Contour-Based Fast Fractional Integration in a Jacobi Fractional Polynomial Basis

We develop a contour-based method for applying Riemann--Liouville fractional integrals in a Jacobi fractional polynomial basis with an endpoint weight. A Gamma-ratio representation reduces the action to contour quadrature and banded ultraspherical solves. For a fixed number of contour nodes, the cost and storage are linear in the polynomial dimension. For fractional equations, we construct selected diagonals of a short rational approximation directly and use their banded factorizations as reusable right preconditioners for Krylov iterations. Numerical experiments show accurate integral actions and substantially reduced solution costs across Abel, multi-order, BBO, deferred-correction, eigenvalue, and boundary-value problems.