Subject: NA Digest, V. 19, # 19 NA Digest Sunday, May 12, 2019 Volume 19 : Issue 19 Today's Editor: Daniel M. Dunlavy Sandia National Labs dmdunla@sandia.gov Today's Topics: New MOOC on Programming for High Performance New Book, Linear Algebra and Learning from Data 2019 Golub Memorial Workshop, China, Apr 2019 Basilisk/Gerris Users' Meeting, France, Jun 2019 NUMTA2019, Italy, Jun 2019 French-German-Swiss Conference on Optimization, France, Sep 2019 Induction Course for New Lecturers in the Mathematical Sciences, Sep 2019 Mathematical Modelling and Computational Methods, Czech Republic, Sep 2019 Mathematical and Computational Aspects of Machine Learning, Italy, Oct 2019 Domain Decomposition Methods (DD26), China, Dec 2019 HPC Asia 2020, Japan, Jan 2020 Software Engineer Position, IBM Postdoc Positions, Mathematical Imaging/Optimization/Inverse Problems, Shenzhen Univ PhD & Postdoc Positions, Uncertainty Quantification, DTU, Denmark Predoc Position, TU-Vienna, Austria Contents, Constructive Approximation, 49 (2) Subscribe, unsubscribe, change address, or for na-digest archives: http://www.netlib.org/na-digest-html/faq.html Submissions for NA Digest: http://icl.utk.edu/na-digest/ ------------------------------------------------------- From: Robert van de Geijn rvdg@cs.utexas.edu Date: May 11, 2019 Subject: New MOOC on Programming for High Performance New MOOC on Programming for High Performance We are excited to announce that edX has opened registration for "LAFF-On Programming for High Performance" [1]. This free-to-audit, four-week, self-paced course developed by UT-Austin faculty Robert van de Geijn, Maggie Myers, and Devangi Parikh starts on June 4, 2019. The course uses the simple but important example of matrix-matrix multiplication to illustrate fundamental techniques for attaining high-performance on modern CPUs. A carefully designed sequence of exercises leads the learner from a naive implementation to one that effectively utilizes instruction level parallelism and culminates in a high-performance, multithreaded implementation. Along the way, it is discovered that careful attention to data movement is key to efficient computing. In other words, learners are exposed to techniques for attaining high performance through carefully scaffolded exercises that illustrate how the BLAS-like Library Instantiation Software (BLIS) [2] implements dgemm, which is itself based on Goto's algorithm [3]. We believe this course is appropriate for a novice yet of interest to an expert. It may be, for example, a great way to get a summer intern quite literally up to speed. Others may want to use it as a component in a class they teach. Some learners may merely come to the conclusion that they should be using high-performance libraries. Others may find they enjoy low level optimization. Please help us spread the word! [1] https://www.edx.org/course/laff-on-programming-for-high-performance [2] Field G. Van Zee, Robert A. van de Geijn. BLIS: A Framework for Rapidly Instantiating BLAS Functionality. ACM TOMS, 2015. [3] Kazushige Goto, Robert A. van de Geijn. Anatomy of high-performance matrix multiplication. ACM TOMS, 2008. ------------------------------------------------------- From: Gilbert Strang gilstrang@gmail.com Date: May 07, 2019 Subject: New Book, Linear Algebra and Learning from Data 2019 / xiv + 432 pages / Hardcover / 978-0692196-38-0 / Wellesley-Cambridge Press / List $87.50 / Individuals $70.00 This textbook explains the structure of deep learning, which constructs a function F(v) that correctly classifies the samples v in the training set. F is created from a chain of much simpler functions f(v) = max (Av + b, 0). The matrices A and the vectors b are the "weights" to be optimized. For applications to images, A is often a well localized convolution matrix. The heart of the book develops -- as clearly as we can -- the linear algebra and optimization methods and probability/statistics that go into the success of the learning function F. Please see detailed contents and complete sections on the book's website http://math.mit.edu/~gs/learningfromdata ------------------------------------------------------- From: Walter Gander gander@inf.ethz.ch Date: May 08, 2019 Subject: 2019 Golub Memorial Workshop, China, Apr 2019 The 2019 Gene Golub Memorial Workshop was held at Lanzhou University, Lanzhou, China, on April 19-23, 2019. This annual international workshop on matrix computations is dedicated to celebrating the life of the late Professor Gene Golub (February 29, 1932 - November 16, 2007) of Stanford University, USA. Gene was one of the very few preeminent scientists/engineers who was elected to both the USA National Academy of Sciences and the USA National Academy of Engineering. At the workshop, we reminisced about another side of Gene, the generous senior educator who always went out of his way to advise and help the next generation. About 60 people attended the workshop. They included senior researchers, young scholars, and graduate students. There were 13 invited speakers: Michele Benzi from Italy, Walter Gander and Martin H. Gutknecht from Switzerland, Wai-Ki Ching, Franklin T. Luk, and Tie-Yong Zeng from Hong Kong, and Zhong-Zhi Bai, Dan-Fu Han, Yu-Mei Huang, Zeng-Qi Wang, Rui-Ping Wen, You-Wei Wen, and Yu-Jiang Wu from the mainland of China. To follow in Gene's footsteps to encourage young researchers, the workshop initiated two Best Presentation awards. Professors Zhong-Zhi Bai, Michele Benzi, Walter Gander, Martin H. Gutknecht, and Franklin T. Luk constituted the Prize Committee. They chose Zeng-Qi Wang of Shanghai Jiao Tong University for the best oral presentation, and Jun-Feng Yin of Tongji University, Shanghai, for the best poster presentation. Each winner received a plaque, a certificate, and a small gift. The two organizers, Zhong-Zhi Bai and Yu-Mei Huang, did an excellent job and provided a nurturing environment for academic exchanges between the presenters and the students. The workshop received generous funding support from Lanzhou University and the Tianyuan Mathematical Center in Northwest China. This is the second time the Golub Memorial Workshop was held in the mainland of China. The first time was April 20-24, 2018, at the Hangzhou Normal University, Hangzhou. Previously, the Workshop was held for ten years from 2008 to 2017 at the Hong Kong Baptist University in Hong Kong. We are proud to show by example that Gene was wrong in his private remark to one of us, "No one will remember me five years after I pass away." For more details, please visit the website: http://lsec.cc.ac.cn/~NASCNAG/ Walter Gander and Franklin Luk ------------------------------------------------------- From: Stephane Popinet popinet@basilisk.fr Date: May 06, 2019 Subject: Basilisk/Gerris Users' Meeting, France, Jun 2019 The 5th Basilisk/Gerris Users' Meeting will be held in Paris from the 17th to 19th June 2019. The meeting is dedicated to the presentation of scientific results obtained using the Basilisk and Gerris free softwares in a broad range of fields including fluid mechanics, engineering, environmental science etc. Have a look at the web page of the previous meeting (http://basilisk.fr/BGUM2017) for examples of presentations. The meeting is free of charge but registration is mandatory. To register just follow the instructions on the meeting web page: http://basilisk.fr/BGUM2019#participants We would like to finalize the program early, so please try to register before the 17th of May. Do not forget to indicate a title for your presentation if you would like to talk. If you have already registered, please note that an extra day (19th) has been added to the meeting and that you are encouraged to vote on what should be covered that day. See: http://basilisk.fr/BGUM2019#informal-session-on-the-19th We will also provide a few travel scholarships for students wishing to attend but with limited funds. See: http://basilisk.fr/BGUM2019#travel-scholarships ------------------------------------------------------- From: Dmitri Kvasov kvadim@dimes.unical.it Date: May 06, 2019 Subject: NUMTA2019, Italy, Jun 2019 NUMTA2019 - the 3rd International Conference and Summer School "Numerical Computations: Theory and Algorithms" will be held in Italy in the All Inclusive "TH Le Castella Village" resort close to Crotone - the city of Pythagoras, from 15th to 21st June 2019. Numerical analysis, continuous optimization, and emerging computations are among the main topics of the Conference (see https://easychair.org/cfp/numta2019 for the NUMTA2019 Call-for-Papers). Plenary lectures will be delivered by Lou D'Alotto (USA), Renato De Leone (Italy), Kalyanmoy Deb (USA), Luca Formaggia (Italy), Jan Hesthaven (Switzerland), Francesca Mazzia (Italy), Panos Pardalos (USA), Michael Vrahatis (Greece), Anatoly Zhigljavsky (UK). Tutorials will be given by Roberto Natalini (Italy) and Yaroslav Sergeyev (Italy). The Springer Young Researcher Prize for the best NUMTA2019 presentation made by a young scientist will be awarded (applications can be sent by May 15, 2019). Proceedings will be published as a special volume of the world-renowned Springer "Lecture Notes in Computer Science" and will contain regularly peer-reviewed short and full papers (to be submitted by May 30, 2019). One-page abstracts of presentations at NUMTA2019 will be published as online Book of abstracts with its own ISBN (to be submitted by May 15, 2019: http://si.dimes.unical.it/~yaro/numta2019/abssubmit.php). Please consult the Conference web-site http://si.dimes.unical.it/~yaro/numta2019 for more details and contact the Organizers at numta2019@dimes.unical.it for information. ------------------------------------------------------- From: Jean-Baptiste Caillau jean-baptiste.caillau@univ-cotedazur.fr Date: May 06, 2019 Subject: French-German-Swiss Conference on Optimization, France, Sep 2019 FGS'2019 Submissions (talks & MS) deadline: May 15, 2019 Registrations are now open, early bird deadline: June 30, 2019 The 19th French-German-Swiss conference on Optimization will take place in Nice from September 17 to 20, 2019. This series of conferences began in 1980 at Oberwolfach and has gathered since French and German colleagues in optimization every other year. It is customary to invite a third country to participate. In 2019, the invited country is Switzerland. Theses conferences usually gather from 100 to 150 mathematicians and are the main meeting of European researchers in optimization in the broad sense. The 2019 edition is organized by LJAD and I3S labs from CNRS, and by Inria Sophia Antipolis Mediterranee center. The conference will be located on the Valrose campus of Universite Cote d'Azur, on the heights of Nice. A limited number of grants has been planned to cover the registration fees for students (undergraduate, graduate or postdoc). Students wanting to apply for these grants shall - send an email to fgs-2019@sciencesconf.org including a cv and asking for a grant - submit a talk (contributed or within a mini-symposium) before May 15, 2019. More information: fgs-2019.sciencesconf.org ------------------------------------------------------- From: Pamela Bye pam.bye@ima.org.uk Date: May 10, 2019 Subject: Induction Course for New Lecturers in the Mathematical Sciences, Sep 2019 Isaac Newton Institute for Mathematical Sciences, Cambridge 18 - 19 September 2019 Through a community initiative supported by the Institute of Mathematics and its Applications, the Isaac Newton Institute for Mathematical Sciences and the Heads of Departments of Mathematics Sciences (HoDoMS) and endorsed by the Royal Statistical Society, the Operational Research Society and the London Mathematical Society, we are delighted to announce that in September 2019 the two-day Induction Course for lecturers new to teaching mathematics and statistics within Higher Education will once again take place. The Induction Course for New Lecturers in the Mathematical Sciences has been designed by the mathematics community so that it is ideally suited for anyone who is new to or has limited experience teaching mathematics or statistics within UK higher education. It will be delivered by individuals with significant experience of teaching in the mathematical sciences and will focus upon the specific details and issues that arise in mathematics and statistics teaching and learning within higher education including topics such as: Lecturing; Supporting student learning; Making teaching interactive; Assessment, examinations and feedback; Linking teaching & research; Using technology to enhance teaching and learning; Using examples and mathematical problem solving. Additionally there will be significant opportunities for delegates to discuss their own ideas, challenges and experiences with the session facilitators so that individual queries can be answered. In the past, attendance has been recognised as contributing towards some introductory institutional programmes in learning and teaching for new staff, and for the 2019 Induction Course accreditation will be provided through the Institute of Mathematics and its Applications relative to the UK Professional Standards Framework for Teaching and Supporting Learning in Higher Education. Website for further information: https://ima.org.uk/11361/induction-course-for-new- lecturers-in-the-mathematical-sciences-2019/ For questions or queries about the academic content or structure of the Induction Course, please contact Michael Grove: m.j.grove@bham.ac.uk For questions or queries about the administration of the Induction Course, please contact the IMA Conference Officer at: conferences@ima.org.uk ------------------------------------------------------- From: Radim Blaheta radim.blaheta@ugn.cas.cz Date: May 12, 2019 Subject: Mathematical Modelling and Computational Methods, Czech Republic, Sep 2019 The 7th Conference MODELLING 2019 on Mathematical Modelling and Computational Methods will be held in Olomouc, Czech Republic, September 16-20, 2019. For the details, see http://www.ugn.cas.cz/link/modelling19 New deadlines: May 31, 2019 for Abstract submission (can be even later after an agreement with conference and minisymposia organizers). Early registration and payment Jun 15, 2019. ------------------------------------------------------- From: Michele Benzi michele.benzi@sns.it Date: May 09, 2019 Subject: Mathematical and Computational Aspects of Machine Learning, Italy, Oct 2019 School on Mathematical and Computational Aspects of Machine Learning (Pisa, 7-11 October 2019) The School will take place at the Ennio De Giorgi Mathematical Research Center ofthe Scuola Normale Superiore in Pisa, Italy, on October 7-11, 2019. The lectures will give a broad overview of the rapidly evolving field of Machine Learning from a mathematical, statistical, and computational perspective. The school is aimed primarily at PhD students, post-docs and young researchers. Lecturers and Topics: - Jean Barbier (ICTP Trieste): Mean Field Theory of High-Dimensional Bayesian Inference - Philipp Grohs (U. of Vienna): Approximation theory, Numerical Analysis and Deep Learning - Gabriel Peyre` (ENS Paris and CNRS): Optimal Transport for Data Science - Lars Ruthotto (Emory U.): Numerical Methods for Deep Learning - Stefano Soatto and Alessandro Achille (UCLA): Representation Learning Participation is free but limited to a maximum of 90 participants. There will be the possibility for a small number (10-12) of participants to give a short communication on their own work in the field of ML. Deadlines: Submission of abstracts: July 31, 2019 [decision by August 15] Application to participate: August 31, 2019 [decision by September 15] For more information and instructions on how to apply, please see http://crm.sns.it/event/451 ------------------------------------------------------- From: Felix Kwok felix_kwok@hkbu.edu.hk Date: May 09, 2019 Subject: Domain Decomposition Methods (DD26), China, Dec 2019 We are pleased to announce that the Chinese University of Hong Kong (CUHK) will be hosting the 26th International Conference on Domain Decomposition Methods (DD26) in Hong Kong, China from December 2 to 6, 2019. The purpose of the conference is to bring together mathematicians, computational scientists and engineers who work in the main themes of Domain Decomposition, including theoretical, algorithmic and implementation aspects of domain decomposition methods, solvers for multiphysics problems, parallel-in-time methods, multigrid and multilevel methods, fast solvers and preconditioning, and applications of such methods in physics and engineering. Registration will open shortly. In the meantime, we would like to extend our invitation for your participation and contributions. We are now ready to receive proposals for minisymposia that fit the themes of Domain Decomposition. Please provide a title, a short description of the subject matter, and a list of speakers for the minisymposium. Minisymposia may span one or more sessions, with each session consisting of four 25-minute talks, with an additional 5 minutes after each talk for questions. Please submit your proposals as a plain text e-mail to dd26@math.cuhk.edu.hk before June 15, 2019. For more information, including the list of confirmed plenary speakers, please visit the conference website at http://www.math.cuhk.edu.hk/conference/dd26, or e-mail us at dd26@math.cuhk.edu.hk. ------------------------------------------------------- From: Takeshi Fukaya fukaya@iic.hokudai.ac.jp Date: May 07, 2019 Subject: HPC Asia 2020, Japan, Jan 2020 HPC Asia 2020 - Call for Papers HPC Asia: International Conference on High Performance Computing in Asia-Pacific Region, January 15-17, 2020, Fukuoka, Japan High performance computing is a key technology to solve large problems in science, engineering, and business by utilizing computing power which has been evolving to the future. HPC Asia, which is an international conference series on HPC technologies in Asia Pacific region, and was held in the past several times in several countries in Asia regional site to discuss the issues on high performance computing and to exchange information of research and development results. Following the success in 2018 and 2019, the International Conference on High Performance Computing in Asia-Pacific Region, or HPC Asia 2020 is organized in order to exchange ideas, case studies, and research results related to all issues of high performance computing. Important Dates: - Abstract submissions due: August 2, 2019 (AoE, Anywhere on Earth, UTC-12) - Paper submissions due: August 9, 2019 (AoE, Anywhere on Earth, UTC-12) - Notification of Acceptance: October 14, 2019 - Camera-ready papers due: November 1, 2019 - Conference: January 15-17, 2020 Conference Web page: http://sighpc.ipsj.or.jp/HPCAsia2020/ E-mail contact: hpcasia2020@sighpc.ipsj.or.jp ------------------------------------------------------- From: Jacob Jelden jacob.jelden@ibm.com Date: May 07, 2019 Subject: Software Engineer Position, IBM The IBM ESSL team is in search of a Software Engineer to be part of an exciting opportunity to work on mathematical software for the next generation of supercomputers.The primary jobis to design, develop and validate high performing hybrid (CPU & GPU) scalable mathematical software. Additionally you may have the opportunity to expand your knowledge by working directly with other IBM teams, partners and customers to collaborate on designs and to solve system level problems related to compilers, GPU programming, Linux kernels and other software technologies. Qualified candidates will be expected to develop new subroutines and tune existing subroutines for new systems, processors, and GPUs as well as assist with other activities related to product delivery (e.g. build, test, packaging, support). If you are interested in this opportunity, please visit http://ibm.biz/227990BR for a more detailed description and submit you application. ------------------------------------------------------- From: J. Lu jianlu1979@163.com Date: May 09, 2019 Subject: Postdoc Positions, Mathematical Imaging/Optimization/Inverse Problems, Shenzhen Univ Postdoc Positions in Shenzhen University, China Location: Shenzhen, China Job Type: Full-Time Duration: 2 years Number of Position: 4 Positions Closing Date: Open Until Filled Description: We are looking for Postdoctoral Researchers in Image/Video Processing/Analysis,Mathematical Imaging, Optimization, Inverse Problems, Wavelet analysis, etc. We have no teaching tasks and check in / check out policy for Postdoctoral Researchers. Promotors: - Prof. Jian Lu (Shenzhen Key Laboratory of Advanced Machine Learning and Applications, College of Mathematics, Shenzhen University) - Prof. Lixin Shen (Syracuse University, USA) The salary is about 330,000 RMB (49,000 US dollars) per year, of which 180,000 RMB per year is after-tax. Those who are interested please send their C.V. to Prof. Dr. Jian Lu, e-mail: jianlu@szu.edu.cn; jianlu1979@163.com ------------------------------------------------------- From: Per Christian Hansen pcha@dtu.dk Date: May 10, 2019 Subject: PhD & Postdoc Positions, Uncertainty Quantification, DTU, Denmark The Technical University of Denmark opens a 3-year PhD position and a 2-year postdoc position, both starting Sept. 1, 2019 (or as soon as possible after that). The positions are part of the research project CUQI, Computational Uncertainty Quantification for Inverse Problems. For details see: https://www.compute.dtu.dk/english/cuqi The project's goal is to create a computational platform, suited for non-experts in academia and industry. A core problem is to formulate appropriate models of the priors for the desired solution and the reconstruction model, and to use these priors in large-scale numerical computations. E.g., we must handle solution priors for edges in an image, and we must incorporate hyperpriors in a reconstruction model to handle uncertain aspects of the physics and the acquisition model. The PhD project focuses on either one or both aspects of UQ, and it involves development of theory as well as computational algorithms, plus evaluation of the results on selected inverse problems. The applicant must have experience with mathematical modeling and analysis, as well as computational algorithms. For more details and to apply: https://tiny.cc/CUQI-PhD-1 The postdoc will be responsible for developing UQ models and methods that can handle a broad class of non-Gaussian and systematic errors in inverse problems, and for implementing and testing these methods by efficient and stable numerical algorithms. The aim is to develop a systematic and user-friendly approach - possibly using suitable approximations. The end result is a first attempt at producing an abstraction level that lets users focus on modeling and data analysis instead of mathematical details and low-level algorithm aspects. For more details and to apply: https://tiny.cc/CUQI-Postdoc-1 ------------------------------------------------------- From: Anton Arnold anton.arnold@tuwien.ac.at Date: May 08, 2019 Subject: Predoc Position, TU-Vienna, Austria - Job vacancy in the research group of Anton Arnold, at TU Vienna, 30 hours/week - Start: 1.6.2019 (or soon after), 4 years - Annual gross salary: 30.077 EUR Scientific topic: Numerical methods for highly oscillatory differential equations. The development of efficient numerical methods for highly oscillatory differential equations (e.g. Schroedinger equation in the classical limit) should be continued. On top of the employment at TU Vienna, an association with the ongoing doctoral school "Dissipation and dispersion in nonlinear partial differential equations" is offered, for details see: http://asc.tuwien.ac.at/~npde/ Requirements: completed master or diploma studies in mathematics; solid knowledge of (partial) differential equations and numerical analysis; programming experience The position comes with a teaching load of max. 4h/week (exercise classes). Application for position # 146.2: via e-mail to susanne.pietsch-brousek@tuwien.ac.at ; deadline 23.05.2019 For further scientific information contact: anton.arnold@tuwien.ac.at http://www.asc.tuwien.ac.at/~arnold/ ------------------------------------------------------- From: Ed Saff Constr.Approx@Vanderbilt.edu Date: May 06, 2019 Subject: Contents, Constructive Approximation, 49 (2) Constructive Approximation Volume 49, Issue 2, April 2019 http://link.springer.com/journal/365 Table of Contents Non-Self-Adjoint Toeplitz Matrices Whose Principal Submatrices Have Real Spectrum, Boris Shapiro and Frantisek Stampach Pluripotential Numerics, Federico Piazzon Behavior of the Roots of the Taylor Polynomials of Riemann's Psi Function with Growing Degreem Robert Jenkins and Ken D. T.-R. McLaughlin Solutions of Complex Differential Equation Having Pre-Given Zeros in the Unit Disc, Janne Groehn The Saff-Varga Width Conjecture and Entire Functions with Simple Exponential Growth, Antonio R. Vargas Optimal Monte Carlo Methods for L^2-Approximation, David Krieg On Mahler's Transcendence Measure for e, Anne-Maria Ernvall-Hytonen, Tapani Matala-aho, and Louna Seppala ------------------------------------------------------- End of Digest **************************