
)
| r
µ
p3q
odd2, r
µ
p1q
odd2
E
. (4.72)
Example 4.24. Consider the following 4-ary semigroup of complex matrices
Sp4q
matr “
A
Smatr | µ
p4q
matr
E
, Smatr “ tau , a “
ˆ
u 0
0 0
˙
, u P C, (4.73)
µ
p4q
matr ra1, a2, a3, a4s “ a1 ` εa2 ` ε2
a3 ` a4, (4.74)
where ε “ e
2π
3 , ε3
“ 1.
We construct the doubles S “
ˆ
a
b
˙
P Smatr ˆ Smatr with the componentwise multiplication (4.1)
which does not change the arity. Then we write the “twist-like” equivalence relation (4.18) for our case
ˆ
a1
b1
˙
„twist
ˆ
a2
b2
˙
ðñ DzPS
`
µp4q
˘˝2 “
a3
1, b3
2, z
‰
“
`
µp4q
˘˝2 “
a3
2, b3
1, z
‰
, ai, bi, z P Smatr. (4.75)
It follows from (4.73)–(4.74), that Sp4q
matr is an idempotent semigroup, because µp4q
ra4
s “ a, @a P Smatr,
moreover, µp4q
ra3
, bs “ b, @a, b P Smatr, and so
`
µp4q
˘˝2
ra3
1, b3
2, zs “
`
µpmq
˘˝2
ra3
2, b3
1, zs “ z identically
for @ai, bi, z P Smatr. Thus, we do not have different equivalence classes at all, and therefore for (4.73)
K
p4,4q
0
´
Sp4q
matr
¯
“ 1, (4.76)
in our multiplicative notation.
Let us consider the case, when the completion group of classes has different arity from that of the initial
semigroup, i.e K
pm,nq
0
`
Spmq
˘
with n ‰ m.
Example 4.25. The representatives of the residue (congruence) class Sres “ rr7ss10 “ t10k ` 7u, k P N0
form a 5-ary semigroup with respect to multiplication (see the arity shape for residue classes rrassb in
DUPLIJ [2017]). We take
Sp5q
res “
@
t10k ` 7u | µp5q
res
D
, (4.77)
where 5-ary multiplication µ
p5q
res is the ordinary product (in Z`) µ
p5q
res ra1, a2, a3, a4, a5s “ a1a2a3a4a5,
ai “ 10ki ` 7, ki P N0.
Now the doubles (3.1) are
Sres “
ˆ
10k ` 7
10l ` 7
˙
P rr7ss10 ˆ rr7ss10 , k, l P N0. (4.78)
– 18 – July 1, 2022 at 00:39

The equivalence relations (4.17) and (4.18) for m “ 5 are
ˆ
10k1 ` 7
10l1 ` 7
˙
„gauge
ˆ
10k2 ` 7
10l2 ` 7
˙
ô Dx,yPSres
ˆ
p10k1 ` 7q4
x
p10l1 ` 7q4
x
˙
“
ˆ
p10k2 ` 7q4
y
p10l2 ` 7q4
y
˙
, ki, li P N0, (4.79)
and
ˆ
10k1 ` 7
10l1 ` 7
˙
„twist
ˆ
10k2 ` 7
10l2 ` 7
˙
ô Dx,yPSres p10k1 ` 7q4
p10l2 ` 7q4
z “ p10k2 ` 7q4
p10l1 ` 7q2
z. (4.80)
It follows from the “gauge” shifts (4.79) that the components of the double (4.78) are mutually prime.
Some of the equivalence relations are (we present only those with k1 ă k2)
ˆ
7
17
˙
„
ˆ
77
187
˙
„
ˆ
147
357
˙
. . . ,
ˆ
7
77
˙
„
ˆ
17
187
˙
„
ˆ
27
297
˙
. . . ,
ˆ
17
47
˙
„
ˆ
187
517
˙
„
ˆ
357
987
˙
. . . ,
ˆ
27
37
˙
„
ˆ
297
407
˙
„
ˆ
567
777
˙
. . . ,
ˆ
57
87
˙
„
ˆ
247
377
˙
„
ˆ
437
667
˙
. . . ,
ˆ
87
377
˙
„
ˆ
147
637
˙
„
ˆ
177
767
˙
. . . .
(4.81)
We denote the equivalence class by
r
S “ r
Sres “
„
a
b

res
“
„
10k ` 7
10l ` 7

res
P Sres ˆ Sres ä„, k, l P N0, (4.82)
and we use the minimal representative, when needed and possible. The polyadic group of the equivalence
classes can be obtained, if we define the classes multiplication and the querelement. If we choose the 5-ary
componentwise multiplication (4.45), then we obtain the group completion of the same arity K
p5,5q
0
´
Sp5q
res
¯
,
as in the previous Example 4.23. The more exotic case is to use for the classes doubles r
Sres the changing
arity product (4.14), which gives us the ternary group of equivalence classes, which is built from the 5-ary
semigroup. Indeed, let r
µ
p3q
res be the ternary multiplication of the equivalence classes defined by (cf. (4.14))
r
µp3q
res
”
r
S1, r
S2, r
S3
ı
“
„
p10k1 ` 7q p10k2 ` 7q p10k3 ` 7q p10l1 ` 7q p10l2 ` 7q
10l3 ` 7

res
, ki, li P N0. (4.83)
The querelement r
S can be determined from the manifest form of the ternary multiplication r
µ
p3q
res of the
equivalence classes, and it has the general form
r
S “
„
ā
b̄

res
” r
µp1q
res
”
r
S
ı
„
„
a3
b2
a2
b3

res
„
„
p10k ` 7q3
p10l ` 7q2
p10k ` 7q2
p10l ` 7q3

res
, a, b P Sres, ki, li P N.
(4.84)
Thus, the ternary group of the classes (4.82)
K
p5,3q
0
`
Sp5q
res
˘
“
A!
r
Sres
)
| r
µp3q
res, r
µp1q
res
E
(4.85)
is the completion group of the 5-ary semigroup Sp5q
res (4.77).
– 19 – July 1, 2022 at 00:39

Acknowledgements. The author is deeply thankful to Joachim Cuntz, Siegfried Echterhoff and Chris-
tian Voigt for long-ago explanations of K-theory which indeed now re-awakened my former interest to
consider higher arities in this promising direction, and grateful to Vladimir Akulov, Mike Hewitt, Dimitrij
Leites, Thomas Nordahl, Vladimir Tkach, Raimund Vogl and Alexander Voronov for useful discussions,
and valuable help.
REFERENCES
BORCEUX, F. (1994). Handbook of categorical algebra 1. Basic category theory, Vol. 50 of Encyclopedia of Mathematics
and its Applications. Cambridge: Cambridge University Press. [Not cited]
BOURBAKI, N. (1998). Elements of Mathematics: Algebra I. Berlin: Springer. [Cited on page 2]
BRANDT, H. (1927). Über eine Verallgemeinerung des Gruppenbegriffes. Math. Annalen 96, 360–367. [Cited on page 2]
BRUCK, R. H. (1966). A Survey on Binary Systems. New York: Springer-Verlag. [Cited on page 2]
CLIFFORD, A. H. AND G. B. PRESTON (1961). The Algebraic Theory of Semigroups, Vol. 1. Providence: Amer. Math. Soc.
[Cited on page 2]
DÖRNTE, W. (1929). Unterschungen über einen verallgemeinerten Gruppenbegriff. Math. Z. 29, 1–19. [Cited on pages 4 and 14]
DUPLIJ, S. (2017). Polyadic integer numbers and finite pm, nq-fields. p-Adic Numbers, Ultrametric Analysis and Appl. 9
(4), 257–281. arXiv:math.RA/1707.00719. [Cited on page 18]
DUPLIJ, S. (2018). Polyadic algebraic structures and their representations. In: S. DUPLIJ (Ed.), Exotic Algebraic and
Geometric Structures in Theoretical Physics, New York: Nova Publishers, pp. 251–308. arXiv:math.RT/1308.4060. [Cited on
pages 2, 9, 10, and 15]
DUPLIJ, S. (2019). Arity shape of polyadic algebraic structures. J. Math. Physics, Analysis, Geometry 15 (1), 3–56. [Cited on
page 8]
DUPLIJ, S. (2021). Higher braid groups and regular semigroups from polyadic-binary correspondence. Mathematics 9 (9),
972. [Cited on pages 4 and 8]
DUPLIJ, S. (2022). Polyadic analogs of direct product. Universe 8 (4), 230. [Cited on pages 8 and 9]
GLEICHGEWICHT, B. AND K. GŁAZEK (1967). Remarks on n-groups as abstract algebras. Colloq. Math. 17, 209–219.
[Cited on page 4]
HAUSMANN, B. A. AND Ø. ORE (1937). Theory of quasigroups. Amer. J. Math. 59, 983–1004. [Cited on page 2]
HUNGERFORD, T. W. (1974). Algebra. New York: Springer. [Not cited]
IANCU, L. (1991). On the category of n-groups. Bul. stiin. Univ. Baia Mare, Seria B 7 (1/2), 9–14. [Not cited]
KAROUBI, M. (1978). K-theory. An introduction. Berlin: Springer-Verlag. [Cited on pages 2, 4, 5, and 6]
LAMBEK, J. (1966). Lectures on Rings and Modules. Providence: Blaisdell. [Not cited]
LANG, S. (1965). Algebra. Reading, Mass.: Addison-Wesley. [Not cited]
MAC LANE, S. (1971). Categories for the Working Mathematician. Berlin: Springer-Verlag. [Not cited]
MICHALSKI, J. (1984a). Free products of n-groups. Fund. Math. 123 (1), 11–20. [Not cited]
MICHALSKI, J. (1984b). On the category of n-groups. Fund. Math. 122 (3), 187–197. [Not cited]
POST, E. L. (1940). Polyadic groups. Trans. Amer. Math. Soc. 48, 208–350. [Cited on pages 2 and 3]
ROSENBERG, J. (1994). Algebraic K-theory and its applications. New York: Springer-Verlag. [Cited on pages 2, 4, 5, 6, and 7]
SHCHUCHKIN, N. A. (2014). Direct product of n-ary groups. Chebysh. Sb. 15 (2), 101–121. [Not cited]
WEIBEL, C. A. (1986). The K-Book. An Introduction to Algebraic K-theory. Providence: AMS. [Cited on pages 4, 5, and 6]
– 20 – July 1, 2022 at 00:39

Steven Duplij, "Polyadic analog of Grothendieck group", arxiv: 2206.14840.pdf

More Related Content

Similar to Steven Duplij, "Polyadic analog of Grothendieck group", arxiv: 2206.14840.pdf (20)

More from Steven Duplij (Stepan Douplii) (20)

Recently uploaded (20)

Steven Duplij, "Polyadic analog of Grothendieck group", arxiv: 2206.14840.pdf