Solving linear functional inequalities with a computer
Computational methods for problems pertaining to difference and functional equations (39-08) Functional equations for real functions (39B22) Functional inequalities, including subadditivity, convexity, etc. (39B62) Inequalities; stochastic orderings (60E15) Numerical methods for functional equations (65Q20) Symbolic computation and algebraic computation (68W30)
The authors discuss a universal procedure and use results concerning stochastic orderings to obtain continuous solutions of inequalities of the form \(a_{1} f(\alpha_{1}x + (1-\alpha_{1}) y) + \cdots + a_{n} f(\alpha_{n}x + (1-\alpha_{n})y) \leq 0, x,y \in I\), where \(x, y\) are any points form the interval \(I \subset \mathbb{R}\), \(f : I \rightarrow \mathbb{R}\) is an unknown function and \(a_{i} \in \mathbb{R}\), \(\alpha_{i} \in [0, 1]\) are given numbers. The authors assume that \(a_{1} + \cdots + a_{n} = 0\) in this paper.\N\NIn the beginning, the authors give the basic tools and provide simple examples to explain their use. It is shown that the expressions the authors study may be written as Stieltjes integrals with respect to suitably constructed functions. The authors claim that a function \(f\) is convex of order \(n\) if \(f[x_{1}, \cdots, x_{n+2}]\geq 0\), \(x_{1}, \cdots, x_{n+2} \in I\). The authors provide a simple example of the application of the theorem that gives a simple sufficient condition for the higher-order convex ordering by considering the inequality \[f(x)-3f\left(\frac{2x+y}{3}\right)+3f\left(\frac{x+2y}{3}\right)-f(y)\leq 0.\] Further, they show that a continuous function \(f : \mathbb{R} \rightarrow \mathbb{R}\) satisfies the inequality \[-2f(x)+3f\left(\frac{3x+y}{4}\right)-2f\left(\frac{x+y}{2}\right)+3f\left(\frac{x+3y}{4}\right)-2f(y)\leq 0\] if and only if it is convex.\N\NMoreover, the authors present the construction of a computer program that solves inequalities of the above type after providing all necessary theoretical details. They divide the procedure in the following three steps: \N\N(1) They show that every continuous solution of the considered inequality is a convex function of some order;\N\N(2) They calculate the number of crossing points; and\N\N(3) They find the numbers \(\alpha_{i}\), that is a crucial step. \N\NEach step is based on several results, properly discussed in the paper. The computer method used to achieve their result is a Python function named \textbf{Linear\(_-\)inequality\((\cdot)\)}. This function includes all the steps from 1 to 3 and works by accepting a considered functional inequality as input.\N\NThe authors explore some results with accompanying graphs and others without graphical representation. For a complete overview of all results, including graphs and the Python source code, refer to the GitHub URL \url{https://github.com/CPOkeke/Solving-linear-functional-inequalities-with-a-computer}. The authors give the following examples on inequalities illustrating the application of the computer approach:\N\N\begin{itemize}\N\item \(\frac{1}{3}\big[f\big(\frac{2+\sqrt{2}}{4}x+\frac{2-\sqrt{2}}{4}y\big)+f\big(\frac{x+y}{2}\big) +f\big(\frac{2-\sqrt{2}}{4}x+\frac{2+\sqrt{2}}{4}y\big)\big] -\frac{1}{12} (f(x)+f(y)) -\frac{5}{12} \big[f\big(\frac{5+\sqrt{5}}{10}x+\frac{5-\sqrt{5}}{10}y\big)+f\big(\frac{5-\sqrt{5}}{10}x+\frac{5+\sqrt{5}}{10}y\big)\big]\leq 0\).\N\N\item \(\frac{4}{9}f\big(\frac{x+y}{2}\big) +\frac{5}{18}\big[f\big(\frac{5+\sqrt{15}}{10}x+\frac{5-\sqrt{15}}{10}y\big)+f\big(\frac{5-\sqrt{15}}{10}x+\frac{5+\sqrt{15}}{10}y\big)\big] -\frac{1}{12} (f(x)+f(y)) -\frac{5}{12} \big[f\big(\frac{5+\sqrt{5}}{10}x+\frac{5-\sqrt{5}}{10}y\big)+f\big(\frac{5-\sqrt{5}}{10}x+\frac{5+\sqrt{5}}{10}y\big)\big]\leq 0\).\N\N\item \(\frac{4}{45}f\big(\frac{x+y}{2}\big) +\frac{5}{18}\big[f\big(\frac{5+\sqrt{15}}{10}x+\frac{5-\sqrt{15}}{10}y\big)+f\big(\frac{5-\sqrt{15}}{10}x+\frac{5+\sqrt{15}}{10}y\big)\big] -\frac{1}{20} (f(x)+f(y)) -\frac{49}{180} \big[f\big(\frac{7+\sqrt{21}}{14}x+\frac{7-\sqrt{21}}{14}y\big)+f\big(\frac{7-\sqrt{21}}{14}x+\frac{7+\sqrt{21}}{14}y\big)\big]\leq 0\).\N\N\item \(\frac{16}{45}f\big(\frac{x+y}{2}\big)+\frac{49}{180} \big[f\big(\frac{7+\sqrt{21}}{14}x+\frac{7-\sqrt{21}}{14}y\big)+f\big(\frac{7-\sqrt{21}}{14}x+\frac{7+\sqrt{21}}{14}y\big)\big]-\frac{1}{30} (f(x)+f(y))-\frac{5}{12}\big[f\big(\frac{5+\sqrt{5}}{10}x+\frac{5-\sqrt{5}}{10}y\big)+f\big(\frac{5-\sqrt{5}}{10}x+\frac{5+\sqrt{5}}{10}y\big)\big] \leq 0\).\N\N\item \(\frac{1}{3}\big[f\big(\frac{2+\sqrt{2}}{4}x+\frac{2-\sqrt{2}}{4}y\big) +f\big(\frac{2-\sqrt{2}}{4}x+\frac{2+\sqrt{2}}{4}y\big)\big] -\frac{1}{9}f\big(\frac{x+y}{2}\big) -\frac{5}{18}\big[f\big(\frac{5+\sqrt{15}}{10}x+\frac{5-\sqrt{15}}{10}y\big)+f\big(\frac{5-\sqrt{15}}{10}x+\frac{5+\sqrt{15}}{10}y\big)\big]\leq 0\).\N\N\item \(\frac{1}{3}\big[f\big(\frac{2+\sqrt{2}}{4}x+\frac{2-\sqrt{2}}{4}y\big) +f\big(\frac{2-\sqrt{2}}{4}x+\frac{2+\sqrt{2}}{4}y\big)\big] -\frac{1}{45}f\big(\frac{x+y}{2}\big) -\frac{1}{20} (f(x)+f(y)) -\frac{49}{180} \big[f\big(\frac{7+\sqrt{21}}{14}x+\frac{7-\sqrt{21}}{14}y\big)+f\big(\frac{7-\sqrt{21}}{14}x+\frac{7+\sqrt{21}}{14}y\big)\big]\leq 0\).\N\N\item \(\frac{5}{12}\big[f\big(\frac{5+\sqrt{5}}{10}x+\frac{5-\sqrt{5}}{10}y\big)+f\big(\frac{5-\sqrt{5}}{10}x+\frac{5+\sqrt{5}}{10}y\big)\big] -\frac{1}{12} (f(x)+f(y)) -\frac{2}{3}f\big(\frac{x+y}{2}\big)\leq 0\).\N\N\item \(\frac{5}{18}\big[f\big(\frac{5+\sqrt{15}}{10}x+\frac{5-\sqrt{15}}{10}y\big)+f\big(\frac{5-\sqrt{15}}{10}x+\frac{5+\sqrt{15}}{10}y\big)\big] -\frac{1}{6} (f(x)+f(y)) -\frac{2}{9}f\big(\frac{x+y}{2}\big)\leq 0\).\N\N\item \(\frac{49}{180} \big[f\big(\frac{7+\sqrt{21}}{14}x+\frac{7-\sqrt{21}}{14}y\big)+f\big(\frac{7-\sqrt{21}}{14}x+\frac{7+\sqrt{21}}{14}y\big)\big]-\frac{7}{60} (f(x)+f(y))-\frac{14}{45}f\big(\frac{x+y}{2}\big) \leq 0\).\N\N\item \(\frac{1}{2} \big[f\big(\frac{3+\sqrt{3}}{6}x+\frac{3-\sqrt{3}}{6}y\big)+f\big(\frac{3-\sqrt{3}}{6}x+\frac{3+\sqrt{3}}{6}y\big)\big]-\frac{1}{3}\big[f\big(\frac{2+\sqrt{2}}{4}x+\frac{2-\sqrt{2}}{4}y\big)+f\big(\frac{x+y}{2}\big) +f\big(\frac{2-\sqrt{2}}{4}x+\frac{2+\sqrt{2}}{4}y\big)\big] \leq 0\).\N\N\item \(3f(\frac{3x+y}{4})+3f(\frac{x+3y}{4})-f(x)-4f(\frac{x+y}{2})-f(y)\leq 0\).\N\N\item \(9f(\frac{2x+y}{3})-8f(x)-f(y)\leq 0\).\N\end{itemize}
- A package for symbolic solution of real functional equations of real variables
- A solution to the problem of Raşa connected with Bernstein polynomials
- A tight Hermite-Hadamard inequality and a generic method for comparison between residuals of inequalities with convex functions
- An extension of Levin-Stečkin's theorem to uniformly convex and superquadratic functions
- Characterization of higher-order monotonicity via integral inequalities
- Computer aided solution of the invariance equation for two-variable Gini means
- Computer aided solution of the invariance equation for two-variable Stolarsky means
- Computer assisted solution of systems of two variable linear functional equations
- Functional equation characterizing polynomial functions and an algorithm
- scientific article; zbMATH DE number 51927 (Why is no real title available?)
- scientific article; zbMATH DE number 1354983 (Why is no real title available?)
- scientific article; zbMATH DE number 5224752 (Why is no real title available?)
- Inequalities
- Inequalities of Hermite-Hadamard type for higher order convex functions, revisited
- Inequalities of the Hermite-Hadamard type involving numerical differentiation formulas
- Ohlin type theorem and its applications for strongly convex set-valued maps
- Ohlin's lemma and some inequalities of the Hermite-Hadamard type
- On a class of functional inequalities, a computer approach
- On a generalization of a theorem of Levin and Stečkin and inequalities of the Hermite-Hadamard-type
- On convex functions of higher order
- On the Ohlin lemma for Hermite-Hadamard-Fejér type inequalities
- Solving linear two variable functional equations with computer
- Stochastic orders
- The interplay between linear functional equations and inequalities
- The s-convex orders among real random variables, with applications
This page was built for publication: Solving linear functional inequalities with a computer
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6887933)
