mnets / pymnet Goto Github PK

Dear bolozna,

I apologise in advance as I am relatively new to python and perhaps the issue I'm experiencing is the result of that rather than a real issue. I really tried to make sure I checked everything, but you never new when learning a new programming language..

I'm trying to get the pymnet library running. I cloned the repository and I have added the library directory to my python path in Spyder. Now I'm running the first lines but run into the error:

ImportError: cannot import name 'sampling' from 'pymnet' (/usr/local/anaconda3/envs/multilayer-networks/lib/python3.7/site-packages/pymnet/init.py)

If you have any idea what I'm doing wrong here and if you could point me in a right direction (documentation?) I'm super grateful.

Thank you very much in advance,
Tessa

Dear authors,

While running your tutorial, I noticed the existence of the class pymnet.net.MultilayerNode:

from pymnet import *
net = MultilayerNetwork(aspects=0)
net.add_node(1)
net[1]

<pymnet.net.MultilayerNode object at 0x000002CD29CAAA50>

which exists here.

Anyway I could not find documentation for it. Since it exposes public APIs, I'd suggest providing such documentation.

Also, I noticed the use of str to name the function that returns the weighted degree of a node. Since the str name is historically assigned to strings in Python, I would suggest renaming it to something else.

Dear authors,

While running the tutorial on automorphisms, I got the error:

import pymnet
net_social = pymnet.MultiplexNetwork(couplings='categorical', fullyInterconnected=False)
net_social["Alice", "Bob", "Friends"] = 1
net_social["Alice", "Carol", "Friends"] = 1
net_social["Bob", "Carol", "Friends"] = 1
net_social["Alice", "Bob", "Married"] = 1

net_transport = pymnet.MultiplexNetwork(couplings='categorical', fullyInterconnected=False)
net_transport["Helsinki", "Turku", "Train"] = 1
net_transport["Helsinki", "Tampere", "Train"] = 1
net_transport["Turku", "Tampere", "Train"] = 1
net_transport["Helsinki", "Turku", "Ferry"] = 1

print(pymnet.get_automorphism_generators(net_social, allowed_aspects=[0]))

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\Users\claud\anaconda3\envs\pymnet_review\Lib\site-packages\pymnet\isomorphisms\__init__.py", line 197, in get_automorphism_generators
    assert len(automorphism_group_generator_backends) > 0, \
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: No backends for automorphism generators were imported!

I understand that the bliss-blind package is needed. Anyway, I could not find it online (also, it should come installed as a dependency on Windows x64 machines, according to the documentation).

Is there a reason for having both pyproject.toml and setup.py? Can you remove setup.py?

openjournals/joss-reviews#6930

Checklist

Primary responsibilities assigned following our discussion, feel free to modify/object.
Once an item is considered done, check its box.
Async discussion in the thread for this issue.

General checks

• Source code publicly available on GitHub
• OSI approved plain-text LICENSE file
• Author contribution

Functionality

• Installation possible via pypi @arashbm
• Cleaner handling of Networkx dependency @arashbm
• Networkx 3.x support (optional) @arashbm
• Graphlet functionality (handling Bliss dependency: make it compatible w/ Python 3, replace by SOTA graph isomorphism library, ...?) @arashbm + @ercco
• Update project requirements/toml to reflect Networkx 3.x support @arashbm

Code quality

• (Almost) full test coverage @ercco (graphlet + sampling functionality), all others for the rest
• Automated testing (GitHub actions) @arashbm
• Codecov integration (optional) @dataspider
• Codeclimate integration (optional) @dataspider

Documentation

• Installation instructions + clearly stated list of dependencies + automated package management @arashbm
• Automated documentation (sphinx + GitHub actions) @arashbm
• Code of conduct (w/ MK contact webpage as default contact point) @dataspider

Writeup

• Separate "publication" branch @dataspider
• Writing folder "paper" with "paper.md" and "paper.bib" in it @dataspider
• Summary of purpose and high-level functionality @bolozna (first draft)
• Statement of need @ercco (first draft)
• State of the field ("related packages", papers using pymnet already) @ercco (first draft)
• Example usage with code examples and pretty figures @ercco + @dataspider
• High-quality writing @dataspider
• Complete references using rmarkdown citation syntax @dataspider
• Check acknowledgements section + add grant number @bolozna

Bells and Whistles

• Docstrings for remaining API methods + add to references.rst @ercco (graphlet + sampling functionality)
• Other CI (incl. updating pypi package) @arashbm
• State-of-the-art example data provision @dataspider
• Example usage (old tutorials -> ipynb tutorials) @dataspider
• Proof-reading all tutorials @dataspider
• Proof-reading all docstrings @dataspider
• Community guidelines (see GitHub's templates) @dataspider @arashbm

Dear authors,

In the tutorial section about multiplex networks, after defining the 2-aspect multiplex network conet you try to add a layer to its third aspect, which results in an error.

conet = MultiplexNetwork(couplings=['categorical','ordinal'])
conet.add_node('node')
conet.add_layer('a',1)
conet.add_layer('b',1)
conet.add_layer(1,2)
conet.add_layer(2,2)
conet.add_layer(3,3)

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\Users\claud\anaconda3\envs\pymnet_review\Lib\site-packages\pymnet\net.py", line 997, in add_layer
    if layer not in self.slices[aspect]:
                    ~~~~~~~~~~~^^^^^^^^
IndexError: list index out of range

This is certainly not essential to this review, but for greater consistency (in the future), I would choose snakecase or camelcase and stick with it. Right now, there is a mix of both.

openjournals/joss-reviews#6930

Please consider fixing these typos in docs/tutorial/networktypes:

These to two ways are equivalent, but [...]

The multilayer networks can in theory be used to represented multiplex networks

JOSS Review Issue: openjournals/joss-reviews#6930

There is a typo here.

The paragraph can be fixed as follows:

Currently, the network is assumed to be unweighted and, if the input file contains a directed network, an undirected one is returned anyway. The directions are removed in a way that i and j are connected if there is either link i->j or j->i (or both).

I have identified an issue in the Reading and Writing Networks section of the documentation. Currently, this section only includes information about the read_ucinet function.

To ensure comprehensive documentation, please include the full API related to reading and writing networks. This will enhance the usability and clarity of the documentation for all users.

I really liked the overview page, but it would be great to 1) link to it on the README 2) add a visual(s) of multilayer networks for newcomers and 3) clarify the language surrounding "aspects". I understand that (like in Multilayer Networks), this is pretty standard terminology, but it might be more consistent to rename aspects to num_layers or something similar.

openjournals/joss-reviews#6930

TypeError: FigureBase.gca() got an unexpected keyword argument 'projection'

As stated in matplotlib/matplotlib#20770 matplotlib's FigureBase.gca() deprecated the 'projection' argument, however Pymnet draw() is still using it in the current version.

There is a typo here:

Fix: substitute permutated with permuted.

When I add edges of a single layer using the code like mnet = MultilayerNetwork(aspects = 3) and mnet[1,2,'gene','gene'] = 1, it reports the error like "KeyError: 'Invalid number of indices.'". But when I change the aspects of the multilayer network to 2 mnet = MultilayerNetwork(aspects = 2) and run the same code to add edges, the program runs correctly. So what is the cause of this error?

Dear authors,

While reviewing your package, I noticed the following unexpected behaviours:

1. Assigning strings to edges is allowed:

from pymnet import *
net = MultilayerNetwork(aspects=0)
net[1,2] = "hello"
net[1,2] 

"hello"

2. Modifying the MultilayerNetwork via a MultilayerNode:

from pymnet import *
net = MultilayerNetwork(aspects=0)
net.add_node(1)
net[1].mnet.add_node(2)
list(net)

[1, 2]

I understand that these behaviors do not break the software, therefore fixing them is not mandatory for this review. Anyway, for the future, I would like to respectively point out:

1. Disallowing this by e.g. throwing and error;
2. Using getters and setters, or name mangling;

Not necessary for this review, but in the future, it could be useful to standardize the docstrings. One example of a standard is numpydoc.

openjournals/joss-reviews#6930

In the docs/tutorial

• the mentioned URL to the GitHub repository works, but it is a redirect. Could you update it?
• the mentioned git clone path is outdated and requires a public SSH key. Could you substitute it with the HTTPS version (i.e. git clone https://github.com/mnets/pymnet.git)?

On Python 3.10.8:

>>> import pymnet as pn
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "Multilayer-networks-library/pymnet/__init__.py", line 1, in <module>
    from .net import MultilayerNetwork,MultiplexNetwork
  File "Multilayer-networks-library/pymnet/net.py", line 810, in <module>
    class MultiplexIntraNetDict(collections.MutableMapping):
AttributeError: module 'collections' has no attribute 'MutableMapping'

Related: https://stackoverflow.com/questions/59636631/importerror-cannot-import-name-mutablemapping-from-collections
Looks like MutableMapping has been moved from collections to collections.abc.

Temporary fix found in the StackOverflow answer: https://stackoverflow.com/a/75382123

Overall the paper looks good. Some comments:

• The "Statement of Need" section was clearly written and helpful.
• There might not be space, but it would be cool to have a visualization of a small empirical multilayer network.
• In the summary, there are a few long and confusing sentences. For example, in the sentence "Multilayer networks (Kivelä et al., 2014) generalize graphs to capture the rich network data often associated with complex systems, allowing us to study a broad range of phenomena using a common representation, using the same multilayer tools and methods" can you remove the comma or end with "using the same representations, tools, and methods." In addition, can you rewrite "For many systems, there is a need to represent richer network data, such as multidimensional features, several types of interactions, numerous layers of hierarchy, and multiple modalities, to capture the structure and dynamics of the systems more accurately than simple graphs would allow." to something like "Many systems contain nodes with multidimensional features, interactions of different types, different levels of hierarchy, and multiple modalities, which are more faithfully captured with multilayer networks than simple graphs."
• The "Main Features and Examples" section was well-written and helpful.

openjournals/joss-reviews#6930

Seems like vscode needed %matplotlib inline to use pymnet. It works now thanks.

AttributeError Traceback (most recent call last)
in ()
23 two_layer_net[4, 4, 'a', 'b'] = 1
24
---> 25 fig = draw(two_layer_net, layout="circular")
.......
AttributeError: 'NoneType' object has no attribute 'text'

There are several long lines (this one, for example) which should be wrapped using a formatter such as black.

In addition, using isort would help standardize import statement order.

openjournals/joss-reviews#6930

I see that there is a brief contribution guide in the README. It would be helpful to add a few issues with the "Good first issues" (For example, see NetworkX, and link to this label in the contribution section.

In the future, it would be great to make the contribution guide a bit more step-by-step (e.g., fork the repository, make your changes, check test coverage, format code, add docs (is there a docstring format?), make a PR, etc.)

openjournals/joss-reviews#6930

Hello
May I ask how to call related packages to solve some related node indicators? For example, the degree and centrality of nodes, etc.

It would be great if this package were better covered. There are one or two modules with very little test coverage, which would be low-hanging fruit to improve coverage.

In addition, it would be great to add minimum coverage requirements for new pull requests using the CodeCov yaml file.

openjournals/joss-reviews#6930

Hi @bolozna ,

Please could you help me out with below mentioned error:
ImportError: cannot import name 'sampling' from partially initialized module 'pymnet' (most likely due to a circular import) (/home/duvuki/anaconda3/envs/colstrike/lib/python3.9/site-packages/pymnet/init.py)

I searched for posts with solutions, and tried commenting from . import sampling (last line) in init.py
Still receiving same error.
Thanks in advance!!

Dear authors,

I noticed that in the tutorial on Using NetworkX functions, some residuals from Python 2 are present:

print net

print map(nx.number_connected_components,models.er(1000,3*[0.005]).A.values())

Moreover the last line, when ported to python 3, does not return [9, 11, 8] as reported:

print(map(nx.number_connected_components,models.er(1000,3*[0.005]).A.values()))

<map object at 0x000001B26309D4B0>

The current paper title is “pymnet: A Multilayer-Networks Library for Python”.

I believe the hyphen in “Multilayer-Networks” is unnecessary. I suggest removing the hyphen and considering alternative titles such as:

1. “pymnet: A Multilayer Networks Library for Python”
2. “pymnet: A Python Library for Multilayer Networks”
3. “pymnet: A Multilayer Networks Library in Python”

Hi, I'm trying to plot a multilayer with six layers, but I only get 3 and a half layers. I have tried modifying the figsize but the other layers are still blank, although the lab of those layers appears. I would appreciate your help to visualize the whole network.

Also, I would like to know how to set the order of layers.

Thanks for advance.

Hello,
After I installed pymnet from git, I run into the below issue when trying run "from pymnet import *":

ImportError: cannot import name 'sampling' from partially initialized module 'pymnet' (most likely due to a circular import)

Can you help me on this? thanks,
Yuanyuan

In the "Related Packages" section of the paper you write the following paragraph:

Could you consider including Py3Plex?

mnet = MultilayerNetwork(aspects=2) #aspects==dimension
mnet[1,2,'a','b','x','y']=1
mnet.add_layer('c',1)
mnet.add_layer('z', 2)
fig=draw(mnet)

I tried example of multi aspects.
But they said,

Traceback (most recent call last):
File "D:/TSMM/2.DigitalHumanities/MultilayerNetworksLibrary/multi_aspect2.py", line 14, in
fig=draw(mnet)
File "D:\TSMM\2.DigitalHumanities\MultilayerNetworksLibrary\pymnet\visuals\drawcore.py", line 159, in draw
assert net.aspects==1
AssertionError

Hi @bolozna

I want the model to be displayed on the web page.And I saw that there is code in pymnet about implementing it with threejs. I called an earlier version of threejs to render successfully.But I found that the content about the layer is not well generated.
Thanks in advance

Can you condense this line to

try:
    from collections import MutableMapping
except ImportError:
    from collections.abc import MutableMapping

I don't love the pass statements. With the above, if neither work, it will throw an error.

openjournals/joss-reviews/#6930

The tutorials are really helpful, but it would be great if there was a tutorial on working with an empirical dataset (See Netzschleuder, for example).

openjournals/joss-reviews#6930

Please consider fixing these typos in docs/overview

Here we take e a similar approach

We typically consider that the number of aspects, $d$, to be constant and not dependent on the size of the data.

That is, the graph is a dictionary where each node [...] is a key and values whose value is another dictionary containing information about the neighbors of each node.

JOSS Review Issue: openjournals/joss-reviews#6930

Dear authors,

If I run your tutorial on Visualizing Networks, the figure does not show:

from pymnet import *
net = models.er_multilayer(5,2,0.2)
fig = draw(net,show=True)

C:\Users\claud\anaconda3\envs\pymnet_review\Lib\site-packages\pymnet\visuals\drawbackends\mpl.py:79: UserWarning: FigureCanvasAgg is non-interactive, and thus cannot be shown
  plt.show()

I tried installing PyQt6 (as suggested here) to no avail.

Would you have any idea on what could the problem be (perhaps it is related to my machine or environment)?

Environment:

(pymnet_review) C:\Users\claud>conda list
# packages in environment at C:\Users\claud\anaconda3\envs\pymnet_review:
#
# Name                    Version                   Build  Channel
bzip2                     1.0.8                h2bbff1b_6
ca-certificates           2024.3.11            haa95532_0
contourpy                 1.2.1                    pypi_0    pypi
cycler                    0.12.1                   pypi_0    pypi
expat                     2.6.2                hd77b12b_0
fonttools                 4.53.0                   pypi_0    pypi
kiwisolver                1.4.5                    pypi_0    pypi
libffi                    3.4.4                hd77b12b_1
matplotlib                3.9.0                    pypi_0    pypi
networkx                  3.3                      pypi_0    pypi
numpy                     2.0.0                    pypi_0    pypi
openssl                   3.0.14               h827c3e9_0
packaging                 24.1                     pypi_0    pypi
pillow                    10.3.0                   pypi_0    pypi
pip                       24.0            py312haa95532_0
pymnet                    0.2.0                    pypi_0    pypi
pyparsing                 3.1.2                    pypi_0    pypi
pyqt6                     6.7.0                    pypi_0    pypi
pyqt6-qt6                 6.7.2                    pypi_0    pypi
pyqt6-sip                 13.6.0                   pypi_0    pypi
python                    3.12.4               h14ffc60_1
python-dateutil           2.9.0.post0              pypi_0    pypi
scipy                     1.13.1                   pypi_0    pypi
setuptools                69.5.1          py312haa95532_0
six                       1.16.0                   pypi_0    pypi
sqlite                    3.45.3               h2bbff1b_0
tk                        8.6.14               h0416ee5_0
tzdata                    2024a                h04d1e81_0
vc                        14.2                 h2eaa2aa_1
vs2015_runtime            14.29.30133          h43f2093_3
wheel                     0.43.0          py312haa95532_0
xz                        5.4.6                h8cc25b3_1
zlib                      1.2.13               h8cc25b3_1

OS: windows 10

IDE: vscode