QED.jl - Strong-field particle physics code

The collision of relativistic electron beams with highly intense, highly energetic and
short-pulsed light will give deep insights into the interactions of electromagnetic fields
and matter at extreme scales. Experimentally, those collisions might be addressed
at upcoming projects like HIBEF 2.0 at the EuropeanXFEL, SYLOS at ELI-ALPS or
LCLS-II at SLAC, to name a few. The precise theoretical description of such collision
experiments is very challenging and not fully covered by the currently available tools,
known from particle physics. We develop the open-source software library QED.jl,
which targets those gaps by
Modelling of (non-linear) Quantum Processes
providing new developments of
higher-order
pair production
state-of-the-art modelling tools w.r.t.
inelastic scattering
Processes
and annihilation
strong-field physics. This includes

• Modelling of particle physics

processes: calculation of Matrix
element and cross section

• Monte-Carlo event generation:

Parallelised drawing of samples
from multivariate distributions,

• Multivariate integration:

Algorithms for highly oscillatory
problems and Monte-Carlo
integration for Total cross sections
Monte-Carlo Event-Generator
Large-Scale Simulation
Classical
Processes
QED.jl is written in the Julia
Laser-Matter Collision Experiments
programming language, which
opens up the usage of modern
language features like just-in-time compilation, multiple-dispatch and meta-
programming to attain efficiency in execution time, where the code is still easy to use
and develop. Consequently, based on the computational demanding tasks given by
the physics use case, necessary advances w.r.t. distributed computing are planed to
be developed using Julia:

• Task scheduling using directed acyclic graphs:

Generation of compute graphs from specific physical models, and optimisation of
the evaluation of such graphs in parallel,

• Code injection:

Extension of Julia compile workflow by injecting problem specific C++ code,

• Hardware-agnostic parallelisation:

Kernel abstractions in Julia, e.g. by using ALPAKA