Signed integers with ordinary addition and multiplication
form the prime example of a ring.
The original motivation for Ring Theory was to
investigate what properties of the integers are retained in other sets
similarly endowed with an additive operator and a multiplication
distributive over it.
(Only for integers can multiplication be defined as repeated addition.)
The locution distributive ring merely distinguishes the
algebraic term studied here from other meanings of the word
ring itself.
There's no such thing as a "non-distributive ring".
The term associative ring is also used
(non-associative rings are better called algebras).
Abstract rings were introduced in 1871 by
Richard Dedekind (1831-1916) under the name of
Ordnung. Dedekind also pioneered the related notions of
module and ideal
(then limited to prime ideals)
in the context of what we would now recognize as the integers of a number field.
In 1892, David Hilbert (1862-1942)
coined the modern name (Zahlring, then
Ring) which first appeared in print in 1897.
The German word Ring and the English word ring
share with circle two connotations which may have inspired Hilbert:
One denotes a group of people, the other evokes
circularity and looping back (as in the rings of nonzero
characteristics which Hilbert was then concerned with).
The latter etymology is dominant, as shown by the fact that it's solely
responsible for the French translation (anneau) which lacks the first connotation.
In 1914,
Abraham Fraenkel (1891-1965)
introduced an axiomatic definition which was far more restrictive than the
current one, as it required the existence of a multiplicative identity and postulated
that every noninvertible element should be a divisor of zero
(which rules out ordinary integers).
In 1917, the Japanese mathematician Masazo Sono (1886-1969) dropped those requirements
but retained commutativity, quoting compatibility with the
field axioms published in 1903 by
L. Eugene Dickson (1874-1954).
In 1921, Emmy Noether (1882-1935) did the same thing
independently. A few years later, she finally dropped the requirement of commutativity.
Rings with a multiplicative neutral element are the most important type of rings and
many authors will only study those. They're variously known as
"rings with 1",
"rings with unity",
"rings with identity",
"unit rings",
"unitary rings" or "unital rings" (which is the only term I'll use).
Do forsake the linguistic monstrosity "rng"
("ring without i", pronounced "rung"). This joke was launched in 2005 as a Wikipedia stub
which has now morphed into a full hoax, mascarading
as fait accompli,
claiming adamantly that the word ring should only be used for unital rings.
"
You do not get to choose whether or not an article on you appears in Wikipedia,
and you have no veto power over its contents.
The article can cast you as a genius or an imbecile, a respected scientist or a crackpot...
a vandal could replace a page, any page, with total gibberish.
The page on Einstein might have a statement inserted to the effect that he was a Nazi collaborator,
or that his theories have been totally discredited, or that he was a silicon-based life form from Proxima Centauri...
Wikipedia does not operate by your rules but by its own conventions;
I suggest you learn to accept it... I can assure you resistance is futile.
" [sic]
Well, whoever shows so little respect for the feelings of fellow human beings
can't be expected to care much for the integrity of mathematical discourse...
Mercifully, nobody takes seriously a similar joke stating that
a semiring ought to be called a
"rig" (ring without negatives).
For any given (abelian) additive group G,
the ring of square zero G(0)
is the ring in which the product of two elements is always zero.
Bourbaki had to call it a pseudo-ring of square zero
(pseudo-anneau de carré nul)
because, in their work, the bare term ring was reserved for
unital rings.
This trivial case plays a key role in the enumeration of finite rings.
Reputable authors will routinely start a specialized discussion with a fair warning
like:
"Throughout this paper, the symbol R will denote a finite associative
ring with multiplicative identity." That's the way to go...
(2006-02-15) Rings (called distributive rings for disambiguation).
Addition, subtraction and multiplication are defined, division needn't be.
(A, +, . ) is a ring when
addition (+) and multiplication (.)
are well-defined internal operations
over the set A with the following properties:
(A,+) is a commutative
group whose neutral element is called 0 (zero).
Multiplication is an associative
(not necessarily commutative) internal operation which is
distributive over addition. That's to say:
"x
"y
"z
x.(y + z) = x.y + x.z and (x + y).z = x.z + y.z
Multiplicative notations
allow the omission of the dot symbol (.).
Optional properties of a ring can be indicated by specific qualifiers:
Unital Ring : There's a multiplicative neutral element: 1.x = x.1 = x
Commutative Ring :
"x
"y x.y = y.x
Integral Ring : The product of two nonzero elements is nonzero.
Division Ring : Any nonzero element has a multiplicative inverse.
There seems to be universal agreement to define an integral domain as
a commutative integral ring.
The current trend in Ring Theory
(echoed by Wikipedia)
is to call a domain any integral ring
and specify that the term integral domain only applies to commutative ones.
I still advise against the term domain outside idioms like integral domain.
This has the added benefit of compatibility with topology,
where the bare term domain normally denotes a connected open set.
A field is normally defined as a commutative
division ring (a division ring where multiplication is commutative) unless otherwise specified.
I regard as synonymous the locutions noncommutative division ring
and skew field (as well as the semi-acceptable
oxymoron of noncommutative field).
Some authors allow commutativity in a skew field,
in part to translate what the French call a field (corps)
which is a division ring, commutative or not.
Lesser Rings :
For the record, some algebraic structures have been defined which are endowed with an addition
and a multiplication distributive over it,
but without some of the other requirements imposed on rings. Examples:
A semiring is built on an additive monoid
instead of an additive group.
This is to say that a semiring contains a zero element
(neutral for addition) but subtraction need not be well-defined.
In a semiring, 0 is postulated to be absorbent
("x, 0.x = x.0 = 0)
which is a theorem in a ring.
The prototypical example of a semiring is
(,+,×)
the set of natural integers, endowed with usual addition and multiplication.
Also common are the tropical semirings
(e.g., the max-plus algebra where -¥
is included as neutral additive element with the new operations defined by
xÅy = max(x,y) and
xÄy = x+y).
Likewise in idempotent semirings
xÅx = x holds for any x.
Those may be called semirings of characteristic one
(the only ring of characteristic one is
the trivial field).
A near-ring is a ring-like structure where the additive group
is not necessarily abelian (i.e., addition needn't be commutative).
Multiplication must still be associative but distributivity may not hold on both sides.
Thus, in a right near-ring multiplication is only
required to right-distribute over addition, which is to say:
"x
"y
"z
(x + y).z = x.z + y.z
For example, over an
additive group G (abelian or not)
the functions from G to G
form a right near-ring if the addition of functions is defined
pointwise and multiplication is understood as the
composition of functions:
"xÎG
( f + g ) (x) =
f (x) + g (x)
( f . g ) (x) =
f ( g (x) )
Those definitions imply right-distributivity
( f + g ) . h = f.h + g.h since:
(2006-02-15) Divisors, divisors of zero and
properzero-divisors :
In some rings, the product of two nonzero elements can be zero.
In a ring, by definition, an element d is said to be a
divisor of a given element a when there is a
nonzero element x such that:
d x = a or x d = a
In particular (with a = 0 ) a divisor of zero
is an element whose product into some nonzero element is equal to zero.
As usual, a proper divisor
can't be equal to the dividend itself.
So, the zero element itself isn't a proper divisor of zero.
Zero is sometimes called a trivial divisor of zero.
The above definition implies that zero isn't at all a divisor of zero in the trivial ring ({0},+,x)
because of the lack of nonzero elements in it.
d is a left divisor of zero when there's a nonzero v such that dv = 0.
d is a right divisor of zero when there's a nonzero u such that ud = 0.
d is a two-sided divisor of zero when both exist (whether u = v or not).
d is a regular element if neither proposition is true.
Many authors, including myself, use the locution
zero-divisor to denote a proper divisor of zero
(more awkwardly dubbed
nontrivial divisor of zero or nonzero divisor of zero).
I strongly recommend the hyphenation to best indicate that this locution must be taken as a whole.
Likewise, left zero-divisors or right zero-divisors are always understood to be nonzero...
There are no zero-divisors in integral rings
(including integral domains, division rings, fields and skew fields).
This convention is often implicitly made
(otherwise it wouldn't make much sense to speak of "rings without zero-divisors")
but it's prudent to state explicitly before a formal argument that a zero-divisor
can't be zero (even in the ring {0}).
Conversely, the locution "divisor of zero" can
be read grammatically as above, which makes it include zero.
Two elements x and y are said to be (mutually) orthogonal when:
x y = 0
and
y x = 0
If some nth power of an element is zero,
that element is said to be nilpotent.
Clearly, a nonzero nilpotent element is a zero-divisor.
x n = 0
The simplest example is
the residue 2 in the ring
/
4
(i.e., the ring formed by the four possible residues of integers
modulo 4)
where:
Those are the names given to elements which are not divisisor
of zero to a specified side. Equivalently, muliplication to one side
by an element which is regular to that side is injective.
For example, if the element r is right-regular,
matching right-factors of r can be cancelled from an equation:
(2019-05-11) Units are invertible elements in a unital ring.
In a finiteunital ring, every non-invertible
element is a divisor of zero.
To put it bluntly, every element divides either zero or unity.
In 1914, Abraham Fraenkel (1891-1965)
had imposed this property as an axiom for all rings,
which ruled out the star of the show (the ordinary integers)
since most integers are neither units nor divisors of zero.
The postulate is obsolete but it's a theorem for finite rings.
Proof :
For any element a of the finite unital ring
A ¹ {0}, if
the map which sends x to a x
is injective,
then it's also surjective (since A is
finite) so there's an
element v whose image is 1, meaning a v = 1.
On the other hand, if that map isn't injective, then two different
elements x and y have the same image and
x-y = v verifies a v = 0
with v ¹ 0.
Likewise, considering the map which sends x to x a, we see
that there's a nonzero element u such that u a
is either 0 or 1.
We can't have u a = 1 and
a v = 0 with a nonzero v
(or else we'd have v = u a v = 0) or vice-versa.
Therefore, either a is a unit
(with the same inverse on both sides u = u a v = v)
or it's a two-sided divisor of zero.
The units of a unital ringA form a multiplicative groupe
denoted A* or U(A) (sometimes E(A)
as the German name for unit is Einheit).
That's just a special case of the group M* formed by
all the invertible elements in a multiplicative monoidM,
so the same notation can be used.
Units in a Ring (7:13)
by Liliana de Castro (Socratica, 2017-09-06).
(2006-02-15) Ideal I in a Ring A An ideal is a multiplicatively absorbent additive subgroup.
An ideal is an additive subgroup that contains
a product whenever it contains a factor.
(Such a thing is called multiplicatively absorbent, or
absorbent for short.)
A subring is a ring contained in another one (using the same operations).
With the traditional definition of a ring adopted here (where we don't require a ring to be unital)
a subring is simply a nonempty subset closed under subtraction and multiplication.
That much is clearly true for an ideal. Thus, ideals can also be defined as absorbent subrings.
However, a subring (or an ideal) of a unital ring need not be unital itself.
Therefore, those who are adamant that only unital rings should be called rings
will reject those subring characterizations. Their loss.
For a left-idealI,
the product ax is in I whenever x is:
"aÎA,
aI Í I
For a right-idealI,
the product xa is in I whenever x is:
"aÎA,
Ia Í I
Unless otherwise specified,
an ideal is both a right-ideal and a left-ideal.
Every ring is a (two-sided) ideal of itself, called the unit ideal.
Any other ideal (one-sided or two-sided) is said to be proper.
All rings with more than one element have at least one proper ideal,
namely the single-element subring {0} which is called the zero ideal.
The ring {0} doesn't have any proper ideals.
The unit ideal and the zero ideal
are both called trivial (whether or not they are the same).
The sum and the intersection of two same-sided ideals
are ideals on that side.
The universal convention due to
Hermann Minkowski (1864-1909) is that
the sum (resp. product) of two sets
is defined to be the set whose elements are sums
(resp. products) of elements from those two sets.
If I is a left-ideal and J is a right-ideal,
then I.J is a double-sided ideal, but neither J.I nor
I Ç J are necessarily one-sided ideals.
A one-sided or two-sided ideal is called maximal
when it's a proper ideal not contained in any other proper
ideal of the same kind. Thus, the whole ring is never a maximal ideal
of itself. {0} is maximal only if the ring is nonzero and
doesn't have any other proper ideals of the relevant kind.
A ring A is said to be a filial ring when any ideal of an ideal
of A is also an ideal of A itself.
Using Wielandt's symbols that's to say:
J ⊲ I ⊲ A Þ J ⊲ A
So, ring A is filial when all its
2-accessible subrings are ideals, defining a subring S
as n-accessible when a sequence Ai
exists which verifies the following (such an S is precisely n-accessible if
not k-accessible for k<n):
S =
A0 ⊲
A1 ⊲ ... ⊲
An = A
The ideal generated (to one side) by some set X is well-defined as the
intersection of all ideals (to that side) containing X,
since any intersection of ideals of a given kind is an ideal of the same kind.
An ideal generated by a single element is called a principal ideal.
One example of such a right-ideal is the set a A of all
right-multiples of the element a in the ring A
(e.g., 2
is the ideal of all even integers).
A ring, like ,
whose ideals are all principal is called a
principal ring.
Such a ring is called a principal integral domain
(abbreviated PID) if it has no proper divisors of zero
(i.e., the product of two nonzero elements is never zero).
Note that Bourbaki requires
a principal ring to be a PID.
In a field, there are only two ideals,
namely {0} and the whole field. They are both principal
(respectively generated by the elements 0 and 1) so a field
is a PID.
(HINT: If an ideal of a field contains a nonzero element
it also contains the product of that element by its inverse, which is 1.)
Every skew field is a PID too.
Ideals were introduced in 1871 by Richard
Dedekind (1831-1916) as he investigated what are now called
completely prime ideals, namely ideals which don't contain a
product unless they contain at least one factor
(e.g., among integers,
the multiples of a prime number have that property).
In commutative rings, those are just prime ideals,
namely ideals which don't contain a product of two ideals unless they contain at least one of them.
A completely prime ideal is always a prime ideal,
but the converse may not be true in the noncommutative case. That distinction was introduced
in 1928 by Wolfgang Krull (1899-1971).
The radical Rad(I) of an ideal I
is the set of all ring elements which have one of their powers in I.
The radical of an ideal is an ideal. If an ideal is
the radical of another it's called a radical ideal.
Every prime ideal is a radical ideal.
Modulo a radical ideal,
there are no nilpotentresidues.
In particular, the ideal Rad({0}), called the nilradical
of A, is the set of all nilpotent elements of A.
It's the intersection of all prime ideals of A.
The Jacobson radical J(A) of a ring A
is the intersection of all the maximal ideals of A.
Since all maximal ideals are prime, the
nilradical is contained in the Jacobson radical.
(2006-02-15) Residue Ring (modulo a given ideal I of a ring A)
The ring A/I which consists of all residue classes moduloI.
Modulo an idealI of a ring A,
the residue-class (or just residue ) [x] of an element x of A
is the set of all elements y of A
for which x-y is in I.
The set of all residues modulo I is denoted A/I.
It's a ring, variously called quotient ring, factor ring,
residue-class ring or simply residue ring.
For example, /
4 is the ring formed by
the four residue classes modulo 4, whose addition
and multiplication tables are shown at right.
(Note that "2" is a nilpotentdivisor of zero.)
+
0
1
2
3
0
0
1
2
3
1
1
2
3
0
2
2
3
0
1
3
3
0
1
2
´
0
1
2
3
0
0
0
0
0
1
0
1
2
3
2
0
2
0
2
3
0
3
2
1
The ring
/ p
=
( / p, +, ×)
is a field if and only if p
is prime.
The notation p
instead of /
p
is not acceptable, as the former is reserved for the
(infinite) ring of p-adic integers.
In particular, the Boolean field
/
2
has just two elements; 0 and 1
(called bits nowadays).
It's used below to construct a nontrivial ring,
which provides concrete examples of many abstract concepts.
(2019-04-19) Ring Homomorphisms and Isomorphisms
Function f from one ring to another, which respects the ring operators:
f (x+y) = f (x) + f (y)
f (x y) = f (x) f (y)
If such an homomorphism is bijective,
it's called an isomorphism.
An isomorphism from one ring to itself is called an automorphism.
If f is an homomorphism from ring A to ring B
and J is a subring of B, then
I = f -1 (J) is a subring.
It's an ideal of A
if J is an ideal of B.
I is called the contraction of J.
Kernel :
In particular, f -1 ({0B}) is an ideal,
called the kernel of f :
If all the elements in this sequence are nonzero, the ring is said to have
zero characteristic.
Otherwise, the vanishing indices are multiples of the least of them,
which is called the characteristic of the ring, denoted char(A).
The only ring of characteristic 1 is the trivial
field (where 1 = 0).
The characteristic of a nontrivial unital ring without zero-divisors
is either zero or a
prime number. (HINT:
any integer (1+1+...) corresponding to a
prime divisor of a
composite characteristic would be a zero-divisor.)
In particular, the characteristic of any nontrivial
field (or skew-field)
is either 0 or a prime number.
The characteristic
of a non-unital ring is defined as the least positive integer
p such that a sum of p identical terms always vanishes
(if there's no such p, then the ring is said to have zero characteristic).
The characteristic of a ring depends only on its
additive group (e.g., if that additive group is the
cyclic groupCn ,
then the characteristic is l(n),
the reduced totient of n).
A finite ring can't have zero characteristic.
The characteristic of a finite ring always divides its number of elements.
(cf. structure of abelian groups).
The former relation is due to commutativity.
The latter relation comes from
Newton's binomial formula, with the added remark
that the binomial coefficient C(p,k)
is divisible by p, if p is prime, unless k is 0 or p.
Thus, the map defined by F(x) = xprespects both addition and multiplication.
This ring automorphism,
is called the Frobenius map, in honor of
F. Georg Frobenius (1849-1917)
who discovered the relevance of such things to
algebraic number theory, in 1880.
(2019-01-27) The SUN ring
(Smallest Unital Noncommutative ring).
It can be represented by 8 triangular
2 by 2 boolean matrices.
We may construct it as a vector space over the integers
modulo 2 by choosing any basis of three independent matrices.
This gives an additive group (of characteristic 2)
isomorphic to the 8 integers from 0 to 7 endowed with
bitwise addition (just like the additive group of the Galois field GF(8)
which see for the explicit addition table).
We'll use the ring isomorphism defined by the following
correspondence between powers of two and lower-triangular binary 2×2 matrices
(we also give equivalent upper-triangular matrices
highlighted in yellow
to make explicit the isomorphism between the two types of triangular matrices).
Upper Triangular
1 0 0 1
0 1 0 0
1 0 0 0
Lower Triangular
1 0 0 1
0 0 1 0
0 0 0 1
Binary Name
1
2
4
Choices for 1 and 2 were engineered
to give the names 0,1,2 and 3
to the four elements whose unique properties make them invariant under anyring automorphism.
Those form a commutative subring whose
multiplication table is thus made identical to the multiplication table of
/
4.
Four possible matrices can then be assigned to 4.
I picked one arbitrarily, as I'm still looking for a good reason to distinguish one as canonical.
×
0 0 0 0
1 0 0 1
0 1 0 0
1 1 0 1
1 0 0 0
0 0 0 1
1 1 0 0
0 1 0 1
0 0 0 0
1 0 0 1
0 0 1 0
1 0 1 1
0 0 0 1
1 0 0 0
0 0 1 1
1 0 1 0
0
1
2
3
4
5
6
7
0 0 0 0
0 0 0 0
0
0
0
0
0
0
0
0
0
1 0 0 1
1 0 0 1
1
0
1
2
3
4
5
6
7
0 1 0 0
0 0 1 0
2
0
2
0
2
0
2
0
2
1 1 0 1
1 0 1 1
3
0
3
2
1
4
7
6
5
1 0 0 0
0 0 0 1
4
0
4
2
6
4
0
6
2
0 0 0 1
1 0 0 0
5
0
5
0
5
0
5
0
5
1 1 0 0
0 0 1 1
6
0
6
2
4
4
2
6
0
0 1 0 1
1 0 1 0
7
0
7
0
7
0
7
0
7
Let's use this example as an opportunity to review the basic concepts:
Element 1 is neutral for multiplication: 1 x = x 1 = x for any x.
The automorphisms and anti-automorphisms of the SUN ring :
0
1
2
3
4
5
6
7
Identity
0
1
2
3
4
5
6
7
Automorphism
0
1
2
3
6
7
4
5
Anti-automorphisms
0
1
2
3
5
4
7
6
0
1
2
3
7
6
5
4
As both anti-automorphisms are involutions, they can be used to give SUN
the structure of an involutive ring, in two different ways.
Using the Cayley-Dickson construction
with either choice yields an algebraic stucture of 16 elements which is not a ring
(the composite structure isn't multiplicatively associative since basic multiplication isn't commutative).
If f (x) is the number of 1's in the diagonal of [the matrix for] x, then:
f (x y) ≤ min ( f (x), f (y) )
´
1
i
j
1
1
i
j
i
i
0
0
j
j
j
j
More generally, for any commutative unital ring A,
we may use the above pattern to define SUN(A) as a
3-dimensional module over A,
where multiplication is determined by the multiplication table at right, for a
basis (1,i,j) of the module.
For short, SUNq = SUN( GF(q) )
when q is a prime-power.
Furthermore, SUN = SUN2 denotes the ring discussed above.
For a prime p, SUNp is the only unital noncommutative ring of order p3.
Alternately, we could have chosen to define the SUN ring as generated by two
distinct elements u and v obeying the following relations:
u2 = u
v2 = v
u v = v
v u = u
With the above numbering, that's only true when {u,v} = {4,6}.
We could also have used the relations below, characterizing {u,v} = {5,7} :
(2019-02-14) Idempotent elements & Peirce decompositions (1870)
An idempotent element splits a ring into a direct sum of 4 components.
An idempotent element of a ring A is a solution of the equation:
x 2 = x
A Boolean ring is a ring where
every element is idempotent.
A unital Boolean ring is called a Boolean algebra.
Every ring has at least one idempotent element (namely 0) and every
unital ring with more than one element has another trivial one (namely 1).
If a unital ring has other idempotent elements (said to be nontrivial)
then it has at least one zero-divisor because,
in a unital ring, the above equation reduces to the following zero product
of two factors, neither of which is zero when x is neither 0 nor 1:
x (1-x) = 0
Left-multiplication or right-multiplication by an idempotent is analogous to
a geometrical projection (projecting a projected image doesn't change it).
This geometrical analogy suggests that a ring can be expressed as a direct sum of the images
of two such complementary projections, as shown next.
Peirce Decompositions (Benjamin Peirce, 1870) :
If e is an idempotent of the unital ring A
(i.e., e2 = e) then A
can be split into a direct sum in two different ways
(using Minkowski notations):
The left Peirce decomposition :
A = e A Å (1-e) A
The right Peirce decomposition :
A = A e Å A (1-e)
In the noncommutative case, those two are distinct and we may apply them successively to
obtain the full (two-sided)
Peirce decomposition of A into four components,
best presented formally as a matrix:
A @
e A e (1-e) A e
e A (1-e) (1-e) A (1-e)
The right-hand-side denotes a set of 2×2 matrices whose coefficients are
in four subsets of A specified by Minkowski operations.
Addition and multiplication of such matrices reduce to component operations in
A according to the usual rules.
(It's a simple exercise to show that this set of matrices is stable under addition
and multiplication.)
The decomposition can be trivial (e.g., with e = 1).
The reader may want to work out the example e = 4
with numbered elements of the SUN ring,
which recovers a representation in terms of triangular boolean matrices:
SUN2 = A @
4 A 4 5 A 4
4 A 5 5 A 5
=
{0,4} {0}
{0,2} {0,5}
More generally, what we did for e1 = e and
e2 = 1-e can be done for any set of n
pairwise orthogonalidempotents
adding up to unity:
The younger Benjamin Peirce (rhyming with terse or
purse) taught mathematics at
Harvard University
for nearly 50 years.
In 1832, he proved that an odd perfect number
(if such a thing exists) can't have fewer than 4 distinct prime factors
(that's been upgraded to 10, now).
So far, the Mathematics
Genealogy Project, found 6038 doctoral descendants of his. This writer descends from
him through a lineage of seven academic generations, namely
His son
Charles
"Santiago" Sanders Peirce (1839-1914) made a
fleeting mark in epistemology
by introducing pragmatism
as a guide for scientific research
(a viewpoint which is now deprecated).
Twenty years after the hints of Hermann Grassmann,
Charles S. Pierce published an axiomatization of natural numbers in a paper entitled
"On the Logic of Number" (1881).
The modern version (Peano's axioms)
is usually credited to Dedekind (1888)
and Peano (1889).
Peirce based his own Algebra of Logic (1885)
on what's now called Peirce's law.
(2019-01-25) Enumeration of Finite Rings
How many different rings with n elements are there?
A ring is unital (resp. commutative)
if and only if all its components are. So,
the same enumeration method applies either to unrestricted rings or to
those which are required to be unital and/or commutative.
The ensuing four possible types of enumerations are tabulated below in separate columns.
The only possible additive group for a ring of prime order p
is the cyclic groupCp.
As the characteristic of a finite ring must divide its order,
it's either 1 or p. This yields only two possible rings. Both are commutative:
The Galois field
GF(p) = Fp =
/p
is a unital ring, of course.
There are 11 different rings of order p2,
for any prime p. Of those,
9 are commutative and 4 are unital (all the unital ones are commutative):
3 such rings are decomposable: (Cp(0))2,
Cp(0) × Fp and (Fp)2
Only the last one is unital. All three are commutative.
5 other rings over the additive group (Cp)2 are indecomposable.
These, too, are algebras of dimension 2 over the finite field
Fp :
The two unital ones are commutative: Fp2 and GR(p,2)
One is commutative but not unital.
The other two are neither commutative nor unital.
The 3 indecomposable rings over the group Cp2 are commutative.
Only
/p2
is unital.
The enumeration for order p3 is far more delicate.
Preliminary results were published in
1947 by Robert F. Ballieu, a leading professor at
Université catholique de Louvain
and a former foreign student at
ENS (Ulm, 1936).
A first attempt at a complete enumeration was published in 1973
by Gilmer & Mott.
It was corrected by Antipkin & Elizarov in 1982:
20 such rings are decomposable and either
32 (for p=2) or 3p+30 (for p≥3) are not.
Number of rings of order n = pk (p prime)
broken down by additive groups.
The entries in bold were first obtained by
Christof Nöbauer before 2002.
Because the corresponding enumerations are
multiplicative functions
(entirely determined by their values at prime-powers) the above table
is sufficient to enumerate rings of all orders below 64
(except 32 for rings not required to be unital).
This yields the following four tables:
By comparing the last two tables, we see that the unique
smallest unital noncommutative ring has
8 elements; it's the SUN ring studied above.
Next up, there are 13 distinct noncommutative unital rings of order 16.
Noncommutative unital rings exist only for orders divisible by a cube :
Indeed, if n = k p3 ,
then one example of a noncommutative unital ring of order n
is given by the direct productSUNp × /k
All unital rings of cubefree
order n (A004709) are commutative
because the above enumerations show that there are as just as many commutative unital rings
of order n as there are unital rings of the same order.
(HINT : That much is true when n is a prime or the square of a prime.)
Likewise all rings of squarefree
order (A005117) are commutative.
Noncommutative rings exist only for orders divisible by a square :
(2019-01-25) Simple rings do not possess any nontrivial ideals.
They're called quasi-simple by Bourbaki.
Ideals are to rings what divisors are to integers or what
normal subgroups are to groups,
inasmuch as they allow a standard decomposition when nontrivial examples exist.
Otherwise, we indicate irreducibility to anything simpler by calling those things simple rings,
primes or simple groups.
{0} isn't considered a simple ring, for the same reason
the number 1 isn't called a prime
or {e} isn't considered a simple group: If this convention wasn't made,
fundamental theorems would be tougher to state (and use).
(2019-02-12) Coprime ideals in a commutative ring A:
Two idealsI and J are called coprime when
I + J = A.
The term comaximal is more rarely used for the same concept,
which generalizes coprime integers in a way
which allows the following generalization of the
Chinese remainder theorem.
(2019-01-18) Primary Ideals
They are to prime ideals what prime-powers are to prime numbers.
A primary decomposition is an expression of an ideal as an intersection of finitely many primary ideals.
It's analogous to the factorization of a positive integer into a product of prime-powers
(fundamental theorem of arithmetic).
The fundamental theorem about primary decompositions is the
Lasker-Noether theorem, which was the main motivation for
the definition of Noetherian rings (introduced below).
By definition, an ideal I is said to be primary
when, for any product x y in it, either x or some power of
y is also in I.
The set
A,
of the sequences whose terms are elements of the ring A
has the structure of a ring (dubbed
formal power series over A)
if endowed with directaddition (the n-th term of a sum being the sum of the n-th terms of the
two addends) and the
Cauchy multiplication defined above.
The set A( )
consists of those sequences which have only finitely many nonzero terms.
It forms a subring of the above ring,
better known as the [univariate]
polynomials over A, denoted A[x]
and discussed next.
(2006-04-06) A[x] :
Ring of formal polynomials over a ring A It's endowed with
component-wise addition and Cauchy multiplication.
A finite sequence of elements of a ring A
(or, equivalently, a sequence with finitely many nonzero elements)
is called a polynomial over A.
The set of all such polynomials is a ring
(often denoted A[x]
where x is a "dummy variable")
which is a subring of the aforementioned ring of
"formal power series",
under direct addition and
Cauchy multiplication.
Each term of the sequence defining a polynomial is called a coefficient.
The degree of a polynomial is the highest rank
of its nonzero coefficients (ranks start at zero).
The null polynomial ("zero") has no nonzero coefficients and its degree
is defined to be -¥ (negative infinity).
Thus, for polynomials over an integral ring
(a ring without divisors of zero)
the degree of a product is always the sum of the degrees of the
factors.
If the ring A is commutative, the polynomials so defined
match very well the intuition we may have acquired by using X
as the symbol for a so-called unknown variable and forming
new expressions by addition and multiplication using X and
some constants taken from the ring A.
In the noncommutative case, this intuition fails unless we imagine that
X commutes with every element of A.
This strange thing is what it means to state that the product
of two polynomials is equal to their Cauchy product.
Formal Polynomials vs. Polynomial Functions :
To a polynomial
(a0 , a1 ... an )
of degree n, we associate a
functionf :
n
f (x) =
å
ai xi
i = 0
However, that function and the polynomial which defines it
are two different things entirely...
For example, over the finite field GF(q),
the distinct polynomials x and xq
correspond to the same function.
In other words, the map from polynomials to
polynomial functions
need not be injective.
However, that map is indeed injective in the special case of polynomials over
ordinary signed integers or any superset thereof,
including rational, real, surreal or complex numbers.
(As is the case for p-adic integers and p-adic numbers.)
Whenever the distinction between a polynomial and its
associated function must be stressed,
the former may be called a formal polynomial.
Similarly, infinite sequences of coefficients
are called formal power series
Unlike polynomials,
those may or may not be associated with a convergentpower series which would define a proper function...
Over a noncommutative ring, the concept of polynomials doesn't break down, but
the above association of a polynomial with a function is dubious at best,
in particular because the value of a product of polynomials at some point need
not be the product of their values at that point.
Polynomial Functions over a Noncommutative Ring
Arguably, some polynomial functions are not even associated with a
formal polynomial in the above sense.
Indeed, we could consider functions which are obtained from
variables and constants by some fixed recipe involving a finite number
of additions and multiplications, without assuming any
type of commutativity among constants or variables.
This topic is under-investigated in the literature.
Twisted Polynomial Ring
It's the set of univariate polynomials with coefficients appearing formally at the left
of the powers of the variable X if multiplication is defined using
a nontrivial automorphism
s in the ring of coefficients (e.g., the
Frobenius map in a commutative ring of coefficients
of nonzero characteristic) with the following
postulated formal identity:
X b = s(b) X
In other words, the n-th coefficient of the twisted product is:
n
cn =
å
ai s ( bn-i )
i = 0
Since s isn't trivial,
multiplication of twisted polynomials isn't commutative even when the ring of coefficients is
(it stays Noetherian).
(2019-01-18) Noetherian Rings
Rings which don't contain any infinite ascending chains of ideals.
Four concepts can be defined, which use the concept of an ascending chain,
defined below. They coincide for commutative rings:
Weakly Noetherian : No infinite ascending chain of ideals exists.
Left-Noetherian : No infinite ascending chain of left-ideals
Right-Noetherian : No infinite ascending chain of right-ideals
Strongly Noetherian : Bothleft-Noetherian and right-Noetherian.
A weakly Noetherian ring need not be either
left-Noetherian or right-Noetherian,
since the lack of a chain of sets which are both left-ideals and right-ideals
doesn't disallow a chain of sets with either property.
An ascending chain of sets is defined as a sequence of sets In where
every set is contained in its successor, which is to say:
I0 Í
I1 Í
I2 Í
I3 Í ...
In Í
In+1 ...
That concept can be defined for any objects endowed with some
partial ordering.
So can the mirror notion of a descending chain.
A chain is dubbed infinite if it
contains infinitely many distinct terms. Otherwise,
there will be an index N after which all terms are equal to each other.
Such a chain is said to be finite, to terminate or to stabilize.
The statement that no infinite ascending (resp. descending)
chain exists is generically known as the
ascending chain condition
(ACC) for the relevant ordering (resp. descending chain condition, DCC).
The non-commutative case is under-investigated because
the Lasker-Noether theorem (discussed in the next section)
doesn't apply to it. Emmy Noether herself gave a non-commutative example of
a right-Noetherian ring where some ideals don't have a
primary decomposition.
Equivalently, Noetherian rings can be defined as:
Rings in which every set of ideals has a maximal element.
(with respect to set-inclusion only; it needn't be a propermaximal ideal).
Finitely-generated rings. (In the commutative case, at least.)
Trivially, all finite rings are Noetherian.
Not all Bézout rings are.
Another example of a (commutative) Noetherian ring is the ring of integers
.
Indeed, all its ideals are of the form
n
(the principal ideal generated by n).
Thus, any ascending chain of ideals corresponds to a decreasing sequence of divisors,
which is necessarily finite, so the ACC holds.
More generally, any principal ring (or PID) is Noetherian.
In particular, every field is a Noetherian ring.
Hilbert's Basis Theorem (Hilbert, 1888) :
Univariate polynomials over a
Noetherian ring form a Noetherian ring.
Corollary (by induction) :
So do multivariate polynomials.
This theorem predates the formal introduction of Noetherian rings
by about thirty years!
David Hilbert (1862-1943)
gave a nonconstructive proof (1888) in the special case of univariate
polynomials over a field (fields are Noetherian) which states the
amazing fact that the ring of polynomials over a field is finitely generated.
I am led to believe that Hilbert used only Noetherian concepts in his proof,
so the modernized theorem remained recognizable and retained its original name
(although the corollary is trivial only when the result is stated in Noetherian terms).
The theorem is true also in the noncommutative case.
Let's give a proof which doesn't depend on commutativity:
Let A be a Noetherian ring (using one of the
four possible definitions presented above).
A[X] is the ring formed by all
univariate polynomials with coefficients in A.
Consider any ascending chain of ideals in A[X] :
I0 Í
I1 Í
I2 Í
I3 Í ...
In Í
In+1 ...
To prove the theorem, we must establish that this chain cannot be infinite
(i.e., it must stabilize after some large enough index).
Well, we may define an ideal Jij in A
as the set of all the coefficients of Xj in the
polynomials of Ii with
degree i or less.
It's not difficult to prove that every J is an ideal (one-sided or
double-sided) in the same sense as every I is.
(That exercise is left to the reader.)
Furthermore, it's clear that, if i ≤ i' and j ≤ j' then
Ji,j Í Ji',j'
In spite of its modern terminology, the above proof is probably quite close to
the nonconstructive proof originally proposed by Hilbert,
who was motivated by the abstract generalization of results in the
theory of invariants
which had been painstakingly derived by
Paul Gordan (1837-1912)
of Clebsch-Gordan_coefficients fame.
Legend has it that Gordan paid a strange compliment to this beautiful proof:
(2019-01-21) Artinian rings are Noetherian (but the converse isn't true).
Rings which don't contain any infinite descending chains of ideals.
For rings, surprisingly, the Artinian descending chain condition (DCC)
implies the Noetherianascending chain condition (ACC)
but turns out to be strictly stronger.
The following statement clarifies the situation and makes the independent study of Artinian rings all but superfluous:
An Artinian rings is a Noetherian ring where every non-invertible element is nilpotent.
Artinian rings may also be characterized as Noetherian rings whose
Krull dimensions are not positive.
The importance of Artinian rings stems from the fact that it's the natural concept to use in
the Artin-Wedderburn classification theorem.
(2019-01-18) Lasker-Noether Theorem (Lasker 1905, Noether 1921)
Every commutativeNoetherian ring is a Lasker ring (Laskerian ring).
By definition, a Lasker ring is a commutative ring in which any
ideal has a primary decomposition
(which is to say that it's the intersection of finitely many primary ideals).
In other words, a generalization of the fundamental theorem of arithmetic
holds in such rings. In 1905, Emanuel Lasker considered only
polynomials and convergent power series over a field.
He proved that the rings they form have that property. In 1921,
Emmy Noether generalized the theorem to all commutative Noetherian rings.
I don't know any non-Noetherian Lasker ring, but that's not ruled out.
Lasker didn't point out the unicity of the decomposition, which was established in 1915 by
Francis Macaulay (1862-1937).
The algorithm commonly used to perform actual primary decompositions
is due to Grete Hermann (1901-1984).
It was part of the doctoral work she completed under Emmy Noether
at Göttingen in 1926.
"Zur Theorie der Module und Ideale" by Emanuel Lasker, Math. Ann. , 60 (1905) pp. 19-116
"Idealtheorie in Ringbereiche" by Emmy Noether, Math. Ann. , 83 (1921) pp. 24-66
Emanuel Lasker (1868-1941,
Ph.D. 1902) was a talented German
mathematician,
mostly remembered as the longest-reigning Chessworld champion of the modern era.
He held the title officially for 27 years (1894-1921) including two periods
(1898-1906 & 1911-1920) when he wasn't challenged, either because of world events
(WWI) or because nobody could
raise the money to do so. Following in the footsteps of his older brother Berthold,
Emanuel Lasker depended on game-playing for financial support from an early age
(mostly chess but also card games; he was a world-class bridge player).
Lasker was one of the strongest Go players
outside of Asia. He had been introduced to Go by his close friend and distant relative
(third cousin twice removed)
Edward Lasker (1885-1981)
the amateur chess champion who co-founded the
American Go Association (AGA) in 1935.
Lasker tried to retire from competitive chess and exhibitions in the late 1920s
but had to return to it once everything he owned was confiscated by the Nazi regime.
He was fairly successful in spite of his age but his health degraded and he died in dire straits in the US.
Lasker's 1905 paper on ring theory marks the high-point of his mathematical
career and represents his only lasting mathematical contribution.
It's David Hilbert (1862-1943)
who had advised him to undertake doctoral studies (1900) at
Erlangen under
Max Noether (1844-1921)
the father of Emmy Noether (1882-1935).
Max Noether was one of the early leaders in algebraic geometry,
at the heart of which Lasker's theorem can be construed to be
(with or without the generalization due to Emmy Noether).
Especially if we recast it in the following form: Any
algebraic variety is the union of
finitely many irreducible components.
Lasker became a close friend of Albert Einstein (1879-1955)
who esteemed him highly. Einstein lamented that Lasker's great talents had been wasted on Chess.
(2019-01-22) GCD Domains
In them, two elements always have a greatest common divisor (GCD).
As the name implies, a GCD domain is a commutative ring.
(2019-01-22) Unique Factorization Domains (UFD)
GCD domains with no infinite ascending chains of principal ideals.
Bourbaki called such a ring a factorial ring,
which is now the only term used in French (namely, anneau factoriel ).
In a ring, two related concept (irreducible and prime)
can be defined which coincide only in the case of a factorial rings (UFD).
In such a structure, a straight counterpart of the
fundamental theorem of arithmetic hold,
which justifies the term of unique factorization domain.
An irreducible element q is an element which only has
trivial factorizations (i.e., it can only be written as a product
of two factors if at least one of those is a unit).
A prime element p is an element which can't divide
a product of two factors unless is divides one of them.
The ring
[Ö5]
(pronounced adjoined
root 5) isn't a UFD
because it contains irreducible elements which aren't prime.
For example, the integer 2 divides the following product
without dividing either factor:
4 = (-1 + Ö5) (1 + Ö5)
In 1847, the failure to distinguish between those two concepts famously led the French physicist
Gabriel Lamé
(1795-1870; X1814)
to propose an erroneous proof of Fermat's last theorem.
A few years earlier (1843) Eduard Kummer (1810-1895)
had already identified this trap, which would become one of the key motivations for
the development of ring theory.
(2006-04-05) GR(q,r) :
Galois ring of characteristic q = pm and rank r
The modulo-q polynomials modulo
an irreducible polynomial modulo p.
Let q be a power of a prime p. Let f
be some monicpolynomial
modulo q, of degree r, which is irreducible modulo p
(i.e., f (x) mod p never vanishes).
The Galois ring of characteristic q and rank r is
defined as:
GR(q,r) =
(/q)[x] / f (x)
Finite noncommutative rings can be described as algebra over Galois rings.
(2014-12-06) Hilbert rings = Jacobson rings
In those, every prime ideal is
an intersection of maximal ideals.
In particular, an Hilbert subring ring is the intersection of all maximal ideals containing it.
They were called Hilbert rings by Serre
and/or Bourbaki because of the
connection with Hilbert's Nullstellensatz. Outside the Bourbaki
sphere of influence, the attribution fo Jacobson seems to be preferred.
The qualifier maximal is sometimes replaced by primitive
(2018-01-25) Local rings (Krull, 1938)
Local algebra is the study of commutative local rings
and their modules.
A local ring is a ring which has only onemaximal ideal.
This concept was introduced in 1938 by Wolfgang Krull (1899-1971)
who called them Stellenringe in German.
The English term local ring was coined by
Oscar Zariski (1899-1986).
(2020-04-30) Topological Ring
When the topology makes addition and multiplication
continuous.
When a ring A is a topological space,
its cartesian squareA×A is considered endowed with the
product topology (of Tychonoff).
The addition and multiplication are functions from A×A
to A. When both are continuous, A is said to be a topological ring.
(2020-05-09) Involutive Ring (*-ring)
What a ring is called when endowed with an involution which is
both an additive automorphism and a multiplicative anti-automorphism.
By definition, an involution is just a
bijection equal to its own inverse.
In the context of *-rings, the image by the aforementioned involution
of an element x is denoted x* (pronounced x-star)
and is called the adjoint of x (or its conjugate).
It's now customary to call the involution the
function which maps an element to its adjoint
(I still call it conjugation at times).
In elementary texts, the word transconjuguate is also used
as a reminder that the adjoint of a matrix
of complex numbers is
the transpose of the matrix formed by the complex conjugates of its elements.
There's little need to do transposition without conjuguation
(or vice-versa). It's a bad idea to use a postfixed star to denote conjugation
without transposition, which would call for another postfix symbol to denote the adjoint
(many physicists use a dagger). I find it convenient to use the star
to denote all adjoints (including complex conjugates).
Especially if hypercomplex numbers
are involved.
That's to say that the following axioms are postulated: