Test4Theory Tops 4 Trillion Events

March 14 2018.

LHC@home's Theory application today passed the milestone of 4 TRILLION simulated events. This project, under its earlier name "Test4Theory", began production in 2011 and was the first BOINC project in the world to use Virtual Machine technology (based on CERN's CernVM system). The photo below, taken at the CERN Computing Centre, shows some of the people who are currently involved in the project:

From left to right: Nikolas Giannakis (CERN), Nils Høimyr (CERN), Laurence Field (CERN), Anton Karneyeu (CERN & INR Moscow), Ben Segal (CERN), Peter Skands (Monash U.)

Peter Skands: Test4Theory is a great illustration of the unique kinds of projects that can happen spontaneously at a place like CERN. In 2010, I had just arrived as a new staff member in the Theory Division at CERN. My research is on "virtual colliders" - essentially computer simulations of the physics that takes place in colliders like the Large Hadron Collider. A colleague introduced me to Ben Segal (also in the photo), who represented a small development team that was looking for scientific use cases for testing a new concept: volunteer computing based on virtual machines. This was a perfect match for me. As a small group of theorists, we did not have access to the same kinds of large-scale computing resources that our experimental colleagues can boast of, so large-scale tests and validations of the type that we can now run in Test4Theory were unheard of at that time. I had not previously considered volunteer computing, as the kinds of scientific software we write are almost exclusively developed and run on Linux operating systems, while most volunteers would have Windows or Mac operating systems. Ben's suggestion of encapsulating our simulations inside a virtual Linux machine that could then be sent to run on any architecture, seemed extremely elegant. We immediately began collaborating, and when Anton Karneyeu (also in the photo) joined the project, things really started moving. Alpha testing started in late 2010 and half a year later the system operated smoothly and continuously with about 100 machines connected from around the globe. Approximately 5 billion collider events had been generated during this testing stage. Beta testing then began, focusing on the scalability of the system and brought the total number of connected machines up to several thousand by 2012. Since then, the system has matured and grown further, allowing us to expand the amount of comparisons we can do between theory and data. These comparisons are at the heart of the project, with results available openly at the mcplots.cern.ch web site, which is used by a wide range of both theorists and experimentalists as a database of validations of the simulations against real-world data. Although I am now based at Monash University (in Melbourne, Australia), I continue to be actively involved in Test4Theory, and mcplots remains the go-to resource for getting a comprehensive view of how well different types of physics models, different levels of approximation, and different parameter settings, compare against the many benchmark measurements that have been performed not just at LHC but at many older collider experiments as well. We are hugely indebted both to the volunteers who are making this possible, and to the software development and IT teams at CERN who helped to transform the project from idea to reality.

Ben Segal: I managed the creation of CERN’s first BOINC project, LHC@home / Sixtrack in 2004. From 2008, I coordinated the BOINC / CernVM system which allowed CERN physics applications to run on LHC@home, starting with the Test4Theory project in 2011 and later the ATLAS, CMS and LHCb experiments. I like to recall that much of the work to create LHC@home since 2004 was done by volunteers, in the spirit of BOINC itself - either by “retirees” like myself and Eric McIntosh, or by CERN staff in “overtime” mode, or by several tens of students, Fellows and visitors who have contributed to the project over the years.

Anton Karneyeu: Joined the MCPLOTS project in 2010 to put a structure in the chaos of various high energy physics simulation codes. Developed tools to run the codes through Test4Theory and web interface to browse the results.

Laurence Field: With a long experience from Grid computing, Laurence lauched the CMS pilot on LHC@home and later joined the LHC@home core team in 2016 to develop and support the daily operations of the volunteer computing service. Developed a sustainable platform for VM applications and managed the service consolidation with BOINC and HTCondor.

Nils Høimyr: I got involved with LHC@home in 2011, when I was asked to host the infrastructure for Test4Theory and LHC@home in my section.  Thanks to the hard work of an enthusiastic team, Test4Theory successfully deployed virtualization with volunteers computing. LHC@home had been running the accelerator physics application Sixtrack since 2004, and could now also host other physics applications running in the CernVM virtual machine. We also got the LHC experiments on board, and the challenges and opportunities offered by volunteer computing made me take on the role as LHC@home project leader. Thanks to the efforts of volunteers, the scientific community at CERN now gets a lot more simulations done than what would be possible on our local computing facilitties, that are fully dedicated to LHC data processing.  It is a pleasure to work with the scientists providing the simulations, as well as to interact with other volunteer computing projects using the BOINC platform. We would like to warmly thank all the volunteers who contribute to LHC@home, and in particular those who help us to debug and beta-test applications.

Nikolas Giannakis: I joined the LHC@home team in 2017 as a graduate student, and my development includes a prototype WebRTC client for BOINC. Being able to work for a project like LHC@home gave me a vision of what Computer Science should be all about. It should be about cooperation, generosity and striving to achieve greatness. Within the great laboratory that is CERN this project was born to give meaning to the saying that the journey is more important than the destination. The 4th trillionth event computed for Test4Theory is a prime example. The computing of these events brought together people from very different teams and most importantly gave people with a passion for science a chance to contribute to the biggest experiment ever done.