The commissioning and operation of future large-scale scientific experiments will challenge current tuning and control methods. Reinforcement learning (RL) algorithms are a promising solution thanks to their capability of autonomously tackling a control problem based on a task parameterized by a reward function. The conventionally utilized machine learning (ML) libraries are not intended for microsecond latency applications, as they mostly optimize for throughput performance. On the other hand, most of the programmable logic implementations are meant for computation acceleration, not being intended to work in a real-time environment. To overcome these limitations of current implementations, RL needs to be deployed on-the-edge, i.e. on to the device gathering the training data. In this paper we present the design and deployment of an experience accumulator system in a particle accelerator. In this system deep-RL algorithms run using hardware acceleration and act within a few microseconds, enabling the use of RL for control of ultra-fast phenomena. The training is performed offline to reduce the number of operations carried out on the acceleration hardware. The proposed architecture was tested in real experimental conditions at the Karlsruhe research accelerator (KARA), serving also as a synchrotron light source, where the system was used to control induced horizontal betatron oscillations in real-time. The results showed a performance comparable to the commercial feedback system available at the accelerator, proving the viability and potential of this approach. Due to the self-learning and reconfiguration capability of this implementation, its seamless application to other control problems is possible. Applications range from particle accelerators to large-scale research and industrial facilities.

Accurate knowledge of wake functions is crucial in accelerator physics, serving as the cornerstone for understanding intra-bunch interactions and for controlling or mitigating instabilities that limit accelerator performance. Haissinski distributions, which describe the steady-state longitudinal bunch density, are intrinsically determined by the wake function experienced by the bunch. While these distributions are typically computed from a given wake function, we investigate the inverse problem: extracting the wake function directly from measured Haissinski distributions. In this theoretical work, we introduce a novel method to reconstruct wake functions by utilizing Haissinski distributions obtained at multiple bunch charges. By combining these profiles into an overdetermined system, we address challenges posed by the inverse problem, which is sensitive to noise and discretization errors. Here, our preliminary results suggest that the use of regularization techniques may help achieve more stable reconstructions of the wake function.

The Karlsruhe Institute of Technology is currently exploring a compact method of longitudinal electron bunch diagnostics with femtosecond resolution that has recently been demonstrated for other parameter ranges. The experimental setup utilizes a THz-based streaking approach with resonator structures, achieving both high compactness and efficiency. In this paper, we report on the experimental observation of streaking signals with our Compact Transverse Deflecting System, which has been successfully tested using two different resonators, an Inverse Split-Ring Resonator and a Tilted-Slit-Resonator.

The Karlsruhe Institute of Technology operates the accelerator test facility Karlsruhe Research Accelerator, which also provides synchrotron radiation at 2.5 GeV. Roughly one third of the wall-plug power is used for cooling. Optimizing the infrastructure for cooling has a huge impact on the overall sustainability. To reduce the environmental impact a thermal well system was installed. It reduces the base heat load by eliminating one of three 500 kW cooling units. This paper describes the challenges, such as iron-manganese rich groundwater, and their solution for our 1 MW passive cooling system. The average energy consumption of 28 kW for the thermal well infrastructure is compensated by a new 540 kWp solar power plant. This paper elaborates on the commissioning of the wells and shows the first results of this overall sustainable cooling concept.

This theoretical study presents an advanced method for longitudinal phase space tomography in electron storage rings, focusing on reconstructing phase space densities from electro-optical spectral decoding (EOSD) measurements that incorporate crystal geometry effects. The EOSD crystal geometry significantly impacts the measurement signal due to signal integration along its length and interference from wake fields and Cherenkov diffraction radiation (ChDR). These effects add challenges to reconstructing the original phase space density from experimental data. To address these challenges, we integrate two theoretical frameworks. First, we employ the Vlasov-Fokker-Planck equation to model the turn-by-turn evolution of the charge density distribution. Second, CST simulations of the bunch profile characterize the electric field inside the crystal, enabling a tailored simulation for the EOSD system at the Karlsruhe Research Accelerator (KARA). By combining these approaches, we propose a refined tomography method that more accurately reconstructs the longitudinal phase space from sensor data, effectively capturing the interplay between bunch dynamics and the EOSD system configuration.

The future circular electron-positron collider (FCC-ee) is designed for highest luminosity to enhance the precision of high-energy particle physics experiments, spanning energies from the Z pole to the $\text{t}\bar{\text{t}}$ threshold. As outlined in its conceptual design report, high-precision measurements of the longitudinal bunch profile are required across multiple operation modes, which presents key challenges for beam instrumentation. As part of the feasibility study, a concept for an electro-optical (EO) bunch profile monitor has been developed to address these challenges, building on the existing EO beam diagnostic at the Karlsruhe Research Accelerator (KARA) at KIT. The first EO monitor prototype for FCC-ee features a novel crystal-holder design using prisms, enabling a single-pass setup crucial for measuring the long bunches during Z operation. This contribution presents the first measurement results of the EO monitor prototype for FCC-ee, which were obtained in the in-air test stand at the CERN Linear Electron Accelerator for Research (CLEAR).

Tuning injectors is a challenging task for the operation of accelerator facilities and synchrotron light sources, particularly during the commissioning phase. Efficient tuning of the transfer line is essential for ensuring optimal beam transport and injection efficiency. This process is further complicated by challenges such as beam misalignment in quadrupole magnets, which can degrade beam quality and disrupt operations. Traditional tuning methods are often time-consuming and insufficient for addressing the complexities of high-dimensional parameter spaces. In this work, we explore the use of advanced AI methods, including Bayesian optimization, to automate and improve the tuning process. Initial results, demonstrated on the transfer line of KARA (Karlsruhe Research Accelerator) at KIT (Karlsruhe Institute of Technology), show promising improvements in beam alignment and transport efficiency, representing first steps toward more efficient and reliable accelerator operation. This study is part of the RF2.0 project, funded by the Horizon Europe program of the European Commission, which focuses on advancing energy-efficient solutions for particle accelerators.

FLUTE is a compact and flexible linac-based accelerator test facility at the Karlsruhe Institute of Technology (KIT) in Germany. It serves as a platform for a variety of accelerator studies and to generate intense short THz pulses for various photon science experiments. Later, FLUTE will be also used as an injector of sub-100 fs bunches into the VLA-cSR (Very Large Acceptance compact Storage Ring), which is part of the cSTART (compact STorage Ring for Accelerator Research and Technology) project currently in the technical design phase at KIT. Recently, FLUTE's high energy section has been installed and commissioned. This includes the linac, bringing FLUTE beyond 5 MeV to full energy, the bunch compressor, and two corresponding diagnostics sections. A metal foil in the last diagnostics section has been used to generate a first THz signal with high energy electron bunches.

Abstract Online tuning of particle accelerators is a complex optimisation problem that continues to require manual intervention by experienced human operators. Autonomous tuning is a rapidly expanding field of research, where learning-based methods like Bayesian optimisation (BO) hold great promise in improving plant performance and reducing tuning times. At the same time, reinforcement learning (RL) is a capable method of learning intelligent controllers, and recent work shows that RL can also be used to train domain-specialised optimisers in so-called reinforcement learning-trained optimisation (RLO). In parallel efforts, both algorithms have found successful adoption in particle accelerator tuning. Here we present a comparative case study, assessing the performance of both algorithms while providing a nuanced analysis of the merits and the practical challenges involved in deploying them to real-world facilities. Our results will help practitioners choose a suitable learning-based tuning algorithm for their tuning tasks, accelerating the adoption of autonomous tuning algorithms, ultimately improving the availability of particle accelerators and pushing their operational limits.

Glass fiber sheet molding compounds are a popular fiber-reinforced polymer because of their good processability. Both fiber mass fraction (FMF) and fiber orientation distribution significantly determine the mechanical properties of manufactured components. Local random variations in both properties occur during prepreg production and molding. This contribution expands a non-destructive measurement model for local FMF using Terahertz spectroscopy towards multiple specimen thicknesses. A thorough measurement uncertainty analysis according to the Guide to the Expression of Uncertainty in Measurement (GUM) shows an uncertainty reduction by 70 % compared to the uncertainty resulting from the manufacturing process. Measurement times of 0.5 s qualify the methodology for an in-line application. Integrating the measurement results into a parametrized, component -specific finite element model reduces the root mean squared error between physical experiments and simulations by 33 and 54 % compared to simulations with nominal fiber mass fractions.

Online tuning of real-world plants is a complex optimisation problem that continues to require manual intervention by experienced human operators. Autonomous tuning is a rapidly expanding field of research, where learning-based methods, such as Reinforcement Learning-trained Optimisation (RLO) and Bayesian optimisation (BO), hold great promise for achieving outstanding plant performance and reducing tuning times. Which algorithm to choose in different scenarios, however, remains an open question. Here we present a comparative study using a routine task in a real particle accelerator as an example, showing that RLO generally outperforms BO, but is not always the best choice. Based on the study's results, we provide a clear set of criteria to guide the choice of algorithm for a given tuning task. These can ease the adoption of learning-based autonomous tuning solutions to the operation of complex real-world plants, ultimately improving the availability and pushing the limits of operability of these facilities, thereby enabling scientific and engineering advancements.

The description of physical processes with many-particle systems is a key approach to the modeling of numerous physical systems. For example in storage rings, where ultrarelativistic particles are agglomerated in dense bunches, the modeling and measurement of their phase-space distribution is of paramount importance: at any time the phase-space distribution not only determines the complete space-time evolution but also provides fundamental performance characteristics for storage ring operation. Here, we demonstrate a non-destructive tomographic imaging technique for the 2D longitudinal phase-space distribution of ultrarelativistic electron bunches. For this purpose, we utilize a unique setup, which streams turn-by-turn near-field measurements of bunch profiles at MHz repetition rates. To demonstrate the feasibility of our method, we induce a non-equilibrium state and show that the phase-space distribution microstructuring as well as the phase-space distribution dynamics can be observed in great detail. Our approach offers a pathway to control ultrashort bunches and supports, as one example, the development of compact accelerators with low energy footprints.

The detection of rapid dynamics in diverse physical systems is traditionally very difficult and strongly dominated by several noise contributions. Laser mode-locking, electron bunches in accelerators, and optical-triggered phases in materials are events that carry important information about the system from which they emerge. By detecting single-shot spectra with high repetition rates over long-time scales, new possibilities and applications to diagnose, control and tailor the spectral dynamics of lasers and electron beams in synchrotron and free-electron laser (FEL) accelerators open up. This contribution focuses on the latest developments of real-time, single-shot, high-repetition-rate detectors and data acquisition systems, with a special focus on emerging technologies and new possibilities in the diagnostics of rogue optical signals.

At the KIT electron storage ring KARA (Karlsruhe Research Accelerator) an electro-optical (EO) near-field monitor is in operation performing single-shot, turn-by-turn measurements of the longitudinal bunch profile using electro-optical spectral decoding (EOSD). In context of the Future Circular Collider Innovation Study (FCCIS), a similar setup is investigated with the aim to monitor the longitudinal bunch profile of each bunch for dedicated top-up injection at the future electron-positron collider FCC-ee. This contribution presents the status of a monitor design adapted to cope with the high-current and high-energy lepton beams foreseen at FCC-ee.

The detection of rapid dynamics in diverse physical systems is traditionally very difficult and strongly dominated by several noise contributions. Laser mode-locking, electron bunches in accelerators and optical-triggered phases in materials are events that carry important information about the system from which they emerge. To understand the underlying dynamics of complex systems often large numbers of single-shot measurements must be acquired continuously over a long time with extremely high temporal resolution. Ultrafast real-time instruments allow the acquisition of large data sets, even for rare events, in a relatively short time period. The real-time measurement of fast single-shot events with large record lengths is one of the most challenging problems in the fields of instrumentation and measurement. In this contribution, the novel ultra-fast and continuous data sampling system THERESA using photonic time-stretch is presented and its performance is discussed. The proposed data acquisition system is based on the latest ZYNQ Radio Frequency System on Chip (ZYNQ-RFSoC) family from Xilinx, which combines an array of fast (GS/s) multi-channel Analog-to-Digital Converters (ADCs) with a Field Programmable Gate Array (FPGA) and a multi-core ARM processor in a single heterogeneous programmable device. The stretched pulse is sampled in parallel by 16 wideband sampling channels operating in time-interleaving mode. The sampled data is transferred by a 100 Gb Ethernet data link to the Data Acquisition (DAQ) compute node for further analysis. The combination of both, the photonic time-stretch and the fast sampling system, is capable of sampling short pulses with femtosecond time resolution. Applications of the new system, hardware implementation and the commissioning of the first system for the electron bunch diagnostics are presented.

Sweat is one of the essential biofluids produced by the human body, and it contains various physiological biomarkers. These biomarkers can indicate human health conditions such as disease and illness. In particular, imbalances in the concentration of electrolytes can indicate the onset of disease. These same imbalances affect the dielectric properties of sweat. In this study, we used attenuated total reflection terahertz time domain spectroscopy to obtain the frequency-dependent dielectric properties of human sweat in a frequency range from 200 GHz to 2.5 THz. We have investigated the variation of dielectric properties of sweat collected from different regions of the human body, and we have observed that the real and imaginary part of dielectric permittivity decreases with the increase in frequency. A combination of left-hand Jonscher and Havriliak-Negami processes is used to model the results and reveal the presence of relaxation processes related to sodium and calcium ions concentrations. This information may help design novel biosensors to understand the human health condition and provide a hydration assessment.

To understand the underlying complex beam dynamics in electron storage rings often large numbers of single-shot measurements must be acquired continuously over a long period of time with extremely high temporal resolution. Photonic time-stretch is a measurement method that is able to overcome speed limitations of conventional digitizers and en-able continuous ultra-fast single-shot terahertz spectroscopy with rates of trillions of consecutive frames. In this con-tribution, a novel ultra-fast data sampling system based on photonic time-stretch is presented and the performance is discussed. THERESA (TeraHErtz REadout SAmpling) is a data acquisition system based on the recent ZYNQ-RFSoC family. THERESA has been developed with an analog bandwidth of up to 20 GHz and a sampling rate of up to 90 GS s −1 . When combined with the photonic time-stretch setup, the system will be able to sample a THz signal with an unprecedented frame rate of 8 Tf s −1 . Continuous acquisition for long observation times will open up new possibilities in the detection of rare events in accelerator physics.

The laser-driven generation of relativistic electron beams in plasma and their acceleration to high energies with GV/m-gradients has been successfully demonstrated. Now, it is time to focus on the application of laser-plasma accelerated (LPA) beams. The "Accelerator Technology HElmholtz iNfrAstructure" (ATHENA) of the Helmholtz Association fosters innovative particle accelerators and high-power laser technology. As part of the ATHENAe pillar several different applications driven by LPAs are to be developed, such as a compact FEL, medical imaging and the first realization of LPA-beam injection into a storage ring. The latter endeavour is conducted in close collaboration between Deutsches Elektronen-Synchrotron (DESY), Karlsruhe Institute of Technology (KIT) and Helmholtz Institute Jena (HIJ). In the cSTART project at KIT, a compact storage ring optimized for short bunches and suitable to accept LPA-based electron bunches is in preparation. In this conference contribution we will introduce the 50 MeV LPA-based injector and give an overview about the project goals. The key parameters of the plasma injector will be presented. Finally, the current status of the project will be summarized.

FLUTE (Ferninfrarot Linac- und Test-Experiment) is a compact linac-based test facility for accelerator R&D and source of intense THz radiation for photon science. In preparation for the next experiments, the electron beam of the injector section of FLUTE has been characterized. In systematic studies the electron beam parameters, e.g., beam energy and emittance, are measured with several diagnostic systems. This knowledge allows the establishment of different operation settings and the optimization of electron beam parameters for future experiments.

The micro-bunching instability occurring in an electron storage ring for short electron bunches leads to the emission of coherent synchrotron radiation (CSR) in the terahertz range. This collective instability can be studied by observing the resulting fluctuations in the emitted CSR power with fast THz detectors. The measurements presented in this contribution were conducted at the KIT electron storage ring KARA (KArlsruhe Research Accelerator) in a dedicated short-bunch operation mode. For the measurements of these instability-induced CSR power fluctuations, an on-chip THz spectrometer providing a spectral fingerprint was used. The mm-sized chip contains eight antennas that are sensitive in different frequency ranges from 50 GHz up to 700 GHz, each connected to a Schottky-diode detector element. The on-chip spectrometer was read-out with the new version of the KArlsruhe Pulse Taking and Ultrafast Readout Electronics system, KAPTURE-2, providing continuous bunch-by-bunch data.