Question about _dual_steps definition.

[MKAbdElrahman]

Since the boundary condition are periodic, the first center point is linked to the last center.

Why you needed to subtract a more additional step grid_size[dim] in centers[-1] - grid_size[dim]

 def _dual_steps(self) -> Coords:
        """Return dual steps of the cells in the :class:`Grid`.
        Returns
        -------
        :class:`Coords`
            Distances between each of the cell centers along each dimension.
        """

        # We need the dual steps to be same size as grid.centers, so we pad with periodicity applied
        grid_size = [(bounds[1] - bounds[0]) for bounds in self.boundaries.to_list]
        dsteps = {}
        for dim, (key, centers) in enumerate(self.centers.dict(exclude={TYPE_TAG_STR}).items()):
            centers_pad = np.concatenate(([centers[-1] - grid_size[dim]], centers))
            dsteps[key] = np.diff(centers_pad)

        return Coords(**dsteps)

[momchilmm]

Yeah, this should be fixed to use the periodicity properly, good catch!

[MKAbdElrahman]

thanks!

[MKAbdElrahman]

I think it is wrong to define a step size this way. The distance h will be same as the domain length in one direction and that would lead to a big finite difference error. The proper step size is (h1 + h_end)/2 where h1 and h_end are the first and last step sizes.

[momchil-flex]

Fixed in 6003315, thanks!