Philipp SL Schäfer

Referring to "02_linear_algebra.ipynb", Figure 2.6 A: Of course the closure property is violated in A, but I was wondering why you used vec1*vec2 to show that, since element wise multiplication...

I have a brief question regarding the computation of q_{ij} in lines: https://github.com/jianhuupenn/SpaGCN/blob/a67402d954572b4f088475c79f3492ec1a6048e4/SpaGCN_package/SpaGCN/models.py#L25 In the paper the formula is:  whereas the code says rather (ignoring alpha)  so is...

Changing order of matrix multiplication evaluation to significantly reduce memory consumption and increase speed.

https://github.com/ulfaslak/py_pcha/blob/f398d28c6a28c4a8121cb75443b221fc58fc10e4/py_pcha/PCHA.py#L153 Why are you using np.ceil here? Because it is possible to select and $i$ that is not in the data leading to an index error.

https://github.com/scverse/squidpy/blob/16a9885976e24778664c81c05a9b592b2bd4a237/src/squidpy/_docs.py#L222-L223 The documentation for `legend_loc` in `spatial_scatter` refers to `matplotlib.legend.Legend` which is wrong I think. The `legend_loc` variable is passed to the scanpy function `_add_categorical_legend` ([code](https://github.com/scverse/scanpy/blob/ad657edfb52e9957b9a93b3a16fc8a87852f3f09/src/scanpy/plotting/_tools/scatterplots.py#L1093)), so it seems like...

If one creates a weighted, undirected graph from an adjacency matrix, wouldn't it make sense to check if the adjacency matrix is symmetric? This check was also introduced in `rigraph`...