graph-lib

0.4.8 verified Fri May 01 auth: no python

A set of diffusion related graph algorithms for Python, including community detection, node ranking, and graph sampling. Current version 0.4.8. Low release cadence.

pip install graph-lib

Common errors

error ModuleNotFoundError: No module named 'graphlib' ↓

cause Trying to import as 'graphlib' which is a stdlib module. The correct import uses underscore: 'graph_lib'.

fix

pip install graph-lib and then import graph_lib (with underscore).

error TypeError: unsupported operand type(s) for +: 'int' and 'list' ↓

cause Passing non-integer node labels that cause internal conversion errors.

fix

Relabel nodes to integers: G = nx.convert_node_labels_to_integers(G)

error AttributeError: module 'graph_lib' has no attribute 'diffusion_map' ↓

cause diffusion_map was removed/deprecated in newer versions. Use spectral_embedding instead.

fix

Use graph_lib.spectral_embedding(G, n_components=None) instead.

Warnings

deprecated Function 'graph_lib.diffusion_map' has been deprecated in v0.4.0. Use 'graph_lib.spectral_embedding' instead. ↓

fix Replace graph_lib.diffusion_map(G) with graph_lib.spectral_embedding(G, n_components=None)

gotcha The package imports as 'graph_lib' (underscore), not 'graphlib'. Importing 'graphlib' silently imports Python's stdlib module which does not raise an error but gives wrong results. ↓

fix Use 'import graph_lib' or 'from graph_lib import ...'

gotcha Graph algorithms expect node labels to be integers. Non-integer labels may cause silent type errors in internal matrix construction. ↓

fix Ensure node labels are integers, e.g., G = nx.convert_node_labels_to_integers(G)

Imports

graph_lib
wrong
```
import graphlib
```
correct
```
import graph_lib
```
graphlib is a stdlib module (topological sort) completely different from graph-lib

Quickstart

Detect two communities in Zachary's karate club graph using spectral partitioning.

import graph_lib
import networkx as nx

G = nx.karate_club_graph()
communities = graph_lib.spectral_partitioning(G, k=2)
print(communities)