Variable-Order Fractional Partial Differential Equations: Computation, Analysis, and Application

University Of South Carolina At Columbia, Columbia SC

Investigators

Abstract

Mathematical modeling and simulation techniques have been widely used in science, engineering, and industry. In this project, we consider a class of models of complex phenomena which exhibit memory effects and long range interactions, with applications in design and manufacturing of visco-elastic materials, anomalous diffusive transport, hydrofracking in gas and oil recovery, bioclogging of porous materials, and the deformation of some materials such as in orthopedic implants and shape memory polymers. The focus is on fractional calculus and specifically on variable order fractional partial differential equations, in which the fractional order may be a function of space, time and even unknown solutions. The research activities will contribute to the analysis, simulation, modeling and application of fractional calculus, and provide advanced interdisciplinary training to students. The project includes training opportunities for graduate students. Fractional partial differential equations (FPDEs), which are characterized by power-law decaying tails, have shown to accurately model complex phenomena of nonlocal nature. However, rigorous mathematical and numerical analysis of variable-order FPDEs is currently less known than that for integer-order PDEs. For instance, it is well known that linear elliptic and parabolic FPDEs imposed on smooth domains with smooth data exhibit weak initial or boundary singularity, which is in sharp contrast to their integer-order analogues. This makes it unrealistic to carry out error estimates of numerical approximations to FPDEs based on the (often untrue) smoothness assumptions of their true solutions. In this project the investigators develop accurate and stable numerical approximations to variable-order FPDEs and their fast solution algorithms, as well as prove their well-posedness and smoothing properties. The investigators will also prove optimal-order error estimates of numerical approximations to variable-order FPDEs without any artificial regularity assumption of their true solutions, but only under the regularity assumptions of their coefficients, variable orders and other related data. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.