Adding Uniform Background Magnetic Field via A_external in Hybrid-PIC Turbulence Simulations

[yanghf263]

Technical Details

• WarpX version: 25.06-8-ge59329031fda (and PICSAR 25.06)
• Simulation mode: Hybrid-PIC (Ohm's law)
• Geometry: 2D (XZ plane)
• Boundary conditions: Periodic in both directions
• Domain size: 50 × 50 ion skin depths, 512 × 512 cells
• Python interface: Using PICMI and pywarpx
• Relevant code: A_external set via Az_external_function = -x*B0 for uniform By field

---

Description

I am running turbulence simulations with the Hybrid-PIC solver in WarpX. My goal is to initialize both a perturbed magnetic field δB and a uniform background magnetic field By = B0 (constant and spatially uniform).

• The perturbed field δB is introduced via AnalyticInitialField (self field).
• The uniform background field By = B0 is intended to be introduced via A_external (external field), specifically by setting Az_external_function = -x*B0 (so that By = B0).

Problem

• When using A_external as above, the simulation aborts during initialization with the error:

  Assertion `Efield_fp[lev][idim]->is_finite()' failed
  

• If I set A_external to a periodic function (e.g., Az_external_function = sin(kx*x)), the simulation runs without issue.
• This suggests that A_external must satisfy periodic boundary conditions.

---

Questions

1. For hybrid-PIC simulations, is it possible to introduce an external field (background magnetic field) directly as a magnetic field, rather than via the vector potential (A_external)?
2. Is there a recommended way to introduce a uniform and constant background magnetic field as an external field (not just as an initial condition) in such simulations?
3. What are the possible physical and numerical differences between introducing the background magnetic field as an external field (via A_external) versus as a self field (via AnalyticInitialField)?

---

Any clarification on the way to introduce a uniform background magnetic field in hybrid-PIC turbulence simulations would be greatly appreciated!

[ax3l]

Fix by @clarkse in #5988