Course: AI for Autonomous Particle Accelerators

Collapse all Expand all

General

Select activity Ankündigungen

Ankündigungen Forum

What this it about?

Select activity Content of this course

Imagine stepping into the control room of a cutting-edge particle accelerator, the hum of sophisticated machinery filling the air. As a Scientific Operator, you are tasked with steering and focusing charged particles onto a diagnostic screen to prepare the beam for experimental research. Welcome to this immersive online course that places you at the helm of a virtual particle accelerator, guiding you through the exhilarating process of preparing a beam, steering it with precision, and setting the stage for groundbreaking experiments.

In this course, you’ll dive into the heart of particle accelerators—marvels of modern science that have shaped our understanding of matter since the 1920s. From the particle gun that sparks the beam to the quadrupole and dipole magnets that sculpt its path, you’ll master the key components and their roles in generating, accelerating, and controlling particle beams. Through a hands-on virtual control panel, you’ll adjust magnet strength and steering angles to focus the beam, watching real-time feedback on a diagnostic screen as a graph reveals its position and sharpness. This isn’t just theory—it’s a front-row seat to the precision that drives discoveries like the Higgs boson particle or advancements in RNA vaccine research at DESY’s PETRA III.

You’ll explore how these systems enable transformative research in physics, material science, and beyond, while uncovering the role of artificial intelligence in revolutionizing accelerator operations. AI techniques, such as reinforcement learning, are significantly reducing tuning times—once a time-consuming task for human experts—allowing scientists to focus on pushing the boundaries of discovery. By experimenting with a simulated accelerator inspired by DESY’s ARES linac, you’ll tweak parameters, measure outcomes, and reflect on how each adjustment shapes the beam’s journey, mirroring real-world challenges faced by accelerator scientists.

Fascinating discoveries

Select activity Research with particle accelerators

The first particle accelerators were built in the 1920s and 1930s. Their primary purpose was to investigate the structure of matter. Physicists want to find out, with the help of particle accelerators, "what holds the world together at its core." For this, charged particles (electrons, ions, or protons) are accelerated. By bombarding matter with these particles, one can understand its structure. Nowadays, particle accelerators exist in all possible sizes (from a few centimeters up to several tens of kilometers, linear or circular). The Deutsches Elektronensynchrotron (DESY) is one of the world's leading accelerator centers, and DESY's large accelerator X-ray sources are internationally sought after for a wide range of applications: PETRA III is one of the world's best storage ring for producing X-rays. FLASH delivers ultra-short pulses of "soft" X-rays and enables unique experiments. And the European XFEL, with its 3.4 km, is the longest linear electron accelerator, producing the most intense X-ray pulses ever. More information is available on the DESY homepage.

Since their invention, particle accelerators have had a profound impact on the advancement of human knowledge through their groundbreaking scientific discoveries. A prominent example is the discovery of the Higgs boson at the Large Hadron Collider (LHC) at CERN in 2012; the empirical evidence for the existence of the Higgs field provided a fundamental component of the Standard Model of particle physics and was awarded the Nobel Prize in Physics in 2013. Accelerators are indispensable not only for particle physics but also find applications in many other fields of science and technology, such as X-ray imaging, spectroscopy, and the study of ultrafast processes at atomic and molecular levels. A recent prominent example is coronavirus research: X-ray sources like those at DESY were perfect tools to investigate the structure, dynamics, and function of SARS-CoV-2, aiding in the development of both drugs and vaccines against COVID-19.

A journey through an accelerator

Select activity Start the journey

Next, we take you on a journey through an accelerator in operation (in reality, better don't do this due to radiation). You will see a linear accelerator for electrons, inspired by the Accelerator Research Experiment at SINBAD (ARES), a linear electron beam accelerator at DESY. You can explore the accelerator from all sides. Click through the individual components to learn more.
Select activity Discover the accelerator

Be an operator

Select activity And was this difficult?

The 3D model shown above is of course a simplified representation of a particle accelerator and particle beam. Real accelerators are significantly larger and have vastly more components. Where the visualization shows only five magnets, a real accelerator might have hundreds or even thousands of magnets. Combined with other components, it is not uncommon for these facilties to have tens if not hundreds of thousands of control parameters. At the same time the charged particle beam in reality is so small that it is invisible to the naked eye. Using this vast number of control parameters, this tiny particle beam needs to be controlled over multiple kilometers of beam pipe down to precisions of a few tens of micrometers. For reference that is about the same as throwning a pea from Hamburg to Munich, hoping to hit a target the size of another pea. What is more, users of particle accelerators continue to ask for ever more demanding experimental setups, requiring even more precise control of the particle beam.

AI for accelerators

Select activity AI assistant

While today, expert human operators spend over 2000 hours per year tuning these control parameters, AI and machine learning methods are now being explored to help enable the demands of novel experiments, while reducing tuning times and making more time available for actual scientific research.

One approach explored at DESY is the so-called Reinforcement Learning, where the tuning of the accelerator is formulated as a game in which the AI agent is rewarded for achieving the desired beam properties as quickly as possible, while being penalized for entering unsafe operating conditions. This method has so far successfully been tested on a tuning task involving an accelerator section of three quadrupole magnets and two dipole magnets, very similar to the one shown in the visualization. Here, the AI agent was able to learn how to attain a desired position and focus of the beam visible on the downstream diagnostic screen in a fraction of the time it would take a human operator. However, like real human experts (probably even worse than them), the AI agent needs a lot of time to learn how to control the beam in the first place. In the mentioned example, the AI agent actually needed about 3 years of non-stop trial and error on the accelerator to learn how to control the beam better than the human experts. This much time simply is not available in a real accelerator, where beam time is a very precious resource. Instead, the AI agent was able to learn of fourty parallel simulations of the accelerator, where in each simulation was sped up by five orders of magnitude. This way it is possible for the AI agent to learn in just under an hour, allthe while using no time on the real accelerator at all.

Conclusion

Particle accelerators are of great benefit to our society in the field of basic research, but also for very current and practical challenges. Particle accelerators are highly complex large-scale devices that are operated by experts with many years of experience. Research is currently being conducted into AI support for operation in order to achieve better precision and reliability, thereby increasing the performance of accelerator facilities and enabling new scientific findings.

Topic outline