Predictive modelling in physical science and engineering is mostly based on solving certain   partial differential equations where the complexity of solutions is dictated by the geometry of the domain. Motivated by the broad applications of explicit solutions for spherical and ellipsoidal domains, in particular, the Eshelby's solution in elasticity, we propose a generalization of ellipsoidal shapes called polynomial inclusions. A polynomial inclusion (or p-inclusion for brevity) of degree k is defined as a smooth, connected, and bounded body whose Newtonian potential is a polynomial   of degree k inside the body. From this viewpoint, ellipsoids are identified as the only p-inclusions of degree two, and many fundamental problems in various physical settings admit simple closed-form solutions for general p-inclusions as for ellipsoids. Therefore, we anticipate that p-inclusions will be   useful for applications including predictive materials models, optimal designs, and inverse problems. However, the existence of p-inclusions beyond degree two is not obvious, not to mention their direct algebraic parameterizations. 

In this work, we explore alternative definitions and properties of p-inclusions in the context of potential theory. Based on the theory of variational inequalities, we show that p-inclusions do exist for certain   polynomials, though a complete characterization remains open.  We   reformulate the determination of surfaces of p-inclusions as nonlocal geometric flows which are convenient for numerical simulations and   studying geometric properties of p-inclusions.  In two dimensions, by the method of conformal mapping we find an explicit algebraic parameterization of p-inclusions. We also propose a few open problems whose solution will deepen our understanding of relations between domain geometry, Newtonian potentials, and solutions to general partial differential equations.   We conclude by presenting examples of applications of p-inclusions in the context of Eshelby inclusion problems and magnet designs. 

15 Jun 2023
2pm - 3pm
RM 2464, HKUST (2/F, Lift # 25/26)
Prof. Liping LIU
Professor, Department of Mathematics & Department of Mechanical Aerospace Engineering, Rutgers University, NJ, USA
Department of Mechanical & Aerospace Engineering
Payment Details
Faculty and staff, PG students
Other Events
26 Jun 2023
Seminar, Lecture, Talk
Department of Mathematics - Seminar on PDE - Some recent results on asymptotically Poincare-Einstein manifolds
Poincar-Einstein manifolds are a class of noncompact Riemannian manifolds with a well- defined boundary at infinity. They appear as the framework of AdS/CFT correspondence in string the...
23 Jun 2023
Seminar, Lecture, Talk
Department of Mathematics - Seminar on Applied Mathematics - Deep Particle: learning invariant measure by a deep neural network minimizing Wasserstein distance on data generated from an interacting particle method
High dimensional partial differential equations (PDE) are challenging to compute by traditional mesh-based methods especially when their solutions have large gradients or concentrations...