A transcendental number is a complex number that is not a root of any nonzero polynomial with rational coefficients. Equivalently, it satisfies no polynomial equation with integer coefficients, except the equation given by the zero polynomial. Numbers that do satisfy such an equation are called algebraic numbers. The best-known transcendental numbers are Euler’s number and the circle constant . Although almost every real number is transcendental in a precise mathematical sense, proving the transcendence of a particular explicitly defined number can be exceptionally difficult.
Definition and distinction from irrationality
A complex number is transcendental if there are no coefficients in the integers, not all zero, such that
Using rational rather than integer coefficients gives the same definition: multiplying through by a common denominator converts any polynomial with rational coefficients into one with integer coefficients.
Every rational number is algebraic. For example, , where and are integers and , is a root of . Consequently, every real transcendental number is an irrational number. The converse is false:
- is irrational but algebraic because it satisfies .
- The golden ratio is algebraic because it satisfies .
- and are both irrational and transcendental.
For nonreal complex numbers, the classification into algebraic and transcendental numbers still applies, but “irrational” is normally reserved for real numbers.
An infinite, nonrepeating decimal expansion establishes irrationality, not transcendence. Algebraic irrational numbers also have nonterminating, nonperiodic decimal expansions. Transcendence is a statement about polynomial equations, not simply about the appearance of digits.
Historical development
The distinction between algebraic and transcendental quantities developed as mathematicians investigated which numbers could be characterized by algebraic equations. In 1844, Joseph Liouville established the existence of transcendental numbers by showing that algebraic irrational numbers cannot be approximated too closely by rational numbers, and constructing numbers that violate that restriction.
Charles Hermite proved the transcendence of in 1873. In 1874, Georg Cantor proved that the real algebraic numbers form a countable set while the real numbers do not. This established, without identifying individual examples, that transcendental numbers vastly outnumber algebraic numbers.
Ferdinand von Lindemann proved the transcendence of in 1882. His result settled the ancient problem of squaring the circle with straightedge and compass.
A major twentieth-century advance came in 1934, when Aleksandr Gelfond and Theodor Schneider independently proved the result now called the Gelfond–Schneider theorem. It resolved the transcendence question posed in the seventh of David Hilbert’s problems.
Explicit examples and methods of proof
Liouville numbers
Liouville’s method begins with a restriction on rational approximation. If is a real algebraic number of degree , then there is a constant , depending on , such that
for every integer and positive integer .
The degree is the degree of the polynomial of least degree over the rationals having as a root. The bound follows by evaluating that polynomial at , obtaining a lower bound for its nonzero value, and relating that value to the distance from to .
A Liouville number is a real irrational number such that, for every positive integer , there are integers and , with , satisfying
Such a number admits rational approximations that are too accurate for any algebraic irrational number of fixed degree. Every Liouville number is therefore transcendental.
The standard example is Liouville’s constant,
where denotes the factorial of . Its decimal digits are in positions , and elsewhere.
Truncating the series after the -th term gives a rational number with denominator . The remaining tail satisfies
As increases, these approximations exceed the accuracy permitted by Liouville’s bound for every fixed algebraic degree.
Liouville’s criterion is sufficient but not necessary for transcendence. Many transcendental numbers, including and , are not Liouville numbers.
Exponential values and the Lindemann–Weierstrass theorem
The Lindemann–Weierstrass theorem states that if are distinct algebraic numbers, then
are linearly independent over the algebraic numbers. Thus, a relation
with algebraic coefficients forces every to be zero.
A central consequence is that is transcendental whenever is a nonzero algebraic number. Taking gives the transcendence of .
The transcendence of follows from this consequence and Euler’s formula. If were algebraic, then would be a nonzero algebraic number, so would be transcendental. But
which is algebraic.
These results connect transcendence theory with the exponential function and complex analysis.
The Gelfond–Schneider theorem
The Gelfond–Schneider theorem states that if is algebraic, , and is algebraic but irrational, then every value of is transcendental.
For positive real , the usual real value is unambiguous. In particular,
is transcendental.
The theorem also proves that is transcendental. In complex exponentiation, one value of is
Here both and are algebraic, while is not rational.
How common are transcendental numbers?
The algebraic numbers form a countable set. There are only countably many polynomials with integer coefficients, and every nonzero polynomial has finitely many complex roots. The union of all these finite root sets is therefore countable.
By contrast, the real numbers are uncountable. Removing the algebraic numbers leaves a set of transcendental numbers with the same cardinality as the entire real line.
There is also a measure-theoretic statement. Every countable subset of the real line has Lebesgue measure zero, so the algebraic numbers have measure zero. Consequently, almost every real number is transcendental. Equivalently, under any probability distribution having a density with respect to Lebesgue measure, a real-valued outcome is transcendental with probability one.
These statements do not supply an easy test for a specified number. A property can hold almost everywhere while remaining difficult to establish for particular constants.
Both algebraic and transcendental real numbers are dense in the real line: every nonempty open interval contains numbers of both kinds. Thus, abundance in measure does not mean that transcendental numbers are separated from algebraic ones by gaps.
Arithmetic properties and algebraic independence
The algebraic numbers form a field: sums, differences, products, and quotients with nonzero denominator remain algebraic. Transcendental numbers do not have corresponding closure properties.
If is transcendental and is algebraic, then:
- is transcendental;
- is transcendental when ;
- is transcendental.
Otherwise, field operations would express in terms of algebraic numbers and make it algebraic.
For two transcendental numbers, however, a sum or product may be algebraic. For example,
although every displayed operand is transcendental.
For every positive integer , is transcendental when is transcendental. More generally, if is a nonconstant polynomial with algebraic coefficients, then is transcendental.
A stronger concept is algebraic independence. Numbers are algebraically independent over the rationals if no nonzero polynomial in variables with rational coefficients satisfies
Transcendence is precisely algebraic independence for a single number. Knowing that two numbers are individually transcendental does not establish their algebraic independence: and , for example, satisfy the polynomial relation .
Geometry and computation
The transcendence of rules out squaring the circle by classical straightedge-and-compass construction. Every length constructible from a unit segment by these operations is algebraic. A square with the same area as a unit circle would have side length , which is transcendental because its square is transcendental.
Transcendence does not imply that a number is impossible to compute. Both and have algorithms that approximate them to any prescribed accuracy. Conversely, because algorithms form a countable set while transcendental real numbers are uncountable, most transcendental numbers are not computable.
Finite numerical data alone cannot establish transcendence. Any finite decimal approximation is compatible with both algebraic and transcendental values. Numerical experiments may suggest relations or guide proofs, but transcendence requires an argument ruling out every nonzero polynomial relation over the rationals.
Open questions
Several elementary-looking constants still resist classification:
- It is unknown whether is transcendental, or even irrational.
- It is unknown whether is transcendental, or even irrational.
- It is unknown whether and are algebraically independent.
- It is unknown whether the Euler–Mascheroni constant is irrational, and therefore also whether it is transcendental.
- The transcendence of , a value of the Riemann zeta function, is unknown, although its irrationality is established.
At least one of and must be transcendental. If both were algebraic, then and would be roots of
and hence would be algebraic over the algebraic numbers, contradicting their transcendence.
These unresolved cases distinguish proving that numbers individually satisfy no polynomial equation from controlling polynomial relations among several constants.