adjoint simulation adjoint source

[gaoyajian]

Source idx 25 at location [ 1.69635e+06, 4.48826e+06, 4.19137e+06 ] was not attached. Please check the coordinates ...
Weird problem though the source are in the center of domain.

[gaoyajian]

this problem sometimes emerges when I set the tensor order to 1 when generating mesh file

[gaoyajian]

adj_src = get_adjoint_source(comm=comm, event=event, iteration=iteration)
w = simulation.Waveform(mesh=mesh, sources=adj_src)
adj_src_copy = w.physics.wave_equation.point_source
del w.physics.wave_equation.point_source
w.adjoint.forward_meta_json_filename = f"REMOTE:{meta}"
w.adjoint.gradient.parameterization = comm.project.salvus_settings[
"gradient_parameterization"
]
w.adjoint.gradient.output_filename = "gradient.h5"
w.adjoint.point_source = adj_src_copy
w.validate()

from Mike:

this is a replacement for these lines: https://github.com/dirkphilip/LASIF_2.0/blob/master/lasif/salvus_utils.py#L217-L227
and should change the VectorPoint3DZNE to SideSetVectorPoint3DZNE online 157

[gaoyajian]

adj_src = [
source.seismology.SideSetVectorPoint3DZNE(
latitude=rec["latitude"],
longitude=rec["longitude"],
side_set_name='r1',
fz=1.0,
fn=1.0,
fe=1.0,
source_time_function=stf.Custom(
filename=adjoint_filename,
dataset_name=f"/{rec['network-code']}_{rec['station-code']}",
),
)

    for rec in adjoint_sources
]

[gaoyajian]

Nice scheme to avoid this fatal error when the mesh using topography and ellipsoid.