2D kCSD: setting gdX changes resolution in Y dimension

[fkloosterman]

Hi,

Thanks for the nice package. I have played with the 2D kCSD example (http://pykcsd.readthedocs.org/en/latest/usecases.html#sample-2d-reconstruction), but came across the following issue. If I set the gdX parameter to a higher value, say 0.5, then the resolution in the Y-dimension and not the X-dimension changes. See figure below and compare to the original in the documentation. Is this a bug, or am I missing something?

Best,
Fabian

[parkag]

Thank you, Fabian!
It looks like there is a bug connected with distributing sources in space. The number of points in each direction is calculated from gdX, gdY
Code sample from KCSD2D:

nx = (self.xmax - self.xmin)/self.gdX + 1
ny = (self.ymax - self.ymin)/self.gdY + 1
lin_x = np.linspace(self.xmin, self.xmax, nx)
lin_y = np.linspace(self.ymin, self.ymax, ny)
self.space_X, self.space_Y = np.meshgrid(lin_x, lin_y)
(self.X_src, self.Y_src, self.R) = sd.make_src_2D(
self.space_X,
self.space_Y,
self.n_sources,
self.ext_x,
self.ext_y,
self.R_init
)

[fkloosterman]

Hi Grzegorz,
Could it have something to do with the call to np.meshgrid, which by default switches the first two dimensions, unless you set the indexing parameter to 'ij'?
-Fabian

[ccluri]

Instead of

lin_x = np.linspace(self.xmin, self.xmax, nx)
lin_y = np.linspace(self.ymin, self.ymax, ny)
self.space_X, self.space_Y = np.meshgrid(lin_x, lin_y)

Use:

 self.space_X, self.space_Y = np.mgrid[xmin:xmax:np.complex(0,nx),
                                       ymin:ymax:np.complex(0,ny)]