An algebraic number is a complex number that is a root of a nonzero polynomial with integer coefficients. Equivalently, it is a complex number for which
where the coefficients are integers, , and . Algebraic numbers include all rational numbers and many irrational and nonreal numbers. A complex number that is not algebraic is called a transcendental number.
Definition and examples
The definition is unchanged if the coefficients are allowed to be rational numbers: multiplying a polynomial with rational coefficients by a common denominator produces a polynomial with integer coefficients having the same roots.
Some basic examples are:
| Number | A polynomial it satisfies | Degree over |
|---|---|---|
| , where and | ||
Thus, being algebraic is not the same as being rational or real. In particular, an irrational number may be algebraic, as is, or transcendental, as is.
The requirement that the polynomial be nonzero is essential: the zero polynomial vanishes at every complex number and would make the definition meaningless.
More generally, an element of a field extension is algebraic over a field if it satisfies a nonzero polynomial with coefficients in . Without an explicitly stated base field, “algebraic number” means algebraic over , the field of rational numbers.
Minimal polynomial and degree
Every algebraic number has a unique minimal polynomial over . It is the monic polynomial of least positive degree satisfying . Here, monic means that the leading coefficient is .
The minimal polynomial is irreducible over : it cannot be written as a product of two positive-degree polynomials with rational coefficients. It also divides every polynomial in that vanishes at .
The degree of an algebraic number is the degree of its minimal polynomial:
Consequently:
- The algebraic numbers of degree are exactly the rational numbers.
- Algebraic numbers of degree are called quadratic algebraic numbers.
- A polynomial equation satisfied by a number need not have the smallest possible degree.
For example, satisfies , but its minimal polynomial is , so its degree is , not .
A minimal polynomial can alternatively be normalized to have integer coefficients with greatest common divisor and positive leading coefficient. In this convention it is called the primitive integer minimal polynomial; it need not be monic.
Arithmetic and number fields
The algebraic numbers are closed under addition, subtraction, multiplication, and division by a nonzero element. They therefore form a field, usually denoted
One explanation of this closure uses field extensions. If and are algebraic, the field generated by them has finite dimension as a vector space over . Every element of such a finite-dimensional extension is algebraic: sufficiently many powers
must be linearly dependent over , giving a polynomial equation for .
A number field is a finite extension of . For a single algebraic number of degree ,
and every element of has a unique expression
Arithmetic can therefore be performed exactly by manipulating rational coefficients and reducing powers using the minimal polynomial. For instance, in ,
Although every individual algebraic number generates a finite extension, itself has infinite degree over .
Conjugates, trace, and norm
The roots of the minimal polynomial of are its algebraic conjugates over . Since has characteristic zero, the roots of an irreducible polynomial over are distinct. An algebraic number of degree therefore has exactly conjugates, including itself.
For example:
- The conjugates of are and .
- The conjugates of are , , and , where .
Algebraic conjugacy is not the same as complex conjugation. A real algebraic number may have nonreal algebraic conjugates, and its algebraic conjugates may include several distinct real numbers.
The sum and product of the conjugates give the trace and norm from to . If
then
The symmetries among conjugates are a central subject of Galois theory.
Algebraic integers
An algebraic integer is a complex number satisfying a monic polynomial with integer coefficients. This is a stronger condition than being an algebraic number.
For example:
- , , and are algebraic integers.
- is algebraic but is not an algebraic integer.
An algebraic number is an algebraic integer exactly when its monic minimal polynomial over has integer coefficients. Among rational numbers, the algebraic integers are precisely the ordinary integers.
Algebraic integers are closed under addition, subtraction, and multiplication, but not generally under division. Within a number field , they form its ring of integers, denoted . These rings are fundamental to algebraic number theory, which studies arithmetic in finite extensions of . Factorization into elements can fail to be unique in these rings, motivating the study of ideals.
Every algebraic number becomes an algebraic integer after multiplication by a suitable nonzero ordinary integer.
Algebraic closure, size, and density
The field is algebraically closed: every nonconstant polynomial whose coefficients are algebraic numbers has a root in , and indeed splits completely there.
To see this, let be a complex root of such a polynomial. Its finitely many coefficients generate a finite extension of . Since is algebraic over , the extension is also finite, so is algebraic over . Thus is an algebraic closure of inside .
Despite containing all roots of all integer-coefficient polynomials, the algebraic numbers form a countable set. There are countably many such polynomials, and each nonzero polynomial has only finitely many roots. Since the real numbers are uncountable, most real numbers, in the sense of cardinality, are transcendental.
The algebraic numbers nevertheless form a dense set in , because they contain every number with rational and . The real algebraic numbers are similarly dense in .
Countability also implies that the real algebraic numbers have Lebesgue measure zero. Density and measure zero are compatible: every real interval contains infinitely many algebraic numbers, but the set occupies no positive length.
Radicals and geometric constructions
Any number obtained from rational numbers by finitely many arithmetic operations and extractions of integer roots is algebraic. The converse is false: some algebraic numbers cannot be expressed by radicals.
For example, the polynomial
is irreducible over and has Galois group . Its roots cannot be expressed by radicals over . This is a statement about exact symbolic expressions, not about the possibility of computing numerical approximations.
Straightedge-and-compass geometry imposes an even narrower restriction. A real constructible number must lie in a tower of quadratic extensions of . Its degree over must therefore be a power of , although that degree condition alone is not sufficient.
The number , having degree , is not constructible. This gives the algebraic obstruction to the classical problem of doubling the cube. The transcendence of rules out squaring the circle by straightedge and compass.
Historical development
The study of algebraic numbers grew from attempts to solve polynomial equations and understand irrational magnitudes. Formulas for cubic and quartic equations in the sixteenth century brought nonreal roots into systematic calculation.
During the nineteenth century, the work of Abel and Galois clarified the limitations of formulas by radicals. Kummer, Dedekind, and Kronecker developed arithmetic theories extending ideas about integers to more general algebraic numbers.
A complementary development was the proof that nonalgebraic numbers exist. Liouville established the existence of transcendental numbers in the 1840s. Hermite proved the transcendence of in 1873, and Lindemann proved the transcendence of in 1882. Cantor’s work on infinite sets showed that algebraic numbers are countable whereas real numbers are not.
Exact representation and computation
An algebraic number can be represented exactly by an integer polynomial together with information specifying which root is intended. For a real root, this can be a rational interval containing that root and no other root of the polynomial. For a nonreal root, an isolating region in the complex plane can serve the same purpose.
Such representations support exact equality testing, arithmetic, comparison of real algebraic numbers, and numerical approximation to arbitrarily high precision. They are useful in symbolic computation, computational geometry, and the exact study of polynomial systems.
A polynomial alone generally does not identify a particular algebraic number because it may have several roots. Conversely, a finite decimal approximation does not determine whether the represented value is algebraic or transcendental. Exact algebraic representation and approximate numerical representation therefore provide different kinds of information.