Posted on by Arthur Guillec Posted in Blogs 2021, PRACE Summmer of HPC 2021, Projects 2021 — No Comments ↓

Hello everyone, we are Arthur &Tristan, two engineering students from Paris and we will compete in the Summer of HPC program.

First a few words about us : 

My name is Arthur. I’m 21 and I’m from the south of France. I studied in Paris for the first 4 years and I will do an exchange semester in Milan next year. I have a particular affection for computer / computational science, telecommunication, astronomy and new manufacturing processes.

Hi my name is Tristan, I am from the French Alps and I am 22 years old. I grew up near the Switzerland border, studied there until high-school, then I moved to Lyon and finally Paris for my studies. Being from the mountains I love ski and sports in general. 

We are both students in our 5th year of Applied Mathematics and Computer Science in the engineering school Polytech Sorbonne. (Below you can see a map of our campus located just next to the Seine in Paris.)

Our project

The aim of this project is to combine two things. First, we want to evaluate all pairwise interactions in a system of N electrical charges. Secondly, we would like to be able to evaluate these interactions on massively parallel systems in order to decrease the computation time and to increase the number of particles in the simulations.

The best sequential algorithms have linear complexity. The problem is therefore to find the best way to parallelize them, to compare the different solutions available. There are many frameworks that can be used for this purpose. Instinctively, one thinks of OpenMP, MPI, etc.

During this project we will focus on HPX. Once the implementation is done, we will try to measure how much of the resources we are able to exploit because most current implementations are unable to take full advantage of the resources available on clusters.

HPX

During this project, we will use a general purpose C++ runtime system for parallel and distributed applications called HPX. It is developed by the STE||AR group at Louisiana State University. This is an alternative to OpenMP/MPI which aims to solve the difficulties encountered by MPI on very large supercomputers

We recommend anyone interested in the subject to refer to the HPX documentation which contains a very clear explanation of the difference between HPX and more conventional high performance computing frameworks.

Posted on by Tiziano Barbari Posted in Blogs 2021 — No Comments ↓

Hello and welcome to my blog!

This blog will be a companion in this journey through High Performance Computing (HPC), and I will try to update it frequently. In this first post I will introduce myself and briefly the project.

About myself

My name is Tiziano, I live in Italy and, when I don’t code, I enjoy playing chess and board games, playing the piano, playing with cats.

Cat trying to relax and his human trying to pet him.

I obtained my BSc in mathematics last September. During my undergraduate studies I grew an interest for math applications, coding, and the algorithmic side of mathematical proofs; I also attended a number of computational laboratories, on physics simulations, cellular automata, graphs, state machines, number theory – to mention a few.

In the meanwhile, thanks to online courses, I approached the broad field of data science, in particular natural language processing (NLP). As a consequence, my bachelor thesis was on weighted automata in NLP and I ended up enrolling in the MSc in data science at Padua University.

The way I approached the world of supercomputers is perhaps unusual: I had courses on mathematical optimisation and, when coding, I am also interested in speed and efficiency. All this led to a chain of thoughts and to some research on code optimisation; I tried out some programming contests, then I found some parallel computing courses, and eventually I discovered HPC and this Summer (School) of HPC, organised by PRACE (Partnership for Advanced Computing in Europe). It is a programme for students interested in working on the topic and willing to undertake HPC projects, as well as to try to popularise HPC itself.

The ALMA (Atacama Large Millimeter Array) correlator, in Atacama Desert (Chile), is a supercomputer used for astronomical research.
By ALMA, CC BY 4.0, https://commons.wikimedia.org/w/index.php?curid=25103046.

I immediately knew that I wanted to participate, so I applied right away! The field in general was very interesting to me and the proposed projects in particular looked intriguing, as far as I could understand. When, after a coding test and a motivation letter, I discovered that I had been accepted, I felt super excited!

About the project

Of all this year’s projects, my colleague Morten and I have been assigned project 2125, which has the slightly cryptic name of “Scaling HMC on large multi-CPU and/or multi-GPGPUs architectures”. We will implement an algorithm known as Hamiltonian Monte Carlo (the HMC in the project title) that is used in statistics, and computational physics, for ‘sampling from probability distributions’, to quote the project description.

As you can probably imagine by now, our code will not run on our personal computers, but on Kay, Ireland’s main supercomputer! That’s right, we will access powerful HPC resources and exploit parallelism to achieve results that would be otherwise unimaginable.

Thanks for your attention so far! Let’s get back to the study of the math involved and to the code, and see how the project evolves. See you soon, for a more detailed description of the project, the updates, and all the technicalities.

Posted on by lzervos Posted in Blogs 2021, PRACE Summmer of HPC 2021, Projects 2021 — No Comments ↓

Welcome to my PRACE Summer of HPC blog.

Hello reader (or bot 😁),

As you probably understood from the title, my name is Lazaros Zervos. I am 23 years old and I was born and raised in Athens, Greece. I recently received my B.Sc. degree from National and Kapodistrian University of Athens with major in Computer Science and I’m also starting my M.Sc. studies in October.

The relation between programming and me began back in 2015. While in high school I was very passionate about Math, but was never introduced to the magic of coding. After studying for the university entrance exams which included a course called “Application Development in Programming Environment”, I knew that Computer Science was my thing.

Then, it was in the 3rd year of my Bachelor’s that I discovered Artificial Intelligence. It combined my 2 interests, Computer Science and Mathematics in a unique way. So here I am, working on a NLP project in the frame of SoHPC program.

Get to know more about me

Since I was young, I was very active both in my “professional” life and outside of it. I like all kinds of sport: Football, Basketball, Tennis, Martial arts, etc. I also play Chess at an amateur but still very good level. I can say with certainty that those hobbies which may seem unimportant for someone, for me they have shaped up at some degree my personality.

Preparing to start my game in the Greek Team Cup Final 8 2017 (it’s me on the left)

My team’s project

During the summer I will work on the Cross-Lingual Transfer learning for biomedical texts project, which is organized and mentored by the Text Mining Unit of Barcelona Supercomputing Center. I will have the pleasure to work on this very interesting project, not alone, but with my teammate Aslihan Uysal, or as we call her Asli.

Nevertheless, a very important aspect of the project is the use of Supercomputers. Model Training in Machine Learning is a data-intensive computing job, so there is a need for High Performance Computing power to have an efficient result in short time. Among others, we’ll have the opportunity to work with one of the most powerful supercomputers in the world, MareNostrum.

You can find out more about it in the video below and if you’re more interested you can also take a virtual tour or even an actual visit should you’re sometime in Barcelona.

Learn more about the specs of MareNostrum after its last upgrade

Conclusion

It’s a great chance for everyone interested in Natural Language Processing to study the results and the work done on the project. Therefore, comment below and keep in touch with the blog to stay up-to-date as new posts are coming in the following weeks.

Posted on by Morten Holm Posted in Blogs 2021, PRACE Summmer of HPC 2021 — No Comments ↓

Welcome back! Pretty cool of you to read part two, if you’re seeing this as the first post, I also encourage you to view the previous blog entry as I’ll attempt to make a post every week 🙂

This past week I got to meet my mentors and the general outline of the project was explained. We were handed 8 articles totalling 933 pages (I think I’ll skip reading a book over the summer), with more material incoming along the way. Further, we have been registered in their systems along with a provided email and applied for access to the Hartree supercomputer resources using ssh. Tiziano and I also met up over zoom to discuss the project, our approach to it as well as our past experiences in coding and the material.

So.. about the project, what on earth are we even doing?

The key to understanding any project is first and foremost to understand its headline so bear with me, the project title is “Scaling HMC on large multi-CPU and/or multi-GPGPUs architectures” and it surely is a mouthful. So let’s start by breaking down the title. Scaling in this sense just means that we can make a larger or smaller system that still functions. A system in this sense could for example be analogous to our bread baking metaphor from the previous post. Where doubling our recipe, preferably still yields us the same result as we want the final result to exhibit the same behaviour whether we’re baking a single bun or bread to feed an entire household. Like any system, everything has a boundary where other stuff tends to takes over, like what would even happen if we tried to bake an earth-sized bread. Besides the weight taking over and collapsing its airy interior, what other boundaries could you imagine? instead of going big, what would happen if we went ultra-small?

Earth-sized bread viewed from the moon.
Photo by NASA and Monika Grabkowska; Edited by me

The next part “HMC” is simply Hamiltonian Monte Carlo, so that is of cause pretty self-explanatory… yeah no, it’s definitely not, but allow me to get back to that part later on as it requires its own dedicated post. For now, we can try to accept the idea as a way to pick a random sample for a model using probabilities.

The last part is large multi-CPU and basically, it is how many CPU‘s can we throw at the problem until it is solved before we die of old age.. or maybe just boredom.

Throwing a CPU at a problem
Photo by Keith Johnston; Edited by me

So if we were to compare this to bread… I’m joking, I’ll lay off the bread for a while. So now we can imagine a factory.. that makes bread and essentially having more CPU’s is like the factory having more workers, the production or in this case computation doubles when you add another worker and overall we get to be more efficient. But like most processes’ it doesn’t really double as we encounter limitations. A process might not be able to run in parallel with other parts and we experience a slowdown. The more we can eliminate bottlenecks and streamline the process, the faster we can make the code run.

For the keen reader, you probably remember there’s another part of the title and that boils down to the architecture part. where we have two different layouts to our factory. The CPU has a specific layout/architecture internally that allows it to perform tasks. However, the General-Purpose Graphics Processing Unit (GPGPU) has another layout that in some cases also performs better than CPU’s. Implementing the GPGPU interface require some additional steps to make it work but essentially perform the same task. This sort of brings us full circle as the scaling part can be that we are limited by what a GPU can store and compute and how we can scale beyond using a single GPU. What we have to do is divide the problem up into bits where each of the bits gets its own GPU.

This weeks conclusion

Explaining concepts can be hard and I hope this approach of setting up these spur of the moment analogies helps grasp the context of the project. Moving on it will most likely get more technical and next weeks blog will be about the Hamiltonian Monte Carlo specifically. But I also encourage you to look at some of the other projects and I’ll try to highlight a few other projects and blogs for you every week.
You could for example check out Alishan and Lazaros, they’re doing a project in “Cross-Lingual Transfer Learning For Biomedical Texts” that involves using machine learning to read medical texts.

Posted on by Miguel de Oliveira Guerreiro Posted in Blogs 2021, PRACE Summmer of HPC 2021 — No Comments ↓
Me at the Barcelona Supercomputing Center, MareNostrum 4, 2019

Who am I?

Hello there! I am Miguel de Oliveira Guerreiro and I come from Lisbon, Portugal. I am currently a Lead Software Engineer and System Architect at OutSystems, a researcher at the Distributed Systems group from INESC-ID, Instituto Superior Técnico and a Visiting Researcher at the Workflows and Distributed Computing Group at the Barcelona Supercomputing Center (Computer Sciences Dept). I’ve graduated in Computer Science and Engineering from the Universidade Nova de Lisboa and the Universitat Autònoma de Barcelona. I am currently enrolled in a MSc in Computer Science at Instituto Superior Técnico with a Major in Distributed Systems.

My main research interests focus on dependable fine-grained distributed systems, fault-tolerance, efficient distributed data storage and large-scale geo-replicated systems.

I have also worked in HPC using tools such as CUDA, OpenMP and MPI, Memkind for the optimization of computational models for natural hazards risk analysis at the HPC Software Engineering Group, also at the Barcelona Supercomputing Center.

Feel free to follow the links below to know more about me and my work:

https://www.bsc.es/de-oliveira-guerreiro-miguel

http://web.tecnico.ulisboa.pt/oliveira.guerreiro/

https://www.linkedin.com/in/migueldeoliveiraguerreiro/

HPC Journey

My first contact with HPC was in 2019, when I left Portugal and moved to Barcelona to work as a researcher at the Barcelona Supercomputing Center. Professor Jordi Pons from the UAB and Dr Mauricio Hanzich from the BSC allowed me to take advantage of this huge opportunity. I started working as a research intern at the HPC Software Engineering Group developing the core of an operational system used to forecast volcanic ash dispersion in real-time. I never thought I would learn so much about volcanos while coding in C++! Since then, I’ve participated in several PRACE trainings as well as some HPC Summer Schools such as PUMPS+AI, etc. I’ve been focusing my research in cloud computing and large-scale distributed systems, however, I still keep my big passion about HPC and some of the knowledge I’ve acquired in this topic allows me to provide better insights when it comes to performance analysis and design for distributed systems.

I still remember the day I’ve seen MareNostrum 4 for the first time… really impressive!

Why SoHPC?

My motivation to join SoHPC was huge since I’ve read about the programme. The opportunity to work with institutions such as CERN, SURFSara, BSC, etc had huge impact in my decision. On top of that, I believe that the contents provided at the training week are of extreme importance nowadays for one that is developing work in big data, large-scale systems, cloud computing or HPC. I absolutely endorse and support this kind of initiatives to promote cutting-edge research and development among students and the community of future scientists. 

I’ve applied to this programme with my friend and colleague João Guerreiro, Physics Engineer. Although we are working in different project, we’ve been sharing the enthusiasm of being part of this course and also knowledge about the concepts learned.

I will be working alongside María Menéndez Herrero in Benchmarking HEP workloads from CERN in different HPC Facilities. The idea, is to understand how the High-Energy Physics workloads run in different types of hardware (CPUs/ GPGPUs) while containerised and add more of those workflows as well to the project.

Importance of HPC

HPC plays a determinant role on how science is evolving nowadays. Some mathematical and physical models just require too much computational power to provide outputs in a useful time-window if executed in regular computers. High-end clusters allow scientists to run highly complex models, and code them in a scalable and efficient way so that the computational dimension of the problems doesn’t limit the ability to perform research on such topics.

It is an honour and a pleasure to be part of something like this. Wish you all a great Summer of HPC. Hopefully by the end of the summer you will be able to surf the waves of supercomputing with way more skill and comfort.

Cheers!

Miguel

Posted on by Eoin Kearney Posted in Blogs 2021, PRACE Summmer of HPC 2021, Projects 2021 — No Comments ↓

Hello, I’m Eoin Kearney. I’m 23, Irish, and I’ve just completed my masters in chemistry in the Erastova group [website here] at the University of Edinburgh in molecular dynamics simulation. These consist of modelling uranium adsorption in swelling clay at various pH setups, using the HPC in Edinburgh, Eddie.

I’d always enjoyed computational work, but science indulged my curiosity for the wider physical world, and so I embarked on my chemistry degree in 2015. A combination of changing trends in science and education towards computers, and COVID-19, set me up for a great computational project this year. Though unexpected initially, I genuinely enjoyed it and once I had a taste was eager to take it forward and build on my experience. On top of that nuclear science has always captivated me and so when I saw PRACE had computational opportunities attached to nuclear fusion, I was sold.

My project, with another student, is titled ‘Computational atomic-scale modelling of materials for fusion reactors.’ It is in the atomistic modelling of tungsten reactor vessel walls, especially around defect sites induced under the extreme conditions of fusion. Its at a similar scale to my masters though more collision focused then on adsorption behavior.

In my free time I enjoy hiking and camping with my friends. Living in Scotland has given me a few opportunities to see some of the lovely, rugged landscape in the highlands, and I’d like to see more of the world. My next goal there is to complete the old pilgrimage route Camino de Santiago in Spain or at least to see the Pyrenees. Otherwise I enjoy reading, though my masters has restricted non-chemistry related topics so far!

So I applied to SoHPC and was luckily accepted, and now here I am. I’ve enjoyed the training week so far. As far as text editors go I’ve always defaulted to nano and its hard to get out of. Vim is intimidating but I’ve finally learned how to exit it, so that’s progress. The other perspectives of different training courses can reinforce basic key concepts.

The next step is learning the personality of the HPC cluster in the Barcelona Supercomputing Center, Mare Nostrom4 [more info here]. Its scary just how many facets I’ve skated by, like checking the available modules on the HPC! Previously I had gotten in the habit of using only those necessary and its interesting to shake up my work and see the diversity of applications these molecular modelling programs get applied to. It will be a lot to learn, but I think that’s the strongest point of PRACE; its a fast and deep dive into a complex area. It is what I will make of it.

I had some time in Scotland to see a good bit of the country, this is up near Ullapool

Posted on by Niels Hvidberg Posted in Blogs 2021 — No Comments ↓
This is me in Greenland in 2019. I was working the Ice-core project EGRIP and
was visiting the Greenland ice-cap, which can be seen in the background

Hello Everyone, friends and newcomers! My name is Niels S. Hvidberg and I am a 23 year old Physics student living in Copenhagen Denmark. This summer I will use High Performance Computing on geospatial data, which in this case will be satellite pictures from the Sentinel-2 mission. In the next section I will talk a bit about my self and the Summer of HPC program. If you already know enough about me, and just want to get in to the good stuff, feel free to scroll down to the section called “The big Two: Python and Linux”.

To the rest of you, I hope you enjoy my story none the less!

Physics… and everything else.

Many questions and ideas in physics often starts with the word: Why? Why is the sky blue? Why does the apple fall? Questions that needs answering. Naturally, as a person who is always curious and wondering, I had to study physics. Now, after completing my bachelors degree, I study Computational Physics at the Niels Bohr Institute in Copenhagen. Programming is for me just another big Why which has to be answered. It is fun and challenging and gives you the high of finally running a program that actually works (Programmers High – See Runners High).

After the summer holiday I will begin my master thesis, in which I will analyze Dansgaard-Oeschger events, found in ice-core records from Greenland. The project will involve a lot of programming, statistics and hopefully some machine learning. Geophysics is a fantastic field in physics. You have the world as you experiment, and since the field spans so much, the research field similarly spans over a wide range of methods to study the earth system.

From the Wikipedia page on Dansgaard-Oeschger events. A graph of the permille heavy water isotopes found in ice from Greenland. The Delta-oxygen-18 curve is proportional to temperature. DO-events in short, is rapid warming followed by a slow cooling. These are observed in Ice-cores from Greenland drilled as part of an ungoing project now called EastGRIP.

Participating in Summer of HPC means I will be working with an exciting project during the summer, developing my programming skills and meeting new people, even if it is online. The product of my work will be summarized in a 5 minute video, and a report together with all the 65 other participants great work.

The big Two: Python and Linux

In my project I will be working in python which means everything can be replicated by you. Along the way, the methods and programming that I will use in the project will be explained. Although python can do a lot, most supercomputers use C-programming to do high performance computing. I will be using Cython, which is a package for python that compiles and run code in C when you run the python program. I am very excited about learning Cython to utilize the power of C in python.

I will also be developing my Linux skills, since all communication to the supercomputer will be through the terminal in Linux. If you do not know what the terminal is, I can only recommend that you install Linux right now, and learn to use a computer like they did in the 90’s. Also you will get to be extra prepared for projects like this.

Friday

See you, and have a great summer!

Posted on by Joao Quintiliano Serio Guerreiro Posted in Blogs 2021 — 2 Comments ↓

Greetings everyone! I’m João Guerreiro, 24 years old from Portugal and am currently finishing my Master’s degree in Physics Engineering at Universidade Nova de Lisboa. My academic interests have always been towards the fields of Mathematics and Physics, with the addition of Computer Science when I had contact with it in college. After breaking the “barrier” of understanding the for loop and discovered the numerous applications and possibilities of Computer Science, I knew it was something that I could not leave aside from my career.

I have heard about SoHPC from a close friend of mine, Miguel Guerreiro, Computer Scientist, who brought the idea of us applying together. Needless to say that the submission of our applications didn’t take long after taking a closer look at the projects. The SoHPC programme offers a great chance to learn and work with top notch institutions in such emerging and interesting topics. I was given the opportunity to embrace in project 2127 – “Maximizing data processing efficiency in the cloud, with a twist for Research Data Management” – of which I am very excited!

With the Internet being one of the most important developments of the 20th century, of which our lives are becoming more and more impractical without it, the innovation of cloud has made its mark. It has made possible, via the Internet, the storage of content, development of apps, the test of software without having to install them in our computers and much more. The project is being organized by SURF and it will give insight on how to handle storage for data processing in scientific research and how to apply them in cloud computing. Several cloud storage systems are to be tested in assessing and finding its best use cases.

So far, the training week has passed where several very useful tools have been explored whose topics focused on Python programming for HPC and parallel programming with OpenMP and MPI. I had the chance of meeting my project mentors that immediately put me at ease. They are showing to be very helpful in getting us acquainted with the topic. I feel great motivation for the weeks ahead and I am eager to learn more about this interesting project!

I would like to congratulate all of you and wish you the best of luck in this SoHPC adventure!

Posted on by Aneta Ivaničová Posted in Blogs 2021 — 2 Comments ↓

“The only way to do great work is to love what you do. If you haven’t found it yet, keep looking. Don’t settle.”

– Steve Jobs
This is me in my Summer of HPC t-shirt.

Hello, I am Aneta, 24 years old, and currently I am an undergraduate computer science student at the Matej Bel University in Banská Bystrica. I decided to take a course in computer science three years ago. Before that, I was mostly focused on humanities, but I figured out that I am bad at memorizing and I wanted some change in my life. This decision, so far is one of the best that I have ever made. Even though, I am still not sure in what field of computer science I want to pursue a career, but I know that I am on the right path.

During my studies there were a few subjects where High Performance Computing (HPC) was discussed and I was introduced to parallel programming. I must state, that I enjoyed all of these subjects, so when one of my university teachers brought up Summer of HPC, it immediately piqued my interest. For those who are not familiar with the concept of HPC, it “refers to computing systems with extremely high computational power that are able to solve hugely complex and demanding problems”, as stated on European Commission’s webpage.

In May I found out that I was accepted by the committee of Summer of HPC and I felt great! I was selected for project number 2101, which is about the analysis of data management policies in HPC architectures. In charge of this project is the Barcelona Supercomputing Center which, needless to say, is located in Barcelona (Spain). This project is part of the MEEP project, which’s objective is to create an emulation software development platform for exascale systems. To put it simply, my job is going to be to analyze and compare ways of data movement, storage and access through the different levels of memory hierarchy in order to achieve high performance and efficiency.

Over the next two months, I hope to learn about how teams work on large projects, to gain new skills that I may put to use in the future and maybe to find some new friends.

If you enjoyed this article, please share it with your colleagues and friends. Also, do not hesitate to leave a comment below.

Posted on by Valentin Trophime Posted in Blogs 2021 — No Comments ↓

Hello there, I’m Valentin Trophime from Grenoble in France. I study computer science in Evry (in the south of Paris) at Télécom SudParis. I will graduate next year with a specialization in distributed services. More personally, in life, I like several things but the best are probably video games (Celeste, Hollow Knight, Overwatch, Super Smash Bros or Sekiro for instance) and movies (The Big Lebowski, Hot Fuzz, Matrix, Promare, The Lord of the Ring). Also I really love think about mathematical problems especially when they are directly connected with computer science. For instance I found fascinating all the questions behind decidability and computability (this one is classic but really nice). I’m curious about those and that’s why I am here in this summer of HPC program in a project of treewidth algorithms in graph theory. In this blog I will show you an introduction to this project from my perspective.

Our project is motivated by the need of simulating a quantum computer. In order to evaluate the performances of real quantum computers we need to compare them to “what we expected”. This expectation is the result of the simulation of the virtual quantum computer inside a classic computer. Simulating such a computer is done by contracting a tensor network this is basically a great multiplication of tensors of different dimensions. There are many orders possible ways to compute this multiplication and some are more efficient than others. For instance consider this simple expression with 4 vectors, these 2 ways of doing the computation don’t need the same amount of basic arithmetic operations. The first (left) order is better because it needs less computations.

Example of why order of multiplications matters.

These expressions can be represented as graphs as you can see in the previous picture. The terms (tensors) of the multiplication are nodes and their degree is equal to their dimension. For instance an isolated vertex is a scalar, a vertex with 1 neighbor is a vector, with 2 neighbors a matrix and so on… Finding the best order for contraction is equivalent to considering the line graph (i.e the reversed graph, edges become vertices and new vertices are connected if their respective edges in the original graph share one vertex) and find the approximate treewidth (definition here, but for short it’s just a number to know if our graph is similar to a tree) using the process of elimination. I know this sounds like magic, the first time but with some drawing it will make more sense. Consider this tensor network graph, first we will create its line graph.

Example of line graph.

Now it’s time to calculate its approximate treewidth with elimination, basically this process is the following: for a given order of vertices delete them in order but when you delete a vertex you connect all of his neighbors together, the approximate treewidth is the maximum degree (number of neighbors) of all nodes when they are deleted. Let’s try on our example, I will choose an arbitrary order (top to bottom, left to right) like so.

This is not rigorously the treewidth because the real one is the minimum of those approximations (for all possible orders). Testing all the orders is obviously too long in term of computation time, if the graph has N vertices, there are N! possibilities. This is known as an NP problem that’s why we usually try to compute approximation or even upper/lower bound and not the real value of the treewidth. Some approximations are based on heuristics or on complex algorithms like quickBB to guess the right order that gives the best answer. That’s why, in the beginning of our project, we will try to implement heuristics before some of the complex algorithms described in papers like this one. Our final goal is to implement those complex algorithms in a julia package with parallelization and optimization when it is possible. I hope this introduction to algorithms of treewidth in graphs, was clear. Thanks for reading, have a nice day or night, bye !

Posted on by Maria Li Lopez Bautista Posted in Blogs 2021, PRACE Summmer of HPC 2021 — No Comments ↓

Hello there! Welcome to my blog, my name is Maria Li López Bautista and I will be posting about my journey in the Summer of HPC program. 
 
With the end of the training week and the start of the projects, it’s time for the presentations. So, here we go…

Introducing me

To begin with, I’m a 20-year-old undergraduate student of theoretical physics. Currently, I’m in my third year and thinking about my final degree project, if it can be said, what a nightmare having to choose! But fortunately, I’ve been accepted to take part of this amazing summer program which is full of knowledge of the programming world and its scientific applications. So, I hope it helps me decide…fingers crossed!!!

Well, let’s leave my student sorrows behind and proceed to proper introduce myself. First of all, I was born and bred in Barcelona, a city with plenty of activities to do and never ending. If I had to describe myself with one word, it would be curious. Because since I was a child, I had the need to know beyond what it seems at first sight. Maybe I should also say that I’m a bit stubborn because once I have an idea fixed in my head I don’t stop until it comes true. Finally, about my hobbies I have to say that I love listening to music and reading science-fiction books as well as hanging out with my friends.

About the Project

SURF Research Cloud: a workspace’s components, including datasets

In the following months, my partner and I will be working on the project 2127 Maximising data processing efficiency in the cloud, with a twist for Research Data Management. As the title implies, the purpose of the project it’s to study the outcomes of different data processing techniques in the cloud to decide which one is best for each case and data type.

To sum up, we’ll learn how to handle storage for data processing in scientific research as well as how to apply them to cloud computing, and how to use Research Data Management principles in practice with real cases. Therefore, we’ll be working in the Research Data Management field and getting into the SURF Research Cloud.

(Virtual) Greetings from Barcelona! I hope to see you around here!

Me in the Summer of HPC Online Version

Kind regards,

Maria

Posted on by Carla Nicolin Schoder Posted in Blogs 2021, PRACE Summmer of HPC 2021, Projects 2021 — No Comments ↓

Hello, from Vienna! My name is Carla Nicolin Schoder,
my master thesis project in Astrophysics, at the University of Vienna is yet not over.
I study the influence of dense galaxy cluster environments on dwarf galaxy behavior,
performed with simulations on the Vienna Scientific Cluster and comparisons to Virgo galaxy cluster observations later.

Background: Owl nebula in narrow-band filter (OIII & Hα) with Vienna Little Telescope (0.8m), Credit: C.Schoder. Front: Infrared-Image of me.

That my life path brought me to Astronomy, computing and nature science,
started as a child in the countryside of Austria in the dark nights.
Even then I was a night owl, so I packed my blanket and my dog and left into the dark,
the fascination of the unbelievable vastness of space and the beauty of the stars,  ignited between me and Astronomy the spark.
 
Significantly later, programming found the way into my heart,
I still remember my tantrums with my first computer, which took half an hour to start.
I have always seen them as a tool to answer my scientific questions,
but in the meantime I discovered, that I love to lose myself in programming sessions.
 
When I started my master project a few months ago, naively thinking I am an Astronomer,
I have not yet guessed that I will be soon a supercomputer programmer.
As it should soon turn out, working with supercomputers will not be limited to my dwarf galaxy investigation,
because I was accepted for this amazing PRACE Summer of HPC, as part of the MPAS atmosphere model project, with Jonas Eschenfelder (project partner) and Evgenij Belikov (project mentor) in collaboration.

Variable resolution MPAS Voronoi mesh; Credit: https://mpas-dev.github.io/

Model Prediction Across Scales (MPAS), is able to model atmospheres globally and within small geographical regions,
to investigate for example air-quality and atmospheric chemistry, or predict hurricanes and weather over the seasons.
The unstructured spherical centroidal Voronoi meshes, though structured in vertical direction, is the special model capability,
because it allows smooth selective mesh refinements, with increased accuracy and flexibility.

The goal of the MPAS summer project is to perform this atmosphere model on ARCHER2, the new UK National Supercomputer,
and to push the MPAS model to it’s scalability limits, hopefully contributing to the models promising future.
Joyfully looking forward to join the PRACE Summer of HPC for the next weeks,
exited to work with UK Supercomputer, Atmosphere Model MPAS and collaborate with other computer geeks.

Posted on by Irem Dundar Posted in Blogs 2021 — 2 Comments ↓

Me

Hi everyone, I am Zeynep Dündar. I am a senior industrial engineering student from Koç University, double majoring with computer engineering. In this phase of my life, I am in search for my passion and I am here to discover my abilities and interest about machine learning and big data.

From My Hometown To PRACE

So, how did I get here?

I was (not) selected to SoHPC

The story of acceptance of me to Summer of HPC program was not a smooth process. From application process to acceptance, I needed to put so much effort to a research for the first time in my life. To say the truth, it was the first time that I was applying to a research program online and I was feeling incompetent to be accepted to this program. However, I screwed up my courage and fill up my application deliberately. After that it was time to wait for the results, and I waited a long time for that. Even 6 days after the due date for the official declarations for the acceptance has passed, there was no reply for my application in my mail box. My hope for getting late acceptance was diminishing day by day. At some point, I gave up waiting and confronted with the truth that I was not selected. It was not bad that much, I wasn’t selected, but at least I had an experience about applying to a research program.

After this self consoling session, I kept going with my daily life, I was searching for the other research opportunities. And on 6th of May I saw a mail in my mail box, It was from one of the members of SoHPC, urging me to approve or not my acceptance for the program. Actually I was accepted as the due date, but the mail was in spam and I did not see. That was the time I felt so much joy with me, all the deliberate efforts has resulted in a positive way, and I was accepted!

Remote Internship in SoHPC

What PRACE means for me?

As you read my lines, I wish you feel the disappointment and the joy all together. By two months I will be conducting a research about “Big data management for better electricity consumption prediction”. Even if we conduct a remote research, I am very glad to be a part of this project. I am aiming to delve into the project and interpret meaningful results and improving my interpersonal skills with my teammate, mentor and SoHPC community. If you are a researcher here like me, I would like you to remember to be grateful to be involved in this program. Or if you are looking for a research or eager to join a program, just find out your interest and search for the opportunities.

Remember, your courage is only power to lead your life!

I will be sharing my experiences here all through this programme during 2 months.

Best,

Zeynep

Posted on by Joseph Sleiman Posted in Blogs 2021, PRACE Summmer of HPC 2021 — No Comments ↓

My name is Joseph, I am 23 years old and I am a Computational Physicist studying at the University of Edinburgh, about to undertake my final Master’s year. It has not been a typical university experience with many interruptions to our learning (strikes, extreme weather, and some sort of pandemic), but my love for the logical simplicity of coding and the mesmerising beauty of Scotland has made up for these roadblocks many times over.

I happened upon PRACE in the midst of my search for internships this summer and found the vast array of projects incredibly appealing, particularly those involving Machine Learning. Having taken an introductory course in Artificial Intelligence, my interest in training models with big data to invert the problem-solving method led to my choosing of Neural Networks in Quantum Chemistry, under the stewardship of Dr. Marián Gall at the Slovak Academy of Sciences. To elaborate on the project, typically scientists discover chemicals in nature and use their known molecular structure to analyse their chemical properties with a quantum-mechanical treatment. However, this process is computationally dense and gets exponentially more expensive the more complicated your molecular structure is. Thus, with the help of artificial neural networks, we will replace the traditional analytical method with a machine learning approach using molecular descriptors via Dscribe as input. Moreover, in the name of high-performance computing (HPC), we will aim to parallelise the neural network using both GPUs and CPUs, comparing their relative speed-ups.

Learning that we would have travelled to Dublin for our training week in a “normal” year was quite the dampener as I have always longed to visit Ireland, but this disappointment was quickly swept aside by the wisdom being presented to us remotely by the Irish Centre for High-End Computing (ICHEC). In particular, the techniques to speed up Python code using tools such as Numba and Cython were especially insightful (and would have been incredibly useful a year ago when I was conducting week-long computer simulations of DNA!). Being able to exploit the power of their supercomputer, Kay, was also very exciting and provided a great opportunity to brush up on my remote desktop Linux commands via software such as Vim, Bash and Slurm.

Once again, I am incredibly eager to begin working on my project and look forward to a very enlightening Summer of HPC!

Posted on by Artem Mavliutov Posted in Blogs 2021, PRACE Summmer of HPC 2021 — 1 Comment ↓

Hey there! My name is Artem, I am a master’s student of Mathematical Engineering at the University of Padova, Italy. This year I am finishing my thesis work on parallel preconditioners, more precisely I am working on GPU implementation of FSAI preconditioner for general matrices. My story with High-Performance Computing (HPC) began with a course from my curriculum (try to guess its name..) since then I was fascinated with parallel algorithms. Moreover, I am interested in numerical methods. Good thing for me that parallel algorithms go hand in hand with numerical methods.

At the beginning of this year, I was not so sure about my future, therefore I decided to look for some opportunities and after a short search with my interests, I already bookmarked the PRACE summer school. Among many attractive projects, I found project 2122 “Numerical simulation of Boltzmann-Nordheim equation” which was in my area of interest and in correlation with my curriculum.

The Boltzmann-Norheim equation or quantum Boltzmann equation describes the time evolution of a gas. When this gas formed by bosons is cooled down to a low temperature we can observe a Bose-Einstein condensate which is one the most striking quantum phenomena in nature. It is also described as the fifth state of matter. The study of this equation is still undergoing and can bring us closer to an understanding of this quantum effect.

Since this equation is potentially written in 7D phase space (1D time + 3D space + 3D velocity), it is a challenge to come up with a numerically fast, accurate, and stable scheme that solves the problem. The purpose of this summer school is to extend the present work of KINEBEC project to a non-homogeneous case.

With the help of PRACE summer of HPC program this year I will join together with my fellow teammate David Knapp the KINEBEC project!

Posted on by Jonas Alexander Eschenfelder Posted in Blogs 2021, PRACE Summmer of HPC 2021, Projects 2021 — 1 Comment ↓
That’s me! Next to a tree, in a park

Hi! So you’re probably wondering why I, a guy who spent the last three years mostly looking at rocks, is now here playing around with the UK’s new national supercomputer? Well, I’m happy to tell you all about that…

But first a little bit about myself. My name is Jonas, I’m originally from Bavaria in Germany but am now in my third year studying geophysics at the Imperial College in London.
My interests in science lie with impact craters around the solar system, the changing environment right here on Earth and how we can use modelling to understand both better.
When I’m not fretting over school work, I like to spend my time watching fast cars go in circles in Formula 1, play Dungeons and Dragons for hours on end with my friends or go hiking if the British weather ever permits it.

As a teenager, I didn’t know what I would later want to do in life. I was interested in all sorts of things, from history and sports to science, everything seemed cool to me. But that changed when I took a year abroad to the US, where I took my first geology course. Learning how you could figure out what a place might have looked like millions of years ago, just by looking at features in the rocks there today fascinated me and I became obsessed with geology. So I applied to Imperial College London and was fortunate enough to get a place at the Earth Science and Engineering department there to pursue an MSci in Geology.
Early in first year though, my interest changed from looking at the actual rocks to the ‘bigger picture’, the Earths ocean and atmosphere interacting across the globe, entire continents moving around and giant meteor crashing into planets suddenly seemed so much more exciting and so I switched to do Geophysics.

A crater cluster on Mars. This cluster also shows signs of ice on the floor
Image from NASA/JPL/UArizona

Before university, I barely knew how to turn my laptop on and CPUs or GPUs only mattered when it came to gaming, but my new direction towards geophysics led me to learn how to. From the first course onwards I enjoyed it, especially seeing the intersection between the computing world and natural sciences excited me.
Because of this, I did a research internship at Imperial in my second year, looking at how clusters of impact craters form from larger meteors fragmenting in Mars’ atmosphere. Seeing how the models influenced our search for patterns and vice versa was exciting and I wanted to learn more about modelling in and of itself. So for next year, my master thesis is going to model how the chemistry in rivers changes across the Clyde Basin in Scotland with the hopes of being able to figure out where pollutants are introduced into the water.

Now you might ask yourself how all of this leads to PRACE and the Summer of HPC and that is a good question. I found out about this program through a professor at university and at first was hesitant to apply since it seemed way too advanced for me. But the projects sounded interesting and so I applied afterall in the hopes to be able to challenge myself, learn about the pinnacle of computing power and understand the background of how some of these large models actually work that I so often just read the results from.
I got accepted and will work with Carla Schoder on the MPAS atmosphere model at the EPCC. Sadly we won’t be able to taste test a deep-fried mars bar in Edinburgh, as the entire program iis remote. But I’m still excited about the project and to work with new and interesting people.

So what’s next for me? Well, last week we finished an intense training week and now I actually know what a supercomputer is and how to operate one. We will work together to figure out how to best test MPAS and then get to play around with it on ARCHER2, the UK’s new national supercomputer. I will of course keep you updated here with all the cool stuff we’ll be doing and I’m learning, so stay tuned.

Posted on by Marc Tunica Rosich Posted in Blogs 2021, PRACE Summmer of HPC 2021, Projects 2021 — No Comments ↓

Hello everybody! My name is Marc  and I am writing from the mediterranean city of Barcelona, land of several personalities such as Antoni Gaudí, Mercè Rodoreda and Leo Messi.  I was born in this city 23 years ago. Currently, I am studying physics and maths at the University of Barcelona.  

About Me

I would describe myself as an active person, maybe a little bit impatient too. I am passionate for science and maths. I am interested in the huge number of opportunities that computer science brings us to study physic systems.  I also play guitar (extremely bad, actually), particularly, I enjoy playing jazz. If I have to choose a film, it is Midnight in Paris by Allen, a book, 1984 by Orwell and a song,  Isadora by Christian Scott.

About Summer of HPC

Regarding my virtual stage in the Summer of HPC, I am involved in the 2120 High Performance Quantum Fields program. Specifically, I am beginning a project about Carbon Nanostructures (CNS). The main goal of this project is to collaborate with the CNS group from the Jülich Supercomputing Centre, performing and creating a C++ library and an algorithm to model CNS using fermionic matrixes. I will learn and try to post about C++,  OpenMP,  GPU programming and, in general, scientific program design.  

Hope to know more about all of you. And if you have planned to visit Barcelona do not hesitate to contact me: I have a list of the best restaurants!! 

Posted on by Miriam Beddig Posted in Blogs 2021, PRACE Summmer of HPC 2021, Projects 2021 — No Comments ↓

Hi all,
this summer I am doing the “Summer of HPC” programme. In this blog, I want to share my experiences and thoughts with you. But first, let me introduce myself:

Introduction

Handstand with new T-Shirt
Me, doing a handstand with the new “Summer of HPC” T-shirt

My name is Miriam and I am 28 years old. Until 2018 I have studied Mathematics. After that, I started working as a software engineer. While I enjoyed the programming part of my work, I wanted to learn more about the theory and applications of Computer Science. Therefore, I decided to do my Masters in Computer Science which I am currently pursuing at the University of Edinburgh.

In my leisure time, I love doing sports, gardening, cuddling my cat and baking cakes.

Motivation

High-performance computing (HPC) combines my interests in Computer Science and Natural Sciences due to its wide range of applications. What I particularly like about HPC is that it is often necessary to use advanced programming and software engineering techniques and that it is therefore closely related to research.

The Summer of HPC programme is a combination of HPC training and practical work. I am very enthusiastic about increasing my knowledge in high-performance computing and getting practical experience in applying this knowledge to real-world problems. Therefore, when I initially heard from the Summer of HPC programme I knew immediately that I want to apply for this programme. I am very pleased that my application was successful and I am very excited about what I will learn during this summer.

First training week

The first training week covered the usage of Python in HPC, OpenMP, and MPI. What I enjoyed most during this week was the mixture of theory and practice. In this way, the learning outcome was accordingly high.

From next week on, I will be working on the project “HPC Implementation of Molecular Surfaces”, organised by the VSC Research Center in Vienna (https://summerofhpc.prace-ri.eu/hpc-implementation-of-molecular-surfaces/).
I am looking forward to all the experience and insights I will gain during the next two months!

Finally – a small quiz for you

To keep my brain working during the first week, I definitely needed enough cake!
So that not only I benefit from this cake, I implemented a small rebus (https://en.wikipedia.org/wiki/Rebus) for you on it (see the picture below).

  • Foto of the workplace
    Foto of my workplace
  • Marzipan-Nut-Cake. Can you guess the phrase in the Rebus?


Can you guess what I have “written” on the cake? You are welcome to comment below if you found a solution. If it is too difficult I can also provide some tips.

Posted on by Jakub Raczyński Posted in Blogs 2021 — No Comments ↓

Hello everyone! My Name is Jakub, I am 23 years old and at the moment I am pursuing a Master’s degree in artificial intelligence at Poznań University Of Technology (Poland). Earlier this year, I graduated with Bachelor’s degree in computer science at the same university.

Few words about me

Besides studying, I like to play competitive video games and solve different kind of puzzles, like Sudoku, Kakuro, Nonograms and all stuff similar to those, especially if it’s somehow related to math or numbers. I also like to go for walks and doing some casual sports like table tennis or volleyball (the only sport I train regularly though is running for trains and buses as I am constantly late for those). In last 18 months I also spent big chunk of my leisure time on analysing data and reading news about COVID-19.

To be honest, I was very lucky to get a chance to apply to Summer of HPC programme, as I was saved by extended application deadline as I was informed about the programme by my new university mentor with whom I work since April, who by the way was one of the participants in year 2016.

Why Summer of HPC?

I decided to apply to Summer of HPC programme, because I was looking for a chance to gain some experience in cooperation on international projects. That was my main motivation but actually Summer of HPC programme has a lot more to offer, even despite the fact we are working remotely because of the pandemic.
First of all, I will have a chance to work on HPC cluster, which is absolutely stunning and still unbelievable for me as I always have been passionate about working with computer clusters which can offer computing power which is unimaginable for someone like me, who have used just his personal computer throughout his life.
Secondly, there was a wide variety of projects to choose – actually there were 33 of them. That’s why I could choose the one which was the most appealing to me – as I study artificial intelligence, I chose one focused on building machine learning applications which are resilient to errors. Moreover, if we will have enough time, we are going to optimise created applications, so we will be able to deploy them on edge devices like mobile phones, which obviously don’t have as much computational power as supercomputers.

My teammate, mentors and workplace

I will work in cooperation with Enes Erciyes, with whom I have already talked a few times and he seems to be a great guy, so I recommend you to check out his introductory post, too. We will be mentored by Leonardo Bautista Gomez and Albert Njoroge Kahira from Barcelona Supercomputing Center (BSC). Supercomputer infrastructure is also provided by BSC – we will work on MareNostrum supercomputer which in June 2021 was ranked as 63rd most powerful supercomputer in the world according to the TOP500 ranking. Most importantly we will have access to Power9 GPU cluster, which is especially useful in training deep learning models.

Wrap-up

I am pretty excited and can’t wait to start the work, especially after great training week organised by Irish Centre for High-End Computing staff during which I gained a lot of theoretical and practical knowledge on HPC.

Thanks for getting to this point, hopefully you have found this post interesting! If this is the case, I invite you to check out updates of this blog, which will be posted in upcoming weeks. Moreover, if you want to ask me any questions, feel free to do so in comment section below.

Posted on by Athanasios Kastoras Posted in Blogs 2021, PRACE Summmer of HPC 2021, Projects 2021 — 6 Comments ↓

Introduction

Hello everyone!
My name is Athanasios Kastoras (but I go by Thanos) and I am really excited about the opportunity I’ve been given, to participate in this year’s PRACE Summer of HPC. I will be working on Precision based differential checkpointing for HPC applications with the Barcelona Supercomputing Center, but more about that later.

Who am I?

I was born on October 19th, 2001, in the city of Volos, Greece, where I was also raised, and now study. I loved maths from a young age, and I was curious about science, but I never got in touch with Computer Science and Programming until I went to University. When I entered the Electrical and Computer Engineering Department at the University of Thessaly, I was seduced by the C programming language and later by the amazing hardware and computer architecture fields. Now I’ve just finished the second year of my studies and I’m already passionate about computational speed, software optimization, and HPC system programming. This is why I applied to the PRACE SoHPC program; because I believe this is the best starting point to begin my journey to the limitless world of high-end computing.

If you think you can do a thing or think you can’t do a thing, you’re right.

Henry Ford

Some facts about me

  • I love coding! Whether we talk about scripting or low-level programming, I can look at lines of code for hours and never be bored.
  • I am a Linux enthusiast! In the last two years, I’ve switched between Ubuntu, Kubuntu, Fedora, CentOS, and Manjaro Linux distributions and I always love to experiment with new Operating Systems on my personal laptop.
  • I practice Karate! Since the age of seven years old, I’ve been practicing Shotokan Karate. Until now, I’ve gained a black belt with a second Dan, many experiences, and exciting skills.
  • I love camping and nature! I always look for a chance to spend time in nature and relax away from the fast city life.

My experience in SoHPC so far

Today, I’ve just finished the SoHPC training week. We learned many interesting things during the whole week, about the Python language in HPC, parallelizing C and Fortran code using OpenMP and MPI, and more importantly, we run experiment programs on an actual supercomputer. It is amazing to know that the code you wrote on your personal laptop is running on a supercomputer in Ireland just with the press of a button (actually the button was a script, but anyway).

This is the Barcelona supercomputing center. I recently learned that it is built inside a church. Isn’t it so cool? Unfortunately, the program is remote, but this place definitely goes on my travel bucket list!

The project I will be working on

During the summer I will be working on a project called precision-based differential checkpointing for HPC applications. I will be dealing with FTI, a library that helps programmers make HPC applications less vulnerable using four levels of checkpointing (i.e. storing the state of the execution to be restored in case of an error). Levels that cover a larger amount of errors are more expensive than those that cover more simple errors, thus it is necessary to find the best balance between them. Our goal will be to implement differential checkpointing in FTI and explore precision-based differential checkpointing. Sounds confusing? During the following blog posts, I will be presenting the project in more detail, but no prior knowledge is needed to understand it.

Finally

I am really excited about this summer and so far it is going really well! Unfortunately, we weren’t able to travel to the sites we will be working with, but I will try to make the best out of this experience. Do you have any experiences like this? If yes, feel free to share your experience or anything you want to discuss in the comment section.

Finally, I hope you enjoyed my post and found it interesting. More posts are coming, so please follow me on my LinkedIn account if you want to be updated.

Thanos

Facebook
Twitter
Follow by Email