News and events

Imperial CMSE provides a forum to advertise and promote news and events aligned with CMSE objectives. Find below a list of upcoming events.

Research Computing Summer School 2019

September 24 - 27, 2019

The 2019 instalment of the Research Computing  Summer School (formerly known as the HPC summer school) will be held on September 25-27 at Huxley 342 and 308, South Kensington campus.  

The focus of this year's summer school is machine learning.  We will have 2.5 days of tutorials followed by an afternoon of exciting presentations of machine learning applications.

The detailed programme is here.  You can register for separate days.  All are welcome!

Laplace equation in a polygon

Speaker: Nick Trefethen, University of Oxford

You are warmly invited to the inaugural lecture for the joint Imperial-UCL numerics seminar series.  The talk will be followed by a wine reception at the Blackett foyer.

Abstract:

Consider the Laplace equation in a polygon with continuous Dirichlet boundary data.  One could compute the solution u with finite elements, based on a two-dimensional representation of the solution, or integral equations, based on a one-dimensional representation.  We propose a "zero-dimensional" representation: u is the real part of a rational function with poles exponentially clustering near each vertex.  Thanks to an effect first identified by Donald Newman in 1964, the convergence is root-exponential as a function of the number of degrees of freedom, i.e. of the form exp(-C*sqrt(N)) with C>1.  In practice, with 40 lines of Matlab code, we can solve problems with 3-8 vertices in a second of laptop time, with 8-digit accuracy all the way up to the singularities in the corners.  Evaluation of the solution takes around 20 microseconds per point.  This is joint work with Abi Gopal.

Research Computing Summer School 2018

September 24 - 27, 2018

The 2018 instalment of the Research Computing  Summer School (formerly known as the HPC summer school) will be held on September 24-27 at Huxley 340 and 341, South Kensington campus.  

The summer school will present a special edition of a Software Carpentry workshop that will teach the basis of Bash shell, Git version system, SQL, database design, R and Python.  In addition, we included an advanced day on Scientific Python and R.  The school will conclude with an introduction to the College's new research data storage service and Imperial Research Software Engineering Showcase.

The detailed programme is here.  You can register for separate days.  All are welcome!

Mini-workshop on Software and Mathematics of Spectral Methods

Firedrake tutorial

Firedrake (firedrakeproject.org) is a system which automates the numerical solution of partial differential equations using the finite element method. Firedrake users write high level mathematical code for the PDEs they wish to solve, and high performance parallel implementations are automatically generated at runtime. Firedrake makes it vastly easier to create high performance simulations, and even automates the creation of an adjoint simulation. Users at Imperial and around the world employ Firedrake for a wide range of forward and inverse simulation challenges.

The Firedrake team will be offering a half-day hands-on Firedrake tutorial at Imperial on 12 January 2018. If you're considering using Firedrake or are just curious about it, you're welcome to come and participate. The tutorial is free but registration is required. For more information and to sign up, go to: http://www.firedrakeproject.org/tutorial_jan_18.html

A Parallel Framework for Generalized N-body Problems

 Speaker: Aparna Chandramowlishwaran, University of California, Irvine

You are warmly invited to this research talk, which should be of wide interest in spanning machine learning, data mining, computational science, and compiler technology.  She gave a terrific talk at a Dagstuhl Seminar a few weeks ago and I am pleased and excited to have her visit us.  If you would like to arrange a meeting, please contact Dr Pablo Gonzalez de Aledo p.gonzalez-de-aledo-marugan@imperial.ac.uk (a senior RA in my group) who is hosting her visit.

Abstract: 

In this talk, I'll present our work on addressing key challenges in developing parallel algorithms and software for the class of N-body problems on current and future platforms. Our goal is to reduce the apparent gap in performance between code generated from high-level forms and that of hand-tuned code, which we address using extensive characterization of the optimization space for these computations and automating the process through domain-specific code generators. These application-specific compilers provide the domain scientists the ability to productively harness the power of these large machines and to enable large-scale scientific simulations and big data applications. 

In this talk, I’ll present Portal, a new high-performance domain-specific language (DSL) and code-generator designed to enable high-performance implementations of N-body problems on modern multicore machines. Our goal in the development of Portal is three-fold, (a) to implement scalable, fast algorithms which have O(n log n} and O(n) complexity, (b) design an intuitive language and API to enable rapid implementations of a variety of algorithms, and (c) to enable parallel large-scale problems to run on current and future architectures. The initial target includes N-body problems in machine learning and data mining such as k-nearest neighbors, kernel density estimation, Expectation Maximization, to name a few which are commonly used in various applications. To our knowledge, there are no known libraries or currently available frameworks that implement parallel machine learning N-body algorithms and Portal aims to fill this gap. Moreover, our language and intermediate algorithm representation are independent of the architecture, making our approach portable and easily extensible for different platforms. 

Bio:

Aparna Chandramowlishwaran is an assistant professor at the University of California, Irvine, in the Department of Electrical Engineering and Computer Science. Her research lab, HPC Factory, is interested in "all-things" high-performance computing with an emphasis on parallel algorithms, performance analysis and tuning, and domain-specific compilers. She is a recipient of the Intel Ph.D. fellowship (2012), ACM/IEEE George Michael Memorial HPC fellowship (2010), Best Paper Award at IEEE Parallel and Distributed Processing Symposium (IPDPS, 2010), and co-recipient of the ACM Gordon Bell Prize (2010) among others. She received her Ph.D. in Computational Science and Engineering from Georgia Institute of Technology (Georgia Tech) in 2013 and was a research scientist at MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) with the Compilers at MIT (COMMIT) research group.

HPC summer school 2017

September 18 - 22, 2017

The third instalment of the HPC summer school will be held on September 18 -22 in SAF 121 and 120, South Kensington campus.

The programme includes a two-day scientific python tutorial for beginners, a one-day workshop on deep learning and a thorough introduction to Amazon Web Services.   In addition, we are planning a Research Software Engineering session, a session in Bioinformatics and Research Data Management.  The last day of summer school is dedicated to quantum technologies.  

You can register for separate workshops or community sessions.  All are welcome!