Topographic error wrong for hexagonal topography with rectangular grid about minisom

[JustGlowing]

Hi there, it's hard to help you with the details you shared. Check that the dimensions are what you expect and make sure that the indexing respects it.

…

On Mon, Oct 24, 2022, 18:33 Marcello Barisonzi ***@***.***> wrote: Hi, I am trying to get the topographic error from SOM with 11x7 neurons, hexagonal topography. When I do, I get this error: `File ~/DEIL_ISC/2021-2022/5-Processing/On-Reserve Facilities/1-CSD_Compare/TopographicError.py:23, in TopographicError.._get_euclidean_coordinates_from_index(som, index) 21 return (-1, -1) 22 y = som._weights.shape[1] ---> 23 coords = som.convert_map_to_euclidean((index % y, int(index/y))) 24 return coords File ~/.local/lib/python3.8/site-packages/minisom.py:243, in MiniSom.convert_map_to_euclidean(self, xy) 237 def convert_map_to_euclidean(self, xy): 238 """Converts map coordinates into euclidean coordinates 239 that reflects the chosen topology. 240 241 Only useful if the topology chosen is not rectangular. 242 """ --> 243 return self._xx.T[xy], self._yy.T[xy] IndexError: index 8 is out of bounds for axis 1 with size 7` I don't think this line of code <https://github.com/JustGlowing/minisom/blob/9a5415a176fde35400eaed57f70d5c51a22c60f1/minisom.py#L564> makes sense: coords = som.convert_map_to_euclidean((index % y, int(index/y))) Shouldn't the parameters be inverted, e.g.: coords = som.convert_map_to_euclidean((int(index/y, index % y))) Anyway, thanks for the amazing work! — Reply to this email directly, view it on GitHub <#152>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABFTNGLKHRMUXJX3XU5RU5LWE3B5FANCNFSM6AAAAAARNGHYS4> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[JanDetroy]

Hey,
I think I have the same Issue. I get a index out of bounds error, when I try to compute the topographical error using a hexagonal grid. Not using a hexagonal grid the topographical error can be computed just fine.
I am calling the som.topographical_error() function with my data as an argument.
te = som.topographic_error(data)
When not using a hexagonal grid the topographical error can be computed. However when using a hexagonal grid I get the following error:

~/.venvs/viewer/lib/python3.7/site-packages/minisom.py in topographic_error(self, data)
    558             return nan
    559         if self.topology == 'hexagonal':
--> 560             return self._topographic_error_hexagonal(data)
...
--> 245         return self._xx.T[xy], self._yy.T[xy]
    246 
    247     def _activate(self, x):

IndexError: index 5 is out of bounds for axis 1 with size 5

Hope that helps somehow.

[JustGlowing]

This must be bug then. I'll reopen the issue. Unfortunately I don't have time to fix this in the short term.

[JustGlowing]

A correction, I can't replicate this error as I'm able to compute the topographic error using the hexagonal topology under different setups. You need to share more details about your situation if you want any support with this.

[JanDetroy]

Hey,
Thanks for the super fast reply. I create a som, initialise it with weights according to the pca and train it. Afterwards I try to compute the topographic error which results in the out of bounds error. This is the code that I am running:

som_size = [6,5]
som = MiniSom(som_size[0], som_size[1], np.size(data,1), sigma=1.5, learning_rate=0.8,topology='hexagonal')
som.pca_weights_init(data)
som.train(data, 100000,verbose=False,random_order=True)

te = som.topographic_error(data)

[JanDetroy]

I don't think this line of code makes sense:

coords = som.convert_map_to_euclidean((index % y, int(index/y)))

Shouldn't the parameters be inverted, e.g.:

coords = som.convert_map_to_euclidean((int(index/y), index % y))

Anyway, thanks for the amazing work!

The change from above fixed the issue for me!