Logconvex

Logconvex functions are mathematical functions that are convex when considered on a logarithmic scale. A function f(x) is said to be logconvex if the function g(x) = log(f(x)) is convex. In other words, for all x1, x2 in the domain of f and for all λ in [0, 1], the inequality f(λx1 + (1-λ)x2) ≤ (f(x1))^λ (f(x2))^(1-λ) holds.

Logconvex functions have several important properties. Firstly, they are always non-negative, as the logarithm of a non-positive number is undefined. Secondly, they are closed under multiplication, meaning that the product of two logconvex functions is also logconvex. This property makes logconvex functions useful in various fields, such as optimization and probability theory.

In optimization, logconvex functions are often used because they preserve the convexity of the objective function under multiplication. This is particularly useful in geometric programming, where the objective function is a posynomial (a sum of monomials with positive coefficients) and the constraints are posynomial inequalities.

In probability theory, logconvex functions are used to model processes where the logarithm of the probability density function is convex. This is useful in Bayesian inference, where the posterior distribution is proportional to the product of the prior distribution and the likelihood function, both of which can be logconvex.

Logconvex functions have also been studied in the context of majorization theory, where they are used to characterize the relationship between different orderings of vectors. In this context, a function is said to be logconvex if it is convex with respect to the majorization ordering.

Despite their usefulness, logconvex functions are not as widely studied as other classes of convex functions, such as Jensen convex functions. This is likely due to the fact that logconvex functions are not closed under addition, which makes them less flexible in certain applications. However, their unique properties make them an important tool in certain areas of mathematics and its applications.