Adaptive integration of 5-convex and 5-concave functions
Szymon Wąsowicz
TL;DR
This work addresses adaptive numerical integration for $5$-convex or $5$-concave functions in $C^6[a,b]$ by leveraging a three-point Gauss quadrature $\mathcal{G}$ and a four-point Lobatto quadrature $\mathcal{L}$. The authors derive a practical composite scheme $Q_n$ with a rigorous error bound $|\int_a^b f- Q_n|\le\tfrac14|\mathcal{L}_n-\mathcal{G}_n|$, and provide a terminating adaptive algorithm that increases the subdivision until $|\mathcal{L}_n-\mathcal{G}_n|\le 4\varepsilon$. Theoretical results show convergence of the composite quadratures and guarantee termination, while numerical experiments on $f(x)=1/x$ and $f(x)=e^x$ demonstrate substantial efficiency gains over prior $3$-convex methods. Implemented in SageMath, the approach offers a robust and faster alternative for high-order convexity classes relevant in numerical analysis and computational mathematics.
Abstract
An adaptive method connected with 3-point Gauss quadrature and 4-point Lobatto quadrature is introduced and investigated for 5-convex functions.
