Orthogonal functions are a set of functions that satisfy the mathematical condition of orthogonality, which means that the integral of the product of any two distinct functions in the set over a specified interval is zero. This concept is fundamental in various areas of mathematics and engineering, particularly in traitement du signal, analyse de Fourier, and analyse fonctionnelle.
En termes plus techniques, deux fonctions, f(x) et g(x), sont dites orthogonales sur un intervalle [a, b] si :
∫ab f(x) g(x) dx = 0
This relationship indicates that the functions do not correlate with each other, which is a desirable property in many applications. For instance, orthogonal functions can be used to construct Série de Fourier, where sine and cosine functions serve as orthogonal bases for representing periodic functions.
Orthogonality is not limited to just two functions; a complete set of orthogonal functions can be used to expand other functions in a way that minimizes redundancy and maximizes efficiency in representation. Examples of such sets include the Legendre polynomials and the Hermite functions. In the context of machine learning, orthogonal functions are also related to various algorithms, including those used for techniques de réduction de dimension and feature extraction, where they help in ensuring that the features are independent and non-redundant.