MIT 6.

972 Algebraic techniques and semidefinite optimization February 16, 2006

Lecture 4
Lecturer: Pablo A. Parrilo Scribe: Pablo A. Parrilo

In this lecture we will review some basic elements of abstract algebra. We also introduce and begin
studying the main objects of our considerations, multivariate polynomials.

1 Review: groups, rings, fields

We present here standard background material on abstract algebra. Most of the definitions are from
[Lan71, CLO97, DF91, BCR98].

Definition 1 A group consists of a set G and a binary operation “·” defined on G, for which the
following conditions are satisfied:
1. Associative: (a · b) · c = a · (b · c), for all a, b, c ∈ G.
2. Identity: There exist 1 ∈ G such that a · 1 = 1 · a = a, for all a ∈ G.
3. Inverse: Given a ∈ G, there exists b ∈ G such that a · b = b · a = 1.

For example, the integers Z form a group under addition, but not under multiplication. Another example
is the set GL(n, R) of real nonsingular n × n matrices, under matrix multiplication.
If we drop the condition on the existence of an inverse, we obtain a monoid. Note that a monoid
always has at least one element, the identity. As an example, given a set S, then the set of all strings of
elements of S is a monoid, where the monoid operation is string concatenation and the identity is the
empty string λ. Another example is given by N0 , with the operation being addition (in this case, the
identity is the zero). Monoids are also known as semigroups with identity.
In a group we only have one binary operation (“multiplication”). We will introduce another operation
(“addition”), and study the structure that results from their interaction.

Definition 2 A commutative ring (with identity) consists of a set k and two binary operations “·” and
“+”, defined on k, for which the following conditions are satisfied:
1. Associative: (a + b) + c = a + (b + c) and (a · b) · c = a · (b · c), for all a, b, c ∈ k.

2. Commutative: a + b = b + a and a · b = b · a, for all a, b ∈ k.

3. Distributive: a · (b + c) = a · b + a · c, for all a, b, c ∈ k.
4. Identities: There exist 0, 1 ∈ k such that a + 0 = a · 1 = a, for all a ∈ k.

5. Additive inverse: Given a ∈ k, there exists b ∈ k such that a + b = 0.

A simple example of a ring are the integers Z under the usual operations. After formally introducing
polynomials, we will see a few more examples of rings.
If we add a requirement for the existence of multiplicative inverses, we obtain fields.
Definition 3 A field consists of a set k and two binary operations “·” and “+”, defined on k, for which
the following conditions are satisfied:
1. Associative: (a + b) + c = a + (b + c) and (a · b) · c = a · (b · c), for all a, b, c ∈ k.
2. Commutative: a + b = b + a and a · b = b · a, for all a, b ∈ k.
3. Distributive: a · (b + c) = a · b + a · c, for all a, b, c ∈ k.

4­1
 4. Identities: There exist 0, 1 ∈ k, where 0 �= 1, such that a + 0 = a · 1 = a, for all a ∈ k.
5. Additive inverse: Given a ∈ k, there exists b ∈ k such that a + b = 0.
6. Multiplicative inverse: Given a ∈ k, a �= 0, there exists c ∈ k such that a · c = 1.
Any field is obviously a commutative ring. Some commonly used fields are the rationals Q, the reals
R and the complex numbers C. There are also Galois or finite fields (the set k has a finite number of
elements), such as Zp , the set of integers modulo p, where p is a prime. Another important field is given
by k(x1 , . . . , xn ), the set of rational functions with coefficients in the field k, with the natural operations.

2 Polynomials and ideals

Consider a given field k, and let x1 , . . . , xn be indeterminates. We can then define polynomials.

Definition 4 A polynomial f in x1 , . . . , xn with coefficients in a field k is a finite linear combination

of monomials: � �
f= cα xα = cα xα αn
1 . . . xn ,
1
cα ∈ k, (1)
α α

where the sum is over a finite number of n­tuples α = (α1 , . . . , αn ), αi ∈ N0 . The set of all polynomials
in x1 , . . . , xn with coefficients in k is denoted k[x1 , . . . , xn ].

It follows from the previous definitions that k[x1 , . . . , xn ], i.e., the set of polynomials in n variables with
coefficients in k, is a commutative ring with identity. We also notice that it is possible (and sometimes,
convenient) to define polynomials where the coefficients belong to a ring with identity, not necessarily
to a field.
�
Definition 5 A form is a polynomial where all the monomials have the same degree d := i αi . In this
case, the polynomial is homogeneous of degree d, since it satisfies f (λx1 , . . . , λxn ) = λd f (x1 , . . . , xn ).

� 0, b �= 0 ⇒ a· b �= 0.
A commutative ring is called an integral domain if it has no zero divisors, i.e. a =
Every field is also an integral domain (why?). Two examples of rings that are not integral domains are
the set of matrices Rn×n , and the set of integers modulo n, when n is a composite number (with the
usual operations). If k is an integral domain, then so is k[x1 , . . . , xn ].

Remark 6 Another important example of a ring (in this case, non­commutative) appears in systems
and control theory, through the ring M(s) of stable proper rational functions. This is the set of matri­
ces (of fixed dimension) whose entries are rational functions of s (i.e., in the field C(s)), are bounded
at infinity, and have all poles in the strict left­half plane. In this algebraic setting (usually called “co­
prime factorization approach”), the question of finding a stabilizing controller is exactly equivalent to the
solvability of a Diophantine equation ax + by = 1.

2.1 Algebraically closed and formally real fields

A very important property of a univariate polynomial p is the existence of a root, i.e., an element x0 for
which p(x0 ) = 0. Depending on the solvability of these equations, we can characterize a particular nice
class of fields.

Definition 7 A field k is algebraically closed if every nonconstant polynomial in k[x] has a root in k.

4­2
If a field is algebraically closed, then it has an infinite number of elements (why?). What can we say about
the most usual fields, C and R? The Fundamental Theorem of Algebra (“every univariate polynomial
has at least one complex root”) shows that C is an algebraically closed field.
However, this is clearly not the case of R, since for instance the polynomial x2 + 1 does not have any
real root. The lack of algebraic closure of R is one of the main sources of complications when dealing
with systems of polynomial equations and inequalities. To deal with the case when the base field is not
algebraically closed, the Artin­Schreier theory of formally real fields was introduced.
The starting point is one of the intrinsic properties of R:
n
�
x2i = 0 =⇒ x1 = . . . = xn = 0. (2)
i=1

A field will be called formally real if it satisfies the above condition (clearly, R and Q are formally real,
but C is not). As we can see from the definition, the theory of formally real fields has very strong
connections with sums of squares, a notion that will reappear in several forms later in the course. For
example, an alternative (but equivalent) statement of (2) is to say that a field is formally real if and
only if the element −1 is not a sum of squares.
A related important notion is that of an ordered field:
Definition 8 A field k is said to be ordered if a relation > is defined on k, that satisfies
1. If a, b ∈ k, then either a > b or a = b or b > a.
2. If a > b, c ∈ k, c > 0 then ac > bc.
3. If a > b, c ∈ k, then a + c > b + c.
A crucial result relating these two notions is the following:
Lemma 9 A field can be ordered if and only if it is formally real.
For a field to be ordered (or equivalently, formally real), it necessarily must have an infinite number of
elements. This is somewhat unfortunate, since this rules out several modular methods for dealing with
real solutions to polynomial inequalities.

2.2 Ideals
We consider next ideals, which are subrings with an “absorbent” property:
Definition 10 Let R be a commutative ring. A subset I ⊂ R is an ideal if it satisfies:
1. 0 ∈ I.
2. If a, b ∈ I, then a + b ∈ I.
3. If a ∈ I and b ∈ R, then a · b ∈ I.
A simple example of an ideal is the set of even integers, considered as a subset of the integer ring Z.
Also, notice that if the ideal I contains the multiplicative identity 1, then I = R.
To introduce another important example of ideals, we need to define the concept of an algebraic
variety as the zero set of a set of polynomial equations:
Definition 11 Let k be a field, and let f1 , . . . , fs be polynomials in k[x1 , . . . , xn ]. Let the set V be

V(f1 , . . . , fs ) = {(a1 , . . . , an ) ∈ k n : fi (a1 , . . . , an ) = 0 ∀1 ≤ i ≤ s}.

We call V(f1 , . . . , fs ) the affine variety defined by f1 , . . . , fs .

4­3
Then, the set of polynomials that vanish in a given variety, i.e.,

I(V ) = {f ∈ k[x1 , . . . , xn ] : f (a1 , . . . , an ) = 0 ∀(a1 , . . . , an ) ∈ V },

is an ideal, called the ideal of V .

By Hilbert’s Basis Theorem [CLO97], k[x1 , . . . , xn ] is a Noetherian ring, i.e., every ideal I ⊂
�sset f1 , . . . , fs ∈ k[x1 , . . . , xn ]
k[x1 , . . . , xn ] is finitely generated. In other words, there always exists a finite
such that for every f ∈ I, we can find gi ∈ k[x1 , . . . , xn ] that verify f = i=1 gi fi .
We also define the radical of an ideal:
√
Definition 12 Let I ⊂ k[x1 , . . . , xn ] be an ideal. The radical of I, denoted I, is the set

{f | f k ∈ I for some integer k ≥ 1}.

√ √
It is clear that I ⊂ I, and it can be shown that I is also a polynomial ideal. A very important result,
that we will see later in some detail, is the following:
√
Theorem 13 (Hilbert’s Nullstellensatz) If I is a polynomial ideal, then I(V(I)) = I.

2.3 Associative algebras

Another important notion, that we will encounter at least twice later in the course, is that of an asso­
ciative algebra.
Definition 14 An associative algebra A over C is a vector space with a C­bilinear operation · : A×A →
A that satisfies
x · (y · z) = (x · y) · z, ∀x, y, z ∈ A.
In general, associative algebras do not need to be commutative (i.e., x · y = y · x). However, that is
an important special case, with many interesting properties. We list below several examples of finite
dimensional associative algebras.
• Full matrix algebra Cn×n , standard product.
• The subalgebra of square matrices with equal row and column sums.
• The n­dimensional algebra generated by a single n × n matrix.
• The group algebra: formal C­linear combination of group elements.
• Polynomial multiplication modulo a zero dimensional ideal.
• The Bose­Mesner algebra of an association scheme.
We will discuss the last three in more detail later in the course.

3 Questions about polynomials

There are many natural questions that we may want to answer about polynomials, even in the univariate
case. Among them, we mention:
• When does a univariate polynomial have only real roots?
• What conditions must it satisfy for all roots to be real?
• When does a polynomial satisfy p(x) ≥ 0 for all x?
We will answer many of these next week.

4­4
References
[BCR98] J. Bochnak, M. Coste, and M­F. Roy. Real Algebraic Geometry. Springer, 1998.
[CLO97] D. A. Cox, J. B. Little, and D. O’Shea. Ideals, varieties, and algorithms: an introduction to
computational algebraic geometry and commutative algebra. Springer, 1997.
[DF91] D. S. Dummit and R. M. Foote. Abstract algebra. Prentice Hall Inc., Englewood Cliffs, NJ,
1991.
[Lan71] S. Lang. Algebra. Addison­Wesley, 1971.

4­5