ValueError: all the input array dimensions except for the concatenation axis must match exactly, but along dimension 0, the array at index 0 has size 12 and the array at index 1 has size 11
The shape of the first array in this concatenate should always be ((params.shape[0]-1)/6, 6) or ((params.shape[0]-1)/4, 4) depending on if fits are circular or not. As far as I can tell this is working as expected.
For n >= 8, this leads to num_gauss being too small, hence the concatenation error. Here's a plot showing this behavior:
Replacing line 2152 with num_gauss = int(np.ceil((p0.shape[0]-1)/6)) should fix this issue, although I'm not sure if hard-coding in 6 will cause errors for circular fitting (when n_params = 4). It looks like it should be fine, but as an alternative the reshaped array could be pre-calculated before the concatenation (something like params_shaped = params[:-1].reshape((-1, n_params))) and then the np.ones array could just be of shape (params_shaped.shape[0], 1).
The concatenation on line 2198 of
utils.fit_gaussianshttps://github.com/sdfunni/SingleOrigin/blob/043dc6ab19d8c897253ac9b375706cddae8c3dca/src/SingleOrigin/utils.py#L2198-L2202 raises the following error for some images whenpeak_grouping_filter is not None:The shape of the first array in this concatenate should always be
((params.shape[0]-1)/6, 6)or((params.shape[0]-1)/4, 4)depending on if fits are circular or not. As far as I can tell this is working as expected.The second array in the concatenation will always have shape
(num_gauss, 1), wherenum_gaussis calculated on line 2152 https://github.com/sdfunni/SingleOrigin/blob/043dc6ab19d8c897253ac9b375706cddae8c3dca/src/SingleOrigin/utils.py#L2152The documentation for this function says
p0should always be shape((6*n)+1,): https://github.com/sdfunni/SingleOrigin/blob/043dc6ab19d8c897253ac9b375706cddae8c3dca/src/SingleOrigin/utils.py#L2122For
n >= 8, this leads tonum_gaussbeing too small, hence the concatenation error. Here's a plot showing this behavior:Replacing line 2152 with
num_gauss = int(np.ceil((p0.shape[0]-1)/6))should fix this issue, although I'm not sure if hard-coding in 6 will cause errors for circular fitting (whenn_params = 4). It looks like it should be fine, but as an alternative the reshaped array could be pre-calculated before the concatenation (something likeparams_shaped = params[:-1].reshape((-1, n_params))) and then thenp.onesarray could just be of shape(params_shaped.shape[0], 1).