NA Digest, V. 18, # 17

NA Digest Sunday, April 29, 2018 Volume 18 : Issue 17

Today's Editor:

Daniel M. Dunlavy
Sandia National Labs

Today's Topics: Subscribe, unsubscribe, change address, or for na-digest archives:

Submissions for NA Digest:

From: Field G. Van Zee
Date: April 28, 2018
Subject: BLIS 0.3.2 now available

We are pleased to announce BLIS 0.3.2, made possible in part thanks to
a flurry of community activity and contributions as well as funding
from NSF and industry.

BLIS is a software framework for rapidly instantiating high-
performance BLAS-like dense linear algebra operations. Aside from
being a developer's tool, the framework also allows all users to build
libraries that implement a superset of the BLAS while providing a BLAS
compatibility interface by default. (An optional CBLAS API is also

BLIS 0.3.2 provides varying levels of optimized kernel support across
a variety of microarchitectures, from the well-supported Intel
Haswell, Broadwell, Skylake, and Kabylake CPUs to others, such as
ARMv7/ARMv8, for which we offer more preliminary support. We also
provide optimizations for AMD's recent Zen microarchitecture as well
as initial real-domain support for the AVX-512 instruction set
(Intel's Skylake-X and Knights Landing).

Other recent developments include fully integrated BLAS test drivers,
a rewritten configure-time hardware detection heuristic, the ability
to build multi-architecture libraries where kernels are selected at
runtime, and implementations of recently-published research on
inducing complex matrix multiplication.

For links to code, wikis, mailing lists, and other information, please
visit the BLIS github website [1].


From: Robert van de Geijn
Date: April 28, 2018
Subject: MOOC on Programming for Correctness

"Program testing can be used to show the presence of bugs, but never
to show their absence!" -- Edsger Dijkstra

On May 15, 2018, "LAFF-On Programming for Correctness" starts
again on the edX platform: .
A short video that launches Week 0 can also be viewed on YouTube: .

This MOOC is a humble attempt to share some of the foundational
mathematics that enables one to derive programs to be correct,
hand-in-hand with their proofs of correctness. We think it can
benefit novices as well as established software scientists.

The course builds upon the "FLAME" project that yielded PLAPACK, the
libflame library, and the BLAS-like Library Instantiation Software
(BLIS). The techniques also influenced the Elemental library for
distributed memory architectures. It exposes some of the "secret
sauce" that has allowed us to rapidly develop portable
high-performance libraries with a high degree of confidence.

We gratefully acknowledge support for this MOOC from MathWorks, the
National Science Foundation, and The University of Texas at Austin.

Date: April 25, 2018
Subject: New Book, Nonlinearity: Ordinary and Fractional Approximations

Series: Studies in Systems, Decision and Control, Springer, 2018
George A. Anastassiou
Nonlinearity: Ordinary and Fractional Approximations by Sublinear and
Max-Product Operators

Approximation by sublinear operators with applications to Max-product
operators is a new trend in approximation theory. These operators are
nonlinear and rational giving very fast and flexible
approximations. In this book we focus more in approximations under the
presence of ordinary and fractional smoothness. We present both the
univariate and multivariate cases. We deal also with approximations
under convexity. The book's results are expected to find applications
in many areas of pure and applied mathematics, especially in
approximation theory and numerical analysis in both ordinary and
fractional sense.

From: Walter GAnder
Date: April 24, 2018
Subject: Gene Golub Memorial Workshop, China, Apr 2018

Zhong-Zhi Bai (Chinese Academy of Sciences, China) and Dan-Fu Han
(Hangzhou Normal University, China) organized an International
Workshop on Matrix Computations to honor the memory of Gene Golub in
Hangzhou, China, on April 20–24, 2018. Former students, collaborators
and friends of Gene from China, Hong Kong, Russia and Switzerland
presented research works on the favorite topics of Gene: linear
equations, inner-outer iterations, secular equations, matrix splitting
and preconditioning, block triangular and skew-Hermitian splitting
methods, SOR-like methods, HSS preconditioner for saddle point
problems, and so on. For each year from 2010 to 2017, Frank Luk and
Walter Gander organized a Golub Memorial Day at the Hong Kong Baptist
University. We are pleased that Zhong-Zhi Bai has agreed to continue
this tradition in China. As we all know, Gene loved to encourage and
promote young talents, including many from China (he also loved
Chinese food!). The students at the Hangzhou workshop were moved and
inspired by the many stories they heard about Gene. They got to
understand why there is an annual SIAM summer school organized in
Gene's name. So the legacy of Gene Golub (1932-2017) is present and
alive! The next Golub Memorial Workshop will be held in Lanzhou,
China, in April 2019.

Franklin Luk and Walter Gander

From: Jochen Schuetz
Date: April 23, 2018
Subject: Hyperbolic Conservation Laws, Belgium, Jun 2018

Hereby, I would like to kindly invite you to participate in the summer
school on hyperbolic conservation laws at the university of Hasselt,
Belgium, from June 25-27, 2018.

Many processes are described by physical laws expressing the
conservation of mass, momentum, or energy. Commonly, the related
mathematical models are systems of partial and ordinary differential
equations. One illustrative example is the air flow around planes
under standard flight conditions. In mathematical terms, this flow is
modelled by the Navier-Stokes or the Euler equations, which form a
particular set of mathematical conservation laws.

The focus of the summer school is on hyperbolic conservation laws.
The mathematical treatment of conservation laws is a complex and
challenging task. In comparison to other types of equations, the
hyperbolic conservation laws may have multiple solutions, which may
become non-smooth. This makes both the mathematical analysis and the
numerical approximation quite cumbersome. Most of the available
results are restricted to simplified and hence non-realistic cases,
like one-dimensional problems. In the general situations the existence
of solutions is still an open research question. Also, in case of
multiple solutions, entropy criteria have to be defined to select the
physically relevant solution.

The goal of this summer school is to give the participants a basic
introduction to the analysis and the numerics of hyperbolic
conservation laws. The focus will be more on an introduction to the
concepts including their applications rather than a complete and
rigorous discussion of the theory. The summer school will therefore be
amenable to mathematicians, engineers and scientists and will be
presented in a genuine interdisciplinary context. The summer school
will be open to participants having a basic knowledge in calculus and
linear algebra courses as they are taught to engineers and
scientists. All sessions will be hands-on, meaning that sufficient
time for exercises and their discussion is planned.

More information and subscription procedures can be found on
or feel free to contact me via
Participation is free of charge!

From: Claudio Canuto
Date: April 26, 2018
Subject: Associate Professor Position, Politecnico di Torino

The Department of Mathematical Sciences (DISMA) of the Politecnico di
Torino, Italy, opens a call for expressions of interest for a tenured
Associate Professor position in the field of discrete or computational
mathematics. The position is supported by a national research grant
obtained by DISMA as "Dipartimento di Eccellenza 2018-2022".

Full details and contacts can be found at

From: Miguel A. Benítez
Date: April 27, 2018
Subject: Fellow Positions, BCAM

BCAM expresses it's interest to host Marie S-Curie Fellows in
Mathematics (MAT), Physics (PHY), Economic Sciences (ECO) and
Information Science and Engineering (ENG). We kindly ask you to
distribute this call among colleagues and potential candidates. For
further information, please visit the following links:

- BCAM expression of Interest: Brief description of the institution
and research lines

- European Participant Portal (Programme Website)

All applications must be submitted on line. Deadline for submissions:
10th August, 2018.

From: Michail Stamatakis
Date: April 23, 2018
Subject: Postdoc Position, Kinetic Monte Carlo Method, Univ College London

Research Associate - Harnessing Emerging Hardware Architectures In
Kinetic Monte-Carlo Simulations - Ref:1712391

The Multiscale Computational Catalysis and Materials Science group,
led by Dr Michail Stamatakis at UCL Chemical Engineering, develops
computational methodologies and software for mesoscale simulations of
catalytic materials, and implements these methods to unravel the
complexity of important reactions and design superior catalysts. More
details can be found at and

We are seeking an experienced and motivated post-doctoral researcher
to develop innovative methods and algorithms for kinetic Monte-Carlo
(KMC) simulations of reactions on catalytic surfaces. These
algorithms will harness modern/emerging hardware architectures such as
GPUs and FPGAs. Within the remit of the project is also the
implementation of these algorithms in our KMC software application
Zacros (

The post is funded for 3 years in the first instance.

From: Daniel B Szyld
Date: April 27, 2018
Subject: Postdoc Position, Parallel Asynchronous Methods for PDEs

The Department of Mathematics, Temple University, is inviting
applications for a postdoctoral position as an NTT Research Assistant
Professor. This is a non-tenure-track appointment. The appointment is
for two years, to start on July 1st, 2018, or at another mutually
agreeable date. The appointment may be extended for a third year.

This research position is part of a project on asynchronous parallel
methods for the solutions of PDEs using Schwarz methods. The ideal
candidate will have knowledge of Domain Decomposition methods and/or
parallel codes, and possibly asynchronous iterations.

In addition to the research project, the candidate is expected to
teach a light teaching load at least for the first year of the

Applications should include the standard AMS Cover Sheet for Academic
Employment, a cover letter, a curriculum vitae, and a statement about
current and future research. A complete application will also include
at least three letters of recommendations. All application materials
should be submitted via

Inquiries may be directed to

Review of applications will begin immediately. Temple University is an
Affirmative Action/Equal Opportunity Employer, and is committed to
increasing the diversity of its faculty. We especially encourage
applications from women, minorities, and other under-represented
groups in the mathematical sciences.

Official Mathjobs posting:

From: Lieven De Lathauwer
Date: April 23, 2018
Subject: PhD and Postdoc Positions, EOS Project KULeuven/VUB/UCL/UMons

We are advertising several PhD and postdoctoral positions for an EOS
project on "Structured Low-Rank Matrix/Tensor Approximation: Numerical
Optimization-Based Algorithms and Applications" (2018-2021). This
excellence-of-science project involves a consortium of four Belgian
universities: KULeuven (L. De Lathauwer, B. De Moor, P. Patrinos,
M. Van Barel), VUB (I. Markovsky, M. Ishteva), UCL (P.A. Absil, F.
Glineur) and UMons (N. Gillis).

We are seeking outstanding candidates who have obtained a master's or
doctoral degree in mathematical engineering, computer science,
electrical engineering, mathematics or physics. Strong candidates have
a proven record in numerical mathematics, optimization, systems &
control, machine learning and/or signal processing. A good knowledge
of linear algebra is mandatory but experience with tensor techniques
is not required. Candidates must be fluent in English.

Further information and application form are available from .

From: Josef Weinbub
Date: April 23, 2018
Subject: PhD Position, Computational Science, TU Wien, Austria

One PhD position (3 years, 40 hours/week) at the Institute for
Microelectronics, TU Wien is available. The area of research is on
high performance surface evolution algorithms and data structures.

The candidate must be skilled in developing C++ code on Linux. In
particular, we are looking for someone who enjoys developing
scientific, numerical software. Previous experience in this area as
well as with parallel programming (e.g. OpenMP) is a plus. Previous
exposure to semiconductor fabrication physics is not(!) required.

Desired Starting Date: June 2018
(but is flexible and can be negotiated)

The position will remain open until filled.

More information:

From: Ainara Gonzalez
Date: April 26, 2018
Subject: PhD Position, High Performance Computing, Project ENABLE

BCAM - Basque Center for Applied Mathematics is looking for 1 PhD
Student in the field of High Performance Computing, under the
supervision of:
- Johan Jansson, CFD Computational Technology research line leader at
BCAM (Spain) and Assistant Professor at KTH (Sweden)
- Franck Girot, Ikerbasque Research Professor at UPV/EHU (Spain)

Research fields: HPC, Mathematical Modelling, Computational Methods.

Contract and offer: 3-year contract, extensible one year more as KTH

Salary and conditions: BCAM will appoint the successful applicant
under an employment contract with a very competitive salary.

Application deadline: 31 May 2018 at 17:00 CET (UTC+1)

How to apply: Send an email to
Please, put in the subject of your email that your are applying for
the ESR6 position and check that you meet all eligibility criteria.

Required documents: CV, letter of interest and 2 recommendation

Envisaged job starting date: 01/02/2019 (at the latest)

Click on the link to see the job offer:

From: Carolin Gietz
Date: April 25, 2018
Subject: PhD Position, Univ of Munster

The Faculty of Mathematics and Computer Science at the University of
Munster, Germany, invites applicants for the Ada Lovelace PhD Position
(salary level 13 TV-L, 75%). The position is temporary for three
years. The regular weekly working time for full employment is 39 hours
and 50 minutes. The Ada Lovelace PhD Programme was initiated by the
program for female professors (Professorinnenprogramm II) of the state
of North Rhine-Westphalia, Germany, to promote gender equalization for
junior scientists. The position comes with additional resources for
travelling and material expenses. Applicants are expected to have a
master's degree or equivalent, which will allow them to pursue
doctoral studies in mathematics or computer sciences at the University
of Munster. Applicants are also expected to have above-average final
grades. Finally, applicants must be able to integrate into an existing
working group or project group. The successful candidate is expected
to graduate within three years. The University of Munster is an equal
opportunity employer and is committed to increasing the proportion of
women academics. Consequently, we actively encourage applications by
women. Female candidates with equivalent qualifications and academic
achievements will be preferentially considered within the framework of
the legal possibilities. We also welcome applications from candidates
with severe disabilities. Disabled candidates with equivalent
qualifications will be preferentially considered. Applications should
include the following documents: a letter of motivation, CV, a short
description of the planned PhD project, degree certificates, and a
letter of recommendation from a group leader in Munster including a
statement on how the applicant will integrate into the local working
group/research group. Complete applications should be sent
electronically as a single PDF file by 1 June 2018 to the following
email address:
Dean of the Faculty of Mathematics and Computer Science (FB 10)
Einsteinstrasse 62
48149 Munster

From: Yuji Nakatsukasa
Date: April 25, 2018
Subject: Student Intern Positions, Linear Algebra or NA, NII, Tokyo

National Institute of Informatics (NII, located in central Tokyo) is
looking for student interns (in their PhD or master's program),
supported by the MOU program. Please visit the webpage for more

I am looking for enthusiastic individuals interested in eigenvalue
problems, least-squares problems, approximation theory or numerical
integration. The specific topic is fairly open-ended and can reflect
the intern's particular interest. Candidates are expected to have some
background in numerical linear algebra or numerical analysis. The
length and period is quite flexible (2-6 months), expected to start
some time this summer.

A (big) advantage is given to students affiliated with our MOU partner
institutions (see
for the list), but exceptional candidates from elsewhere will be
considered. The deadline for application is May 29th, 2018. Please
contact me at if you are interested in working
with me, at least a week before the deadline.

From: Claude Brezinski
Date: April 23, 2018
Subject: Contents, Numerical Algorithms, 78 (1)

Table of Contents
Numerical Algorithms, Vol. 78, No. 1

Pipeline implementations of Neumann-Neumann and Dirichlet-Neumann
waveform relaxation methods, Benjamin W. Ong, Bankim C. Mandal

An efficient gradient method with approximate optimal stepsize for
large-scale unconstrained optimization, Zexian Liu, Hongwei Liu

Implementation of reduced gradient with bisection algorithms for
non-convex optimization problem via stochastic perturbation, Abdelkrim
El Mouatasim

Efficient implementation of symplectic implicit Runge-Kutta schemes
with simplified Newton iterations, Mikel Antonana, Joseba Makazaga,
Ander Murua

A new integrable convergence acceleration algorithm for computing
Brezinski-Durbin-Redivo-Zaglia's sequence transformation via
pfaffians, Xiang-Ke Chang, Yi He, Xing-Biao Hu, Shi-Hao Li

An algorithm for the inversion of Laplace transforms using Puiseux
expansions, Tongke Wang, Yuesheng Gu, Zhiyue Zhang

Deterministic sparse FFT for M-sparse vectors, Gerlind Plonka, Katrin
Wannenwetsch, Annie Cuyt, Wen-shin Lee

An efficient alternating direction method of multipliers for optimal
control problems constrained by random Helmholtz equations, Jingshi
Li, Xiaoshen Wang, Kai Zhang

Iterative methods for solving proximal split minimization problems,
M. Abbas, M. Al Shahrani, Q. H. Ansari, O. S. Iyiola, Y. Shehu

Trivariate near-best blending spline quasi-interpolation operators,
D. Barrera, C. Dagnino, M. J. Ibanez, S. Remogna

An eighth-order exponentially fitted two-step hybrid method of
explicit type for solving orbital and oscillatory problems,
J.M. Franco, L. Randez

Relaxed two points projection method for solving the multiple-sets
split equality problem, Ya-zheng Dang, Jian Yao, Yan Gao

A partitioned finite element scheme based on Gauge-Uzawa method for
time-dependent MHD equations, Qing Zhang, Haiyan Su, Xinlong Feng

A modified generalized shift-splitting preconditioner for nonsymmetric
saddle point problems, Zheng-Ge Huang, Li-Gong Wang, Zhong Xu,
Jing-Jing Cui

A supernodal block factorized sparse approximate inverse for
non-symmetric linear systems, Massimiliano Ferronato, Giorgio Pini

End of Digest