News
Article Featuring Hans' ICIAM-Presented Research In Spanish News Magazine
Hans' research on parallel-in-time methods applied to computational science and data science problems was featured in an article that appeared in the Spanish national-scale Sunday news magazine XL Semanal, on the occasion of the ICIAM 2019 conference in Valencia. |
Panel on Mathematics and Machine Learning at ICIAM2019
Hans De Sterck will be a panelist at the Special Panel on "The Future of Mathematics in the Age of Machine Learning" at the ICIAM2019 conference in Valencia, Spain, on 16 July 2019. Hans will specifically talk about "Computational Science and Engineering and Machine Learning". |
Invited Lecture at ICIAM 2019
Hans De Sterck gives an Invited Lecture at the ICIAM 2019 conference on "Scalable Solvers for Computational Science and Data Science: Multilevel, Nonlinearly Preconditioned, and Parallel-in-Time", 17 July 2019, Valencia, Spain. |
Alexander Howse defends his PhD thesis (September 29, 2017)
Waterloo PhD student Alexander Howse defended his PhD thesis on "Nonlinear Preconditioning Methods for Optimization and Parallel-In-Time Methods for 1D Scalar Hyperbolic Partial Differential Equations" on September 29, 2017. |
February 2016: Postdoctoral Position, Scalable Solvers for CFD on Adaptive Grids
Candidates are sought for a Postdoctoral Research Fellow Position at the School of Mathematical Sciences, Monash University, Melbourne, Australia. The position focuses on developing novel numerical methods and software for parallel simulation of compressible fluids on adaptive grids. Specific research topics of interest include parallel linear and nonlinear solvers for high-order accurate implicit time integration, and parallel adaptive grid refinement with error estimation. The fellow will be supervised by Professor Hans De Sterck, in a project that involves collaboration with Professor Clinton Groth from the University of Toronto. We are looking for accomplished candidates with a PhD in applied/computational mathematics, mechanical/aerospace engineering, or computational science, and with specialization in one or more of numerical methods for PDEs (hyperbolic conservation laws), scalable solvers, or computational fluid dynamics. Experience with parallel programming in C++ is desirable. This is one of two postdoctoral positions in an interdisciplinary project on "Advanced Simulation Methods for the Coupled Solar Interior and Atmosphere". The second postdoctoral fellow will use the advanced simulation techniques developed in the project to address challenging questions in the area of wave propagation in the solar interior and atmosphere. Monash University is a leading Australian research university that ranks in the top-100 globally. Monash University is located in Melbourne, which is a major cosmopolitan centre and has been named the world's most liveable city for five years in a row. This postdoctoral position is for two years, with an intended start date between June and September 2016 (negotiable). The annual salary is approximately A$80,000. For more information and to apply online, please visit http://tinyurl.com/Monash-SciCom-postdocs The closing date for applications is 31 March 2016. Please contact hans.desterck@monash.edu with any questions. |
Monash Workshop on Numerical PDEs, 15-19 Feb, 2016
The schedule is now available for the first "Monash Workshop on Numerical PDEs" that Jerome Droniou and Hans De Sterck are organizing at Monash University's School of Mathematical Sciences on February 15-19 2016. event website: https://monashpde.eventbrite.com.au The program includes 9 exciting invited presentations from international experts in the field, and about 20 contributed presentations. |
Our paper on population-level modeling of the smoking epidemic appears in BMC Public Health
Submitted Paper: A New Line Search Speeds Up LBFGS for Parallel Logistic Regression in Spark
Accepted for SIAM Data Mining conference 2016 - We submitted "A polynomial expansion line search for large-scale unconstrained minimization of smooth L2-regularized loss functions, with implementation in Apache Spark" (Hynes and De Sterck). See http://arxiv.org/abs/1510.08345. Our new line search uses a simple polynomial expansion idea, but for smooth objective functions (like logistic regression) it makes the line search more accurate than commonly used approaches, which decreases the number of iterations required by LBFGS and leads to speed-up in parallel of 30% and more. |
SIAM News Article on "Data Science and How to Teach It"
I wrote a SIAM News Article with Chris Johnson on Data Science, and How to Teach It. The data revolution is shaping up to become one of the great new
quantitative endeavours of our time, and, as in all quantitative fields,
mathematics is poised to play an important role. We make the point that there are important synergies between
data science and computational science, which make it clear that educational
programs in areas like “computational and data science” or “mathematics
of data and computation” hold significant promise for interdisciplinary
success. |