special function, any of a class of mathematical functions that arise in the solution of various classical problems of physics. These problems generally involve the flow of electromagnetic, acoustic, or thermal energy. Different scientists might not completely agree on which functions are to be included among the special functions, although there would certainly be very substantial overlap.
At first glance, the physical problems mentioned above seem to be very limited in scope. From a mathematical point of view, however, different representations have to be sought, depending on the configuration of the physical system for which these problems are to be solved. For example, in studying propagation of heat in a metallic bar, one could consider a bar with a rectangular cross section, a round cross section, an elliptical cross section, or even more-complicated cross sections; the bar might be straight or curved. Every one of these situations, while dealing with the same type of physical problem, leads to somewhat different mathematical equations.
The equations to be solved are partial differential equations. To apprehend how these equations come about, one can consider a straight rod along which there is a uniform flow of heat. Let u(x, t) denote the temperature of the rod at time t and location x, and let q(x, t) denote the rate of heat flow. The expression ∂q/∂x denotes the rate at which the rate of heat flow changes per unit length and therefore measures the rate at which heat is accumulating at a given point x at time t. If heat is accumulating, the temperature at that point is rising, and the rate is denoted by ∂u/∂t. The principle of conservation of energy leads to ∂q/∂x = k(∂u/∂t), where k is the specific heat of the rod. This means that the rate at which heat is accumulating at a point is proportional to the rate at which the temperature is increasing. A second relationship between q and u is obtained from Newton’s law of cooling, which states that q = K(∂u/∂x). The latter is a mathematical way of asserting that the steeper the temperature gradient (the rate of change of temperature per unit length), the higher the rate of heat flow. Elimination of q between these equations leads to ∂2u/∂x2 = (k/K)(∂u/∂t), the partial differential equation for one-dimensional heat flow.
The partial differential equation for heat flow in three dimensions takes the form ∂2u/∂x2 + ∂2u/∂y2 + ∂2u/∂z2 = (k/K)(∂u/∂t); the latter equation is often written ∇2u = (k/K)(∂u/∂t), where the symbol ∇, called del or nabla, is known as the Laplace operator. ∇ also enters the partial differential equation dealing with wave-propagation problems, which has the form ∇2u = (1/c2)(∂2u/∂t2), where c is the speed at which the wave propagates.
Partial differential equations are harder to solve than ordinary differential equations, but the partial differential equations associated with wave propagation and heat flow can be reduced to a system of ordinary differential equations through a process known as separation of variables. These ordinary differential equations depend on the choice of coordinate system, which in turn is influenced by the physical configuration of the problem. The solutions of these ordinary differential equations form the majority of the special functions of mathematical physics.
For example, in solving the equations of heat flow or wave propagation in cylindrical coordinates, the method of separation of variables leads to Bessel’s differential equation, a solution of which is the Bessel function, denoted by Jn(x).
Among the many other special functions that satisfy second-order differential equations are the spherical harmonics (of which the Legendre polynomials are a special case), the Tchebychev polynomials, the Hermite polynomials, the Jacobi polynomials, the Laguerre polynomials, the Whittaker functions, and the parabolic cylinder functions. As with the Bessel functions, one can study their infinite series, recursion formulas, generating functions, asymptotic series, integral representations, and other properties. Attempts have been made to unify this rich topic, but not one has been completely successful. In spite of the many similarities among these functions, each has some unique properties that must be studied separately. But some relationships can be developed by introducing yet another special function, the hypergeometric function, which satisfies the differential equation z(1 − z) d2y/dx2 + [c − (a + b + 1)z] dy/dx − aby = 0. Some of the special functions can be expressed in terms of the hypergeometric function.
While it is true, both historically and practically, that the special functions and their applications arise primarily in mathematical physics, they do have many other uses in both pure and applied mathematics. Bessel functions are useful in solving certain types of random-walk problems. They also find application in the theory of numbers. The hypergeometric functions are useful in constructing so-called conformal mappings of polygonal regions whose sides are circular arcs.