ERT (secondary) inversion mesh

[romiele]

In standard ERT simulation, using ERTManager, the parameter mesh is embedded in a larger forward (secondary) mesh containing constant or, anyway, arbitrarily set resistivity values.

However, it seems like this is not happening when using ERTManager for inversion. If I look at the "response" function of ERTManager in this case, I see that self.Mesh() contains both parameters' region (area of interest of the inversion) and the secondary mesh area, but the resistivity values contained in "mod" are only those of the parameters mesh.

By default, what resistivity values are assigned to forward-mesh cells that lie outside the parameter mesh, and where in the code is this defined?
Or, alternatively, how are boundary effects handled in the forward modeling carried out during the inversion?

Specifically, I am referring to the code snippet for the forward solver of ERTManager

def response(self, mod):
        """Forward response (apparent resistivity)."""
        # ensure the mesh is initialized
        self.mesh()

        if self.complex() and self._conjImag:
            pg.warn('flip imaginary part for response calc')
            mod = self.flipImagPart(mod)

        resp = self._core.response(mod)

        if self.complex() and self._conjImag:
            pg.warn('backflip imaginary part after response calc')
            resp = self.flipImagPart(resp)

        return resp

For example, when running the inversion example in https://www.pygimli.org/_examples_auto/3_ert/plot_01_ert_2d_mod_inv.html#sphx-glr-examples-auto-3-ert-plot-01-ert-2d-mod-inv-py and analyzing the variables of this functions,

    IPdb [1]: self.mesh()
    Out  [1]: Mesh: Nodes: 22786 Cells: 26116 Boundaries: 24514

and

    IPdb [2]: mod.shape
    Out  [2]: (4800,)

[halbmy]

fop.mesh() is a (usually spatially refined) mesh including the outer boundary region. This is filled by prolongation from the (inner) parameter mesh (that can be retrieved by mgr.paraDomain). This procedure is in the C++ kernel. Definitely a good idea to have a Python implementation for it.

[romiele]

Thank you very much for your quick answer.
Is the prolongation function copying / extending the outermost resistivity values populating the inner mesh (basically at the boundaries) or is it something different?

If any documentation on this topic is available online or published, I would also be interested in reading it. Thanks again!