arXiv Science⌕ Search

arXiv subjects

Matthias Remta

Publications and source records attributed to Matthias Remta.

3 recordsLinked to original sources

Surrogate-assisted beam distribution shaping at the CERN Proton Synchrotron and Super Proton Synchrotron

The Super Proton Synchrotron at CERN delivers protons to the North Area experimental facility at 400 GeV/c. The extracted beam is split twice vertically in the transfer line with Lambertson septa to serve three fixed-target experiments simultaneously. A significant fraction of the beam is lost during these splitting processes, driving radioactive activation of the Lambertson septa. This contribution investigates reducing the splitting losses by optimising the vertical beam distribution before extraction. The developed method relies on resonant excitation with the transverse damper, reshaping the Gaussian vertical profile into a uniform and ultimately bimodal distribution, with the excitation amplitude acting as a single continuous control knob. A loss reduction of up to 23.5 % was measured. In simulations, the vertical amplitude detuning is established as a second knob that reduces the resulting beam envelope. This technique was verified with measurements at the CERN Proton Synchrotron. Surrogate models based on generative deep learning were developed for efficient real-time validation and Bayesian system identification.

physics.acc-ph↗

Flow-based surrogate models for particle tracking

Particle tracking is a fundamental tool for particle-accelerator design and optimisation. Conventional tracking routines provide high accuracy but are computationally demanding, especially when simulating large particle ensembles or long time spans. As a result, optimising moderate- to high-dimensional parameter spaces is challenging, and real-time surrogate models remain out of reach for many applications. This contribution introduces a surrogate-modelling approach based on conditional flow matching (CFM). A CFM model is trained on tracking simulations of CERN's Proton Synchrotron (PS) over a 10-dimensional parameter space. The trained model reproduces final phase-space distributions with a median squared maximum mean discrepancy MMD$^2$ of $3\times 10^{-4}$ and mean inference time of 0.04 s, a speed-up of three orders of magnitude over conventional tracking. To capture distribution-dependent dynamics that vanilla CFM cannot represent, we extend the model with cross-attention over the initial particle ensemble (Cross-Attention-CFM) and demonstrate that this extension recovers performance on a space-charge benchmark in the PS, where vanilla CFM degrades. Finally, we introduce Hybrid-CFM, in which a small number of conventionally-tracked particles are used to inform the model. On the same 10-dimensional PS task, Hybrid-CFM with 100 auxiliary particles trained on 200 distributions matches the vanilla CFM trained on 1500, and improves the worst-case (90th-percentile) MMD$^2$ by roughly a factor of four, substantially reducing the upfront cost of building a surrogate.

physics.acc-ph↗

Particle tracking with physics-informed deep learning methods

Simulating the motion of charged particles in electromagnetic fields is essential for designing and optimising particle accelerators. Conventional tools rely on symplectic integration schemes, which provide high accuracy but are computationally expensive. As a consequence, optimisation in moderate to high-dimensional parameter spaces as well as simulations of tens of thousands to millions of particles can be computationally prohibitive. This contribution explores the possibilities of employing modern machine-learning based tools, in particular SympNet and DeepONet, to enable fast particle simulations. A major novelty is the modification of the conventional SympNet architecture to enable learning of parametric Hamiltonian dynamics. The models are trained and tested on a toy setup of a circular accelerator comprising two different types of quadrupole magnets with varying field strengths. All models achieved faster inference than the symplectic integrator, at the expanse of significantly reduced accuracy. The SympNet implementation achieved the lowest mean squared error. Additionally, a DeepONet was employed to predict the evolution of particle densities, derived from the single-particle simulations.

physics.acc-ph↗