EB and AMR: Segfault when Using Embedded Boundaries with Mesh Refine

[ArmourPiercer1]

Hi everyone,

I'm new to the WarpX and I want to simulate the space charge effect in a vacuum tube. The problem involves some complex electrode shapes, and the zone that my beam passes by is much smaller than the tube, so I wish to use the embedded boundary and the mesh refinement at the same time.

The initial MLMG solver works well on the problem. However, the program reports a segfault when it comes to STEP 1. I wonder if it is caused by some mistakes in my input file, and how can I solve that?

I have tried the following methods, but none of them worked:
(1) Move the refined regions away from the embedded boundary
(2) Let the refined regions completely include the embedded boundary
(3) Move the particle initializing regions such that it is fully / partially / not overlaped with the refined region
(4) Set guard cells of the refined patch manually

Here is a simple test example Script.py (which input file generated by the program automatically is input_file.txt). The test works well with amr.max_level=0, but if the parameter is changed to 1, the output will become:

Level 0: dt = 2.286926927e-13 ; dx = 0.00025 ; dy = 0.00025 ; dz = 0.00025
Level 1: dt = 2.286926927e-13 ; dx = 0.000125 ; dy = 0.000125 ; dz = 0.000125

Grids Summary:
  Level 0   1 grids  1048576 cells  100 % of domain
  Level 1   1 grids  1769472 cells  21.09375 % of domain

-------------------------------------------------------------------------------
--------------------------- MAIN EM PIC PARAMETERS ----------------------------
-------------------------------------------------------------------------------
Precision:            | DOUBLE
Particle precision:   | DOUBLE
Geometry:             | 3D (XYZ)
Operation mode:       | Electromagnetic
                      | - vacuum
-------------------------------------------------------------------------------
Current Deposition:   | Esirkepov
Guard cells           | - ng_alloc_J  = (5,5,5)
 (allocated for J)    |
Particle Pusher:      | Boris
Charge Deposition:    | standard
Field Gathering:      | energy-conserving
Particle Shape Factor:| 3
-------------------------------------------------------------------------------
Maxwell Solver:       | Yee
                      | - current correction is ON
                      | - staggered grid
Guard cells           | - ng_alloc_EB = (4,4,4)
 (allocated for E/B)  |
-------------------------------------------------------------------------------
For full input parameters, see the file: warpx_used_inputs

MLMG: # of AMR levels: 1
      # of MG levels on the coarsest AMR level: 4
MLMG: Initial rhs               = 0
MLMG: Initial residual (resid0) = 4.360626566e+11
MLMG: Iteration   1 Fine resid/resid0 = 0.04536132682
MLMG: Iteration   2 Fine resid/resid0 = 0.001462461101
MLMG: Iteration   3 Fine resid/resid0 = 5.212717643e-05
MLMG: Iteration   4 Fine resid/resid0 = 1.70154524e-06
MLMG: Iteration   5 Fine resid/resid0 = 5.832183471e-08
MLMG: Iteration   6 Fine resid/resid0 = 1.808757471e-09
MLMG: Iteration   7 Fine resid/resid0 = 6.342648775e-11
MLMG: Iteration   8 Fine resid/resid0 = 2.834913371e-12
MLMG: Final Iter. 8 resid, resid/resid0 = 1.236199856, 2.834913371e-12
MLMG: Timers: Solve = 0.2371373 Iter = 0.2210851 Bottom = 0.1257459
MLMG: # of AMR levels: 1
      # of MG levels on the coarsest AMR level: 5
MLMG: Initial rhs               = 0
MLMG: Initial residual (resid0) = 1.218354195e+11
MLMG: Iteration   1 Fine resid/resid0 = 0.01176605151
MLMG: Iteration   2 Fine resid/resid0 = 0.0003176746371
MLMG: Iteration   3 Fine resid/resid0 = 1.739124407e-05
MLMG: Iteration   4 Fine resid/resid0 = 9.569522914e-07
MLMG: Iteration   5 Fine resid/resid0 = 5.02190693e-08
MLMG: Iteration   6 Fine resid/resid0 = 2.592496322e-09
MLMG: Iteration   7 Fine resid/resid0 = 1.328635995e-10
MLMG: Iteration   8 Fine resid/resid0 = 6.78127539e-12
MLMG: Final Iter. 8 resid, resid/resid0 = 0.8261995316, 6.78127539e-12
MLMG: Timers: Solve = 0.2783111 Iter = 0.2541071 Bottom = 0.0439858
--- INFO    : Writing plotfile diags/diag1000000
STEP 1 starts ...
Segfault
See Backtrace.0.0 file for details
SIGABRT
See Backtrace.0.0 file for details