Inconsistency for sensitivity calculation in 2-layer models

[mpessel]

Dear pyGIMLi Developers,

I am trying to compute the sensitivity for a 2-layer model using the example plot_05_ert_4_point_sensitivities.py as a reference. However, I am encountering an inconsistency between the dimensions of my mesh and the Jacobian matrix.

world = mt.createWorld(start=[0, 0], end=[30, -10], worldMarker=True, layers=[-5])
for pos in scheme.sensorPositions():
    world.createNode(pos)
mesh = mt.createMesh(world, area=.05, quality=33, marker=1)
rhomap = [[1, 100.],[2, 10.]]
data = ert.simulate(mesh, scheme=scheme, res=rhomap)
fop = ert.ERTModelling()
fop.setData(data)
fop.setMesh(mesh)
model = np.ones(mesh.cellCount())
fop.createJacobian(model)
print("mesh.cellCount(): ",mesh.cellCount())
print(mesh.cellCount(), fop.jacobian())

10379
RMatrix: 3 x 5216

[halbmy]

Obviously you have 3 data. The number of model parameters represents only the part with marker 2 (mesh.submesh(marker=2).cellCount()) because by default the region with the smallest marker is treated as background. You have to disable this by fop.region(1).setBackground(False).