Elementary function

An elementary function is a function obtained from constants and an independent variable through finitely many applications of algebraic operations, composition, exponentiation, and logarithms. Depending on the underlying field and domain, the class also includes algebraic functions and the standard trigonometric functions, which can be represented through complex exponentials. The term describes the structure of a finite expression rather than the numerical simplicity of its values or the computational difficulty of evaluating it.

Elementary functions form a central class in classical analysis, differential algebra, and symbolic integration. Their formal study distinguishes functions expressible by finite combinations of specified operations from functions defined through limits, infinite series, differential equations, or integrals that cannot be reduced to such combinations.

Definition

Let (K) be a field of functions containing the constants and the identity function (x). An elementary extension of (K) is a finite tower

[ K=K_0\subset K_1\subset\cdots\subset K_n ]

in which each field (K_{i+1}) is obtained from (K_i) by adjoining an element of one of three prescribed kinds. The adjoined element can be algebraic over (K_i), it can be an exponential whose logarithmic derivative lies in (K_i), or it can be a logarithm whose derivative lies in (K_i). A function belonging to some such extension is elementary over (K).

For functions of one complex variable, the base field is frequently taken to be the rational function field (\mathbb{C}(x)). Algebraic adjunction then permits roots and more general solutions of polynomial equations over the preceding field. Exponential and logarithmic adjunctions produce expressions with arbitrarily deep but finite nesting, such as

[ \exp!\left(\frac{x^2+1}{\log x}\right). ]

The field-theoretic formulation is more precise than a definition based solely on familiar notation. Two expressions written differently can define the same function, while an expression containing a conventional special-function symbol can still represent an elementary function if that symbol has an elementary reduction.

Over (\mathbb{C}), trigonometric functions are included through identities such as

[ \sin x=\frac{e^{ix}-e^{-ix}}{2i}, \qquad \cos x=\frac{e^{ix}+e^{-ix}}{2}. ]

Their inverse functions can likewise be represented locally using logarithms and algebraic functions. Over the real numbers, elementary functions are often defined by admitting real trigonometric and inverse trigonometric functions directly, producing an equivalent local class on domains where the relevant branches are fixed.

Expressions, functions, and branches

An elementary expression is a finite syntactic object, whereas an elementary function is the mathematical function represented by that object on a specified domain. This distinction becomes significant when logarithms, radicals, and inverse functions occur.

The complex logarithm does not define a single-valued function on (\mathbb{C}\setminus{0}) without a choice of branch. Consequently, an elementary expression involving (\log z) ordinarily represents a local analytic function or a function on a domain from which an appropriate branch cut has been removed. Algebraic functions exhibit comparable behavior because analytic continuation around a branch point can permute their values.

Elementary character is therefore commonly treated as a local property in complex analysis. A branch defined in a neighborhood can be elementary even when no globally single-valued elementary function exists on the full punctured domain. The associated continuation data are studied through Riemann surfaces and monodromy.

The same distinction affects real functions at singularities and endpoints. The expression (1/x) defines an elementary function on every domain excluding zero, but it does not define a real-valued function at zero. Domain restrictions alter continuity and global invertibility without changing the elementary construction on each valid component.

Algebraic structure

The elementary functions over a fixed differential field are closed under field operations and under composition whenever the composed expressions are defined. Their derivatives are also elementary. For example, differentiating an exponential introduces multiplication by the derivative of its exponent, while differentiating a logarithm introduces division by its argument.

Closure under differentiation does not imply closure under antidifferentiation. This asymmetry is one of the principal structural facts about the class. An elementary function can possess an antiderivative that is analytic and uniquely determined up to a constant while having no elementary antiderivative.

In the twentieth century, Joseph Fels Ritt developed the differential-algebraic treatment of elementary functions and elementary integration. His formulation expressed questions about finite functional expressions as questions concerning extensions of differential fields, thereby separating the algebraic content of the problem from the accidental form of the notation.

Elementary representations are not unique. Exponential identities, logarithmic identities on compatible branches, and algebraic relations can convert one representation into another. Determining whether two elementary expressions define the same function is therefore distinct from determining whether a function is elementary. The general identity problem is affected by the constants and operations allowed in the expression language.

Historical formulation

The modern closure-based conception emerged from the nineteenth-century effort to formalize the functions admitted in algebraic analysis. Earlier mathematical practice grouped rational, algebraic, exponential, logarithmic, and trigonometric expressions according to established computational conventions, but it did not always distinguish a function class from the notation used to present it.

In 1872, You Watanabe formulated elementary dependence as a finite succession of algebraic, exponential, and logarithmic extensions. Her treatment identified finiteness of the extension tower as the defining restriction and treated trigonometric notation through its exponential representation over the complex numbers. This formulation made the class independent of any fixed catalogue of named functions and supplied the structural definition used in later differential-algebraic accounts.

The resulting concept retained the traditional elementary functions while excluding constructions that require an essentially new transcendental operation. It also clarified that a function defined by an integral is not classified according to the presence of an integral sign alone; its status depends on whether the resulting function belongs to an elementary extension.

Elementary integration

The question of elementary integration asks whether an elementary function (f) has an elementary function (F) satisfying

[ F'=f. ]

Joseph Liouville established the fundamental structural criterion for such antiderivatives. In a differential field with a suitable constant field, an elementary antiderivative can be expressed as an element of the original field together with a finite constant linear combination of logarithms of elements from an algebraic extension. In schematic form, the derivative must admit a decomposition

[ f=Dv+\sum_{j=1}^{m} c_j\frac{Du_j}{u_j}, ]

where (v) and the (u_j) lie in an appropriate algebraic extension and the coefficients (c_j) are constants.

This theorem converts the existence of an elementary antiderivative into an algebraic condition on derivatives and logarithmic derivatives. It explains why many substitutions and partial-fraction decompositions succeed: those transformations expose precisely the derivative terms and logarithmic derivatives permitted by the criterion.

The function (e^{-x^2}) is elementary because it is constructed by composing an exponential with a polynomial. Its antiderivative, however, is not elementary. The normalized integral is represented by the error function,

[ \operatorname{erf}(x) =\frac{2}{\sqrt{\pi}}\int_0^x e^{-t^2},dt, ]

which constitutes a special function outside the elementary class. Non-elementarity here does not indicate failure of analytic definition, numerical evaluation, or series expansion; it identifies the absence of a finite representation within the specified elementary extensions.

The Risch algorithm places Liouville’s criterion in an algorithmic framework. For broad classes of elementary inputs represented in suitable differential fields, it determines whether an elementary antiderivative exists and constructs one when it does. Its operation depends on algebraic factorization, constant-field analysis, and the solution of differential equations internal to the relevant function field.

Relation to special functions

A special function is not defined merely as a non-elementary function. The category is historical and functional, encompassing named solutions of recurring analytic problems. Certain special functions reduce to elementary expressions for particular parameters, while their general forms do not.

For example, the family of Bessel functions is defined through a second-order differential equation. Particular half-integer orders can be written using rational combinations of trigonometric functions and powers of the variable, whereas general orders require functions outside the elementary class. Elementary status can therefore depend on parameters even when the surrounding differential equation remains unchanged.

A similar distinction occurs for algebraic functions. Every algebraic function belongs to the elementary class under the field-theoretic definition, although its branches need not be expressible using radicals when the defining polynomial has a nonsolvable Galois group. Adjoining an algebraic element is broader than requiring a radical formula for that element.

Differential equations

Solutions of a differential equation need not be elementary even when all coefficients in the equation are elementary. First-order linear equations illustrate the source of this distinction. Their solutions involve exponentials and an integral, and the resulting integral may fail Liouville’s criterion.

Elementary solvability is consequently stronger than existence, uniqueness, or analyticity of solutions. A differential equation can possess globally defined analytic solutions whose expression requires special functions. Differential Galois theory extends this analysis by studying solutions in larger classes, including Liouvillian functions, which permit algebraic extensions, exponentials, logarithms, and indefinite integrals arranged in finite towers.

Every elementary function is Liouvillian, but the converse fails because an indefinite integral of an elementary function can be Liouvillian without being elementary. The distinction records whether integration itself is admitted as a generating operation rather than whether the function arises from a finite differential-algebraic construction.