Congratulations to the following 13 members of the SIAM community who will receive awards at the 2023 SIAM Conference Computational Science and Engineering (CSE23). A representative for each prize will give a talk at the conference. Additional information about each recipient, including Q&As, can be found below.
Matthew J. Colbrook, Andrew Horning, and Alex Townsend
Matthew J. Colbrook (Cambridge University), Andrew Horning (Massachusetts Institute of Technology), and Alex Townsend (Cornell University) are the 2023 recipients of the SIAM Activity Group on Computational Science and Engineering Best Paper Prize for their work in the SIAM Review paper "Computing Spectral Measures of Self-Adjoint Operators," published in 2021.
They will be recognized at the 2023 SIAM Conference on Computational Science and Engineering (CSE23) and Horning will present a talk about the paper on Monday, February 27 at 4:30 p.m. CET.
The
SIAM Activity Group on Computational Science and Engineering awards this prize every two years to the author(s) of the best paper, as determined by the prize committee, on the development and use of mathematical and computational tools and methods for solving problems that may arise in broad areas of science, engineering, technology, and society. The prize recognizes a paper that makes an outstanding and potentially long-lasting contribution to the field. The selection criteria emphasize multidisciplinary work opening up new areas of research, and potential broad impact, in addition to novelty, creativity, and overall scientific advancement and quality.
Matthew J. Colbrook is an assistant professor at the University of Cambridge. Before that, he was a Junior Research Fellow at Cambridge, an FSMP Fellow at Paris’ École Normale Supérieure, and earned his Ph.D. at Cambridge. He studies the analysis and development of algorithms related to approximation theory, spectral theory, solutions of PDEs, neural networks, data-driven dynamical systems, optimization, inverse problems, and a framework for determining the boundaries of what is and is not computationally possible (Solvability Complexity Index). His work on spectral computations has also won the IMA's Lighthill-Thwaites Prize, and his 2020 Ph.D. thesis, “The Foundations of Infinite-Dimensional Spectral Computations,” earned him the 2022 SIAM Richard C. DiPrima Prize. Learn more about Matthew J. Colbrook.
Andrew Horning is an applied mathematics instructor at the Massachusetts Institute of Technology (MIT) working in computational science and numerical analysis. Motivated by applications in physical applied math and engineering design, he develops computational methods that preserve and exploit key structures in complex large-scale and infinite-dimensional systems. Prior to MIT, Horning completed dual undergraduate degrees in physics and mathematics at Rensselaer Polytechnic Institute (2016) and a Ph.D. at the Center for Applied Mathematics at Cornell University (2021). He will be on the tenure-track job market in 2024.
Learn more about Andrew Horning.
Alex Townsend is an associate professor in the department of mathematics at Cornell University. His research is in computational mathematics and focuses on spectral methods, low-rank techniques, infinite-dimensional computations, and theoretical aspects of deep learning. Before Cornell, he was an applied mathematics instructor at MIT (2014-2016) and a DPhil student at the University of Oxford (2010-2014). He was awarded a Simons Fellows in Mathematics (2022), an NSF CAREER grant (2021), a SIGEST paper award (2019), the SIAM Activity Group on Linear Algebra Early Career Prize in applicable linear algebra (2018), and the Leslie Fox Prize for Numerical Analysis (2015). Townsend has been a member of SIAM since 2013. Learn more about Alex Townsend.
Q: Why are you all excited to receive the award?
A: While we do research for the excitement of exploration and discovery, it always feels good to have your work appreciated by the broader research community. We are excited about the exposure this award gives to infinite-dimensional computations, a topic that we love! In recent years, there has been a growing appreciation for infinite-dimensional computations. We enjoy the challenge of mixing numerical and functional analysis.
Finally, such prizes are extremely useful for early career researchers in terms of showcasing research and helping with job applications when they come on the job market. The opportunity to give a plenary talk at SIAM CSE23 provides an excellent platform.
Q: Could you tell us about the research that won your team the award?
A: When you listen to a piece of music, the sound signal consists of a sum of individual simple frequencies. The Fourier transform reveals these components and splits the system into simpler parts. More generally, this idea can be applied to linear operators and is known as diagonalization, where an operator is decomposed into simple constituent parts via its spectrum. While the spectrum of a finite matrix consists only of discrete eigenvalues, many infinite-dimensional operators in mathematical analysis and physical applications include a continuous spectral component. In particular, eigenvalues and eigenvectors do not diagonalize operators with continuous spectra. Instead, one must consider other quantities— so-called spectral measures.
In our paper, we develop an algorithm for computing smoothed approximations of spectral measures associated with general self-adjoint operators, requiring only a numerical solver for shifted linear equations and inner products. We use a combination of rational approximation theory and infinite-dimensional solve-then-discretize computations to derive high-order methods. Moreover, we figure out what is computationally possible and provide convergence theorems using the Solvability Complexity Index. The algorithm is publicly available in SpecSolve, a software package written in MATLAB.
Q: What does your team's work mean to the public?
A: Spectral measures provide insightful characterizations of complex, infinite-dimensional systems. They are intimately linked to correlation in stochastic processes and signal-processing, scattering cross-sections in particle physics, the local density-of-states in crystalline materials, mode decompositions in dynamical systems, and many other quantities. For example, since the publication of our work, our algorithm has been used by physicists to verify the existence of new quantum phenomena (bulk localized transport states). We hope our work continues to open new paths for scientists to explore infinite-dimensional systems and catalyzes discovery, understanding, and innovation.
Q: What does being a member of SIAM mean to your team?
A: To us, SIAM is not only a professional organization but also a community of like-minded researchers. Mathematics is not something done in isolation! We enjoy attending SIAM Annual Meeting and SIAM Conferences on Computational Science and Engineering and Dynamical Systems. Moreover, we publish in SIAM journals and greatly appreciate the attention to format and copy editing that makes the printed paper so beautifully typeset. We would like to take this opportunity to thank others in this community with whom we have interacted, especially our collaborators. It is an honor to be awarded this best paper prize, and it’s humbling to be placed alongside excellent previous winners.
Kaibo Hu
Kaibo Hu, University of Oxford, is the 2023 recipient of the SIAM Activity Group on Computational Science and Engineering Early Career Prize for contributions to the finite element exterior calculus, particularly structure-preserving numerical algorithms for magnetohydrodynamics. Hu will be recognized at the 2023 SIAM Conference on Computational Science and Engineering (CSE23) and will present a talk on Monday, February 27 at 5:00 p.m. CET.
The SIAM Activity Group on Computational Science and Engineering awards the prize every two years to one post-Ph.D. early career researcher in the field of computational science and engineering for outstanding, influential, and potentially long-lasting contributions to the field within seven years of receiving the Ph.D. or equivalent degree as of January 1 of the award year. The contributions must be contained in a paper or papers published in English in peer-reviewed journals.
Hu is a Royal Society University Research Fellow at the Mathematical Institute and Christ Church College of University of Oxford. He received his undergraduate degree from Nankai University in 2012 and Ph.D. at Peking University in 2017. After that, he was in postdoctoral positions at University of Oslo (2017-2018) and University of Minnesota (2018-2021), and with a Hooke Research Fellowship at Oxford (2021-2022). He was in the Simons Collaboration on Localization of Waves while he was in Minnesota. His current research is focused on structure-preserving numerical methods, particularly finite element exterior calculus and applications. Learn more about Kaibo Hu.
Q: Why are you excited to receive the award?
A: I am deeply honored and would like to thank SIAM for the award. The computational sciences and engineering field is a large and diverse community. I am excited about the community’s growing interest in theories and practices of structure-preservation and finite element exterior calculus. The award is also a reminder for me of the generous support that I receive from my teachers, colleagues, and the community.
Q: Could you tell us about the research that won you the award?
A: The research is about structure-preserving algorithms for magnetohydrodynamics (MHD). MHD is governed by a coupled PDE system, and solving it numerically is a challenging task. These equations have several nice structures, such as the magnetic Gauss law and helicity conservation. These properties are not only important for understanding the underlying physics but are also important for numerics in many circumstances. However, these properties may be lost in numerical computation due to discretization errors. The research investigated finite element methods that preserve these structures precisely at the discrete level. Particularly, topological and geometric ideas incorporated in finite element exterior calculus are crucial for dealing with these nonlinear equations.
Q: What does your work mean to the public?
A: MHD computation plays an important role in astrophysics and in the study of fusion energy, which has the potential of being a sustainable source of energy. It is exciting to hear about a major breakthrough made in this direction recently. Algorithmic advances also showcase that mathematics is both elegant and relevant to the real world.
Q: What does being a member of SIAM mean to you?
A: Being a member of SIAM provided me with opportunities to meet people with similar passions. While I completed my Ph.D., I gave a talk at my local SIAM student chapter – one of the first scientific talks I ever gave. The experience was both valuable and enjoyable. It is also a great pleasure to read and publish articles in SIAM's high-quality journals.
Devin A. Matthews and Field G. Van Zee
Devin A. Matthews, Southern Methodist University, and Field G. Van Zee, University of Texas at Austin, are the 2023 recipients of the James H. Wilkinson Prize for Numerical Software for the development of BLIS, a portable open-source software framework that facilitates rapid instantiation of high-performance BLAS and BLAS-like operations targeting modern CPUs.
They will be recognized at the 2023 SIAM Conference on Computational Science and Engineering (CSE23) and Matthews will present a talk about the software on Monday, February 27 at 4:00 p.m. CET.
The James H. Wilkinson Prize for Numerical Software is awarded every four years to the authors of an outstanding piece of numerical software, or to individuals who have made an outstanding contribution to an existing piece of numerical software. The prize is awarded for an entry that best addresses all phases of the preparation of high-quality numerical software. It is intended to recognize innovative software in scientific computing and to encourage researchers in the earlier stages of their career. It has been awarded at the International Council for Industrial and Applied Mathematics (ICIAM) meetings since 1991. Since 2019, SIAM awards the prize every four years at the SIAM Conference on Computational Science and Engineering
Devin A. Matthews obtained his bachelor’s degree in the department of chemistry and biochemistry from the University of Texas (UT) at Austin in 2010, and subsequently his Ph.D. as a Department of Energy Computational Science Graduate Fellow in 2014. After graduating, he held a postdoctoral position in the Oden Institute for Computational Engineering and Sciences as an Arnold O. Beckman Postdoctoral Fellow. In 2018, he started as a tenure-track assistant professor in the department of chemistry at Southern Methodist University (SMU), as well as a Faculty Fellow of the SMU Center for Research Computing. His research focuses on using and developing accurate theoretical methods to study molecules, reactions, clusters, and extended systems. His specialization is in ab initio methods based on quantum mechanics, combining concepts and techniques from chemistry, physics, mathematics, and computer science. Learn more about Devin A. Matthews.
Field G. Van Zee obtained a B.S. in 2003 and an M.S. in 2006 from the department of computer sciences at the University of Texas at Austin. He is currently a research scientist in the department of computer sciences and the Oden Institute. His research has focused on exploring ways to optimize and accelerate various numerical operations, ranging from matrix multiplication to application of Householder reflectors to the QR algorithm and eigenvalue and singular value decompositions. For the last 10 years, he has focused on building the technical and social infrastructure to support and sustain BLIS and its community into a bright future of continued innovation. Learn more about Field G. Van Zee.
Q: Why are you both excited to receive the award?
A: We often work in the shadows. Linear algebra operations are usually buried at the bottom of the computational stack, and so most people may not even be aware our software is there. A major award like the James H. Wilkinson Prize is great validation that we’re really making a difference to the HPC community and beyond. We also hope that this extra publicity will help to get everyone as excited as we are about the great new features which are added to BLIS all the time!
Q: Could you tell us about the research that won your team the award?
A: On one hand, BLIS is a BLAS library; that is, it is used to perform operations like matrix multiplication as rapidly as possible on a wide range of CPU architectures. However, BLIS is also the result of decades of advances in understanding computer architectures at a fundamental level using formal derivation of matrix algorithms, analytical performance modeling, cache-aware layering of algorithms, and other techniques. This has allowed us to create a clean separation of architecture-dependent and algorithm-dependent components, which makes porting BLIS to new architectures quick, easy, and efficient. We have also used the separation to enable new high-performance implementations of algorithms such as tensor contraction, k-nearest neighbors, Gaussian kernel summation, Strassen’s algorithm, and related methods – all reusing the same underlying approach and high-performance micro-kernels.
Q: What does your team's work mean to the public?
A: Our primary goal is to make dense linear algebra (DLA) faster. Whether people know it or not, linear algebra is everywhere, both in research like modeling and machine learning, as well as in things like video games. This means that people all over the world stand to benefit from the DLA implementations already included in BLIS. But we also aim to support people in the computational trenches as they work to optimize new algorithms and improve portability across architectures. We really see BLIS as a framework for high-performance algorithms, where DLA is just the beginning of what is possible algorithmically.
Q: What does being a member of SIAM mean to your team?
A: SIAM membership is quite valuable to us in that it provides access to SIAM conferences and other events. We have found SIAM Conferences on Parallel Processing for Scientific Computing and Computational Engineering and Science to be excellent venues to share our research and meet new collaborators. SIAM also publishes a number of top-notch research journals such as SIAM Journal on Scientific Computing, which have been effective places to publish our results. And of course, you have to be a member to be eligible for the awards!
Carol S. Woodward, Cody J. Balos, Peter N. Brown, David J. Gardner, Alan C. Hindmarsh, Daniel R. Reynolds, and Radu Serban
The SUNDIALS Core Development Group, consisting of Carol S. Woodward, Cody J. Balos, Peter N. Brown, David J. Gardner, Alan C. Hindmarsh, Daniel R. Reynolds, and Radu Serban, are the 2023 recipients of the SIAM/ACM Prize in Computational Science and Engineering. The group received the award for innovative research and development of nonlinear and differential/algebraic equation solvers for high-performance computing that provides unique, critical capabilities in the scientific software ecosystem.
They will be recognized at the 2023 SIAM Conference on Computational Science and Engineering (CSE23) and Woodward will present a talk on Monday, February 27 at 5:30 p.m. CET.
The SIAM/ACM Prize in Computational Science and Engineering is awarded every two years by SIAM and the
Association for Computing Machinery (ACM) in the area of computational science to one individual or a group of individuals in recognition of outstanding contributions to the development and use of mathematical and computational tools and methods for the solution of science and engineering problems.
Carol S. Woodward is a Distinguished Member of the technical staff at Lawrence Livermore National Laboratory (LLNL). Prior to joining LLNL, she received her Ph.D. in computational and applied mathematics at Rice University and her bachelor’s degree in mathematics from Louisiana State University. Her research interests include numerical methods for nonlinear partial differential equations, nonlinear and linear solvers, time integration methods, numerical software development, and parallel computing. She leads the SUNDIALS team and actively participates in several projects to transition modern numerical methods into large-scale scientific computing application codes. She is a Fellow of SIAM and of the Association for Women in Mathematics. Learn more about Carol S. Woodward.
Cody J. Balos is a computational scientist in the Center for Applied Scientific Computing at LLNL where he is a core developer of the SUNDIALS project and the Extreme-scale Scientific Software Development Kit (xSDK). He is also a member of the FASTMath SciDAC-5 Institute and NUCLEI SciDAC-5 partnership. His research interests include time integration algorithms for Exascale computing and next-generation platforms, numerical methods for multiphysics and multiscale problems, scientific machine learning, and sustainable scientific software. He holds a bachelor’s degree in computer engineering from the University of the Pacific and a master’s degree in computational and applied mathematics from the University of Washington, Seattle. Learn more about Cody J. Balos.
Peter N. Brown has been a staff member at Lawrence Livermore National Laboratory since 1987 and a member of the Center for Applied Scientific Computing at LLNL since its creation.He received his Ph.D. in mathematics from Tulane University in 1978 and was a tenured professor at the University of Houston before coming to LLNL. He is an expert in numerical methods for linear and nonlinear systems, and for the past 18 years, he has worked on deterministic methods for neutron and radiation transport. He is currently a member of the WCI Deterministic Transport Project team and his current research interests include the development of scalable solution methods for neutral particle transport described by the Boltzmann transport equation. Learn more about Peter N. Brown.
David J. Gardner is a computational scientist in the Center for Applied Scientific Computing at Lawrence Livermore National Laboratory. He joined LLNL in 2014 after completing his Ph.D. in computational and applied mathematics at Southern Methodist University (SMU) where he also received a B.S. in mathematics. Prior to attending SMU, Gardner graduated from Brookhaven College with an A.S. in mathematics. His research focuses on numerical methods and software for simulating nonlinear partial differential equations on high-performance computing systems with an emphasis on time integrators and nonlinear solvers for multiscale, multiphysics applications. Learn more about David J. Gardner.
Alan C. Hindmarsh worked as a mathematician at Lawrence Livermore National Laboratory from 1968 until his retirement in 2002. He received his B.S. degree from the California Institute of Technology in 1964 and Ph.D. from Stanford University in 1968. At LLNL, his research interests centered on numerical methods for ordinary differential equations, especially stiff systems. He has authored and co-authored dozens of general-purpose software packages for ODE systems and related problem areas. Since retiring, he has continued at LLNL on guest status. Learn more about Alan C. Hindmarsh.
Daniel R. Reynolds is a professor and Chair of the Department of Mathematics at Southern Methodist University. His research focuses on the development and application of robust time integrators and iterative nonlinear and linear solvers for large-scale multiphysics systems — simulations comprised of multiple interacting physical processes. He received a Ph.D. in computational and applied mathematics from Rice University in 2003. Prior to joining SMU, Reynolds held postdoctoral research positions in the department of mathematics and the Center for Astrophysics and Space Sciences at the University of California, San Diego, and in the Center for Applied Scientific Computing at Lawrence Livermore National Laboratory. Learn more about Daniel R. Reynolds.
Radu Serban is a senior scientist in the department of mechanical engineering at the University of Wisconsin-Madison. He received his Ph.D. from the University of Iowa in 1998 under the supervision of Edward J. Haug and his bachelor’s degree from the Polytechnic Institute of Bucharest in 1992. Prior to joining UW-Madison, Serban was a postdoctoral researcher at the University of California, Santa Barbara, a computational scientist in the Center for Applied Scientific Computing at LLNL, and worked for a start-up in Silicon Valley. His research interests are in computational dynamics, scientific computing, sensitivity analysis for dynamical systems, and mathematical software. He is one of the technical leads and the main contributor to the Chrono open-source multi-physics library. Learn more about Radu Serban.
Q: Why are you all excited to receive the award?
A: We are really excited and honored to receive this award because it recognizes the impact of work on adaptive methods for ODEs, DAEs, and Newton-Krylov methods, as well as the extra effort to harden the methods into robust, documented general use software. We want to thank the numerous people who have contributed in small and large ways to SUNDIALS and the methods inside it, as well as all of the institutions, application teams, and community members that provided the support that made this work possible. We derive tremendous satisfaction from the flow of notifications we receive about our work being cited and used in so many interesting and varied applications.
Q: Could you tell us about the research that won your team the award?
A: SUNDIALS is known worldwide as one of the most effective and efficient software libraries for time integrators and nonlinear solvers. Time integration methods implemented within SUNDIALS include both step and order adaptive linear multistep methods with forward and adjoint sensitivities as well as step adaptive multistage Runge-Kutta methods, including Implicit/Explicit (ImEx) additive Runge-Kutta methods and several variants of multirate methods. In addition, SUNDIALS includes several solvers for nonlinear, algebraic systems including globalized Newton-Krylov and accelerated fixed point methods. The packages are written to be flexible for use on numerous computing platforms with differing programming models, including serial, distributed, and shared memory parallel, and GPU-based systems. SUNDIALS is freely available and includes detailed user manuals to aid users in installation, interfacing, and application of the packages.
The SUNDIALS core development team has worked on methods for time integration and nonlinear solvers for decades and some were pioneers in the development of methods and software for stiff ODE and DAE systems, as well as the Newton-Krylov methods that enabled efficient implicit solvers on large-scale computing platforms. Recently, team members have developed novel multirate methods for systems with multiple separated time scales. Team members have implemented these methods into highly efficient, portable, and well-documented packages that allow nonexperts to benefit from these and related methods.
The packages in SUNDIALS have been used worldwide in a myriad of simulation-dependent applications, from watershed models to fusion device models to reacting flow simulations. Users come from government labs, academia, and industry. In addition, SUNDIALS has been incorporated into numerous third-party libraries.
Q: What does your team's work mean to the public?
A: Simulation of scientific and engineered systems has become increasingly necessary to address fundamental challenges resulting from society’s increasing reliance on technology. To meet the needs of evolving complex simulations, numerical methods have been developed and hardened into software packages that can be included in these simulations. By providing such a package, the work of the SUNDIALS core development team allows scientists and engineers to focus on scientific discoveries or engineering for the public by providing a reliable tool to solve the mathematical equations fundamental to modeling and simulating physical phenomena. In addition, high-quality, publicly funded, and open-source software like SUNDIALS forms the basis for numerous tools with more widespread usage — i.e., Python, Mathematica, and MATLAB.
Q: What does being a member of SIAM mean to your team?
A: SIAM is the premier professional organization for the development and use of numerical software, and we have all found a home at SIAM over the years. We are especially appreciative of the support SIAM gives to the computational science community through the SIAM Activity Group on Computational Science and Engineering. SIAM’s Conferences on Computational Science and Engineering, in particular, have enabled us to learn more about the work going on outside our institutions, to find new topics of research, to inspire new ideas, and meet people in the field who work on similar methods and software. Having the ability to interact with other numerical software developers has significantly strengthened SUNDIALS over the years.