Rapid Low-Noise Simulation of Ultra-bright 10 GeV Electron Bunches in Laser Plasma Accelerators

The BELLA project at LBNL seeks to develop ~10 GeV laser-plasma accelerator stages that will produce ultra-short, low-divergence electron bunches with energy spread of ~1%, slice energy spread of ~0.1%. A beam of sufficient brightness for collider applications can be used to drive a free electron laser and so this is a near-term experimental goal. Simulation support is required to reduce technical risk and increase the chances of experimental success. However, traditional particle-in-cell (PIC) simulations suffer from high-frequency particle noise, which artificially increases the emittance and energy spread of the simulated electron bunch. Fundamentally new techniques are required to adequately suppress numerical noise. We are adding relevant, novel algorithms to the parallel VORPAL framework, including a quasistatic treatment of the beam self-fields, careful initialization of externally-injected electron bunches, 2D cylindrically symmetric versions of existing algorithms and full Vlasov-Maxwell on 3D and 4D meshes. We are carefully benchmarking these algorithms with both PIC and fluid representations of the plasma, both in the lab frame and in Lorentz-boosted frames, using physical parameters relevant to the BELLA Project. The reduction in particle-driven noise and associated artificial emittance growth is dramatic. For high charge, ultra-bright relativistic electron beams, we showed that a quasistatic treatment of the self-fields (analogous to what is done in standard tracking codes) correctly captures relativistic cancellation of transverse forces and essentially eliminates artificial emittance growth from particle noise. We showed that this approach can be combined with standard PIC or fluid treatments of the plasma, in the lab frame or a Lorentz-boosted frame, for 2D production simulations of laser-plasma accelerators. Algorithms implemented during Phase I will be generalized to work in 3D and with all relevant VORPAL features. We will add an electromagnetic update with all E and B components collocated at mesh nodes. The laser envelope model will be modified to work in 2D r-z geometry for fast simulations. When testing and benchmarking, we will use BELLA Project parameters and also explore its expected performance. We will generalize the Vlasov-Poisson algorithms under development to work for relativistic Vlasov-Maxwell, and will use them to simulate colliding pulse injection in one and two spatial dimensions, on three and four dimensional meshes, respectively. We will make VORPAL easier to use. Commercial Applications and Other Benefits: Implementation and testing of the proposed algorithms and capabilities in VORPAL will create additional opportunities for sales to laser-plasma groups around the world. Each year, more such groups are using VORPAL. The addition of relativistic Vlasov-Maxwell capabilities will more generally extend the scope of applications that can be modeled. In addition to being a key component of the DOE advanced accelerator program, laser-plasma accelerators may also contribute to next-generation light sources.