pygarn

Forward- and backward operations on graphs with a lot of fuzzyness.

Install via

• pip: pip install pygarn
• poetry: poetry add pygarn
• or add in your conda environment:

name: sur-your-env-name
channels:
- defaults
dependencies:
- python>=3.8
- pip
- pip:
  - pygarn

Visuals

Examples

import networkx as nx
from pygarn.base import RandomVertexSelector
from pygarn.growth import AddCompleteGraph

n_vertices_initial = 20
g_initial = nx.erdos_renyi_graph(n_vertices_initial, 0.3)

op_add_kcomplete = AddCompleteGraph(
    size=3,
    sources=RandomVertexSelector(min=1, max=3),
    targets=RandomVertexSelector(min=1, max=3),
)

g_new = op_add_kcomplete.forward(g_initial)

g_orig = op_add_kcomplete.backward(g_new)

# Should be highly likely:
assert nx.is_isomorphic(g_orig, g_initial)

import networkx as nx
from pygarn.base import VertexDegreeSelector
from pygarn.growth import AddVertex

n_vertices_initial = 20
g_initial = nx.erdos_renyi_graph(n_vertices_initial, 0.3)
n_edges_initial = len(g_initial.edges)
degrees_initial = [(v, d) for v, d in g_initial.degree()]

selector = VertexDegreeSelector()
op_add = AddVertex()
n_rounds = 5

g_current = g_initial.copy()
for _ in range(n_rounds):
    g_current = op_add.forward(g_current)