Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Home Topics Titles Quotes Blog Featured Help
Hui Liu
University of Calgary
Calgary, Canada

Hui Liu works at the University of Calgary, Canada, as a research associate. He holds a PhD degree in Computational Mathematics and Parallel Computing from the Chinese Academy of Sciences (2010), and a BSc degree in Computational Mathematics from the University of Science and Technology of China (USTC, 2005).

In 2005, after fulfilling his bachelor’s degree in computational mathematics, he was accepted for a PhD scholarship at the Academy of Mathematics and Systems Science, Chinese Academy of Sciences, where he studied computational mathematics and parallel computing. During his PhD program, he studied adaptive finite element methods (h-, p-, hp-adaptive methods), parallel computing, dynamic load balancing, encoding and decoding of Hilbert space-filling curves, and algorithm design. He developed encoding and decoding algorithms for arbitrary Hilbert space-filling curves.

In 2010, he joined the Reservoir Simulation Group, University of Calgary, and worked on GPU computing. It is well known that linear solvers occupy most simulation time during black oil simulations, and if linear solvers are accelerated, reservoir simulations can be sped up. He worked on the acceleration of linear solvers and preconditioners using GPUs, including Krylov subspace solvers, algebraic multigrid solvers, and various preconditioners. He implemented two linear solver packages: one for a single GPU and another for multi-GPUs.

In 2013, he started a parallel platform project to support the development of large-scale reservoir simulations. The platform is designed for distributed-memory parallel systems and uses MPI for communications. It provides gridding, load balancing, mapping, parallel linear solvers, preconditioners, well modeling, visualization, keyword parsing, and parallel input and output. The platform has been utilized for black oil models, compositional models, and thermal models. New parallel preconditioners specialized for reservoir simulation have been developed. The platform and parallel reservoir simulators are scalable, and large-scale reservoir models with billions of grid cells can be simulated.

His research interests include numerical methods for partial differential equations, reservoir simulation, linear/nonlinear solvers, algebraic multigrid solvers, preconditioners, parallel computing, and GPU computing. He has had several papers published in the Journal of Computational Mathematics, Journal of Computational Physics, Numerical Linear Algebra with Applications, and Computers & Mathematics with Applications.


A direct tridiagonal solver based on Givens rotations for GPU architectures
Venetis I., Kouris A., Sobczyk A., Gallopoulos E., Sameh A.  Parallel Computing 49(C): 101-116, 2015. Type: Article

Tridiagonal linear systems arise from discretizations of differential equations. The associated matrices have non-zero elements only on the main diagonal and on the two diagonals directly above and below the main diagonal. In practice, very large ...


A computational approach to nonparametric regression: bootstrapping CMARS method
Yazıcı C., Yerlikaya-Özkurt F., Batmaz İ.  Machine Learning 101(1-3): 211-230, 2015. Type: Article

A well-defined model can relate phenomena and conclusions, which can enhance our understanding of knowledge and help our work decisively. In statistics, one popular research topic is to formulate mathematical models using existing data. Many metho...


Simulating rigid body fracture with surface meshes
Zhu Y., Bridson R., Greif C.  ACM Transactions on Graphics 34(4): 1-11, 2015. Type: Article

Simulation of rigid body fracturing has been a hot research topic in computer graphics and applied sciences, such as collision detection and the simulation of explosions. Many methods have been proposed and applied to rigid body fracturing, includ...


Performance modeling for hierarchical graph partitioning in heterogeneous multi-core environment
Chan S., Ling T., Aubanel E.  Parallel Computing 46(C): 78-97, 2015. Type: Article

Graph partitioning is an interesting topic in parallel scientific computing that has been studied for decades. For many numerical methods, such as finite element methods (FEMs), finite difference methods, and finite volume methods, a mesh has to b...


 Uniformly convergent hybrid schemes for solutions and derivatives in quasilinear singularly perturbed BVPs
Zheng Q., Li X., Gao Y.  Applied Numerical Mathematics 91(C): 46-59, 2015. Type: Article

A singularly perturbed boundary-value problem is a boundary-value problem that contains a small parameter whose value cannot be approximated by setting to zero. The one-dimensional singularly perturbed quasilinear convection-diffusion problem is c...


Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright © 2000-2016 ThinkLoud, Inc.
Terms of Use
| Privacy Policy