galib 2.0.dev0__tar.gz → 2.1__tar.gz

{galib-2.0.dev0 → galib-2.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: galib
-Version: 2.0.dev0
+Version: 2.1
 Summary: A library for graph analysis in Python / NumPy.
 Project-URL: Homepage, https://github.com/gorkazl/pyGAlib
 Project-URL: Source Code, https://github.com/gorkazl/pyGAlib/tree/master/src/galib
@@ -36,8 +36,6 @@ https://pypi.org/project/galib/)
 
 # pyGAlib – Graph Analysis Library in Python / NumPy
 
-> **Development branch**. Version 2 of pyGAlib under construction. Heavy changes expected in this branch. For download, please install the PyPI version or the one in the master branch of this repository. Both are installed via `pip`, see instructions in the corresponding README.md file. 
-
 pyGAlib is a library to generate and study graphs and complex networks in Python. It treats networks as adjacency matrices in order to take advantage of faster NumPy
 array manipulations. The library is easy to install, use, modify and extend.
 
@@ -253,6 +251,27 @@ limitations under the License.
 -----------------------------------------------------------------
 ### WHAT IS NEW
 
+##### June 26, 2026 (Release of Version 2.1)
+
+* **New functions** added:
+    * Support functions `is_directed()` and `is_weighted()` added to _metrics.py_ module, to facilitate working with weighted graphs.
+* **Support to generate and randomize weigthed networks** added. New functions included to _models.py_ module:
+    * `SeedRandomWeights()` adds random weights (sampled from a distribution of choice) to the links of an existing (di)graph.
+    * `ShuffleWeights()` conserves the links of a (di)graph in-place, but randomly reassigns their weights.
+* **"Syntactic sugar" functions** were added to generate some common random / weighted graphs (_models.py_):
+    * `ErdosRenyiGraph_Like()` generates an Erdös-Rényi (random) graph of same size and link probability as a given input network. Optionally, it also seeds random weights to the links, from a distribution of choice.
+    * `RandomGraph_Like()` generates a random graph of same size and number of links as a given input network. Optionally, it also seeds random weights to the links, from a distribution of choice.
+    * `WeightedErdosRenyiGraph()` generates a weighted Erdos-Renyi graph with link weights sampled from a distribution of choice.
+    * `WeightedRandomGraph()` generates a weighted random graph with link weights sampled from a distribution of choice.
+* New **example notebooks** added:
+    * [RichClub_Undirected.ipynb](Examples/RichClub_Undirected.ipynb) shows how to identify the presence of a rich-club in empirical (di)graphs.
+    * [WeightedGraphs_Intro.ipynb](Examples/WeightedGraphs_Intro.ipynb) illustrates how to use the new functions to work with weighted (di)graphs.
+* Minor changes and bug fixes:
+    * In *models.py* module, function renamed from `ModularHeterogeneousGraph()` to `ModularGraph()`.
+    * Bug fix in `ErdosRenyiGraph()` function that prevented addition of self-loops even with `selfloops = True` option set.
+    * `outdtype` option removed from graph generation functions in *models.py* module. For consistency, all graph generators return 2D arrays of `np.uint8` type, for the binary cases and `np.float64` for the weighted (di)graphs.
+
+
 ##### November 10, 2025 (Release of Version 2)
 
 Stable version 2.0 checked, validated and released.
@@ -263,42 +282,6 @@ Stable version 2.0 checked, validated and released.
 * Bug fixes to adapt to the various changes in Python and NumPy since last release.
 * Sample and validation scripts in the "*Examples/*" folder revised and adapted to recent changes in Python and NumPy. 
 
-##### March 14, 2024
-Small bugs fixed:
-
-- Normalization of `galib.metrics.Modularity()` function corrected.
-- Fixed the new  aliases for `int` and `float` in *Numpy*. All arrays are now declared as `np.int64` or `np.float64`, and individual numbers as standard Python `int` or `float`. 
-
-##### February 7, 2022
-Minor bug fixes. A remaining Python 2 to Python 3 conversion error was fixed, since standard library function `range()` no longer returns a list, but an iterator object.
-
-##### June 15, 2020
-Docstrings corrected. Function `k_DensityW()` was added in module *metrics.py* to calculate the k-density in networks with weighted links, which is needed to evaluate potential formation of rich-club structures in weigthed networks.
-
-##### July 12, 2019
-Version 1.1.0 released. Section for classic and deterministic graphs added to the *models.py* module. New generators `PathGraph()`, `StarGraph()` and `CompleteGraph()` included.
-
-##### July 6, 2019
-For clarity, function `RichClub()` has been splitted into two functions: `RichClub()`and `k_Density()`. The reason is that the output of old `RichClub()` was basically the k-density for all degrees, from 0 to kmax. Now this is done by `k_Density()` and the new `RichClub()` function identifies the set of nodes (hubs) for which k-density overcomes a given value.
-
-##### June 15, 2019
-GAlib has been registered in PyPI ([https://pypi.org/project/galib/](https://pypi.org/project/galib/)). Direct installation and version management using `pip` is thus available.
-
-##### January 29, 2019
-New in Version 1.0.1:
-
-- Minor corrections overall.
-- Function to generate Ravasz-Barabási hierarchical networks added to *models.py* module.
-
-##### December 3, 2018
-Release of Version 1.0.0 of pyGAlib. The library is now shaped as a proper Python package and is installable using standard tools. The structure of the package has been renewed and contains the following modules: 
-
-- *metrics.py*: Common graph metrics (degrees, clustering, graph distance, etc)
-- *models.py*: Generation of synthetic networks and randomization.
-- *tools.py*: Miscelaneous helper functions.
-- *metrics_numba.py*: Uses the Numba package to accelerate calculation of some metrics.
-- *models_numba.py*: Uses the Numba package to accelerate generation of some graph models.
-- *extra.py*: Additional measures and functionalities related to network analysis.
  
 See the file *CHANGELOG.md* for a complete history of changes.
 

{galib-2.0.dev0 → galib-2.1}/README.md

@@ -6,8 +6,6 @@ https://pypi.org/project/galib/)
 
 # pyGAlib – Graph Analysis Library in Python / NumPy
 
-> **Development branch**. Version 2 of pyGAlib under construction. Heavy changes expected in this branch. For download, please install the PyPI version or the one in the master branch of this repository. Both are installed via `pip`, see instructions in the corresponding README.md file. 
-
 pyGAlib is a library to generate and study graphs and complex networks in Python. It treats networks as adjacency matrices in order to take advantage of faster NumPy
 array manipulations. The library is easy to install, use, modify and extend.
 
@@ -223,6 +221,27 @@ limitations under the License.
 -----------------------------------------------------------------
 ### WHAT IS NEW
 
+##### June 26, 2026 (Release of Version 2.1)
+
+* **New functions** added:
+    * Support functions `is_directed()` and `is_weighted()` added to _metrics.py_ module, to facilitate working with weighted graphs.
+* **Support to generate and randomize weigthed networks** added. New functions included to _models.py_ module:
+    * `SeedRandomWeights()` adds random weights (sampled from a distribution of choice) to the links of an existing (di)graph.
+    * `ShuffleWeights()` conserves the links of a (di)graph in-place, but randomly reassigns their weights.
+* **"Syntactic sugar" functions** were added to generate some common random / weighted graphs (_models.py_):
+    * `ErdosRenyiGraph_Like()` generates an Erdös-Rényi (random) graph of same size and link probability as a given input network. Optionally, it also seeds random weights to the links, from a distribution of choice.
+    * `RandomGraph_Like()` generates a random graph of same size and number of links as a given input network. Optionally, it also seeds random weights to the links, from a distribution of choice.
+    * `WeightedErdosRenyiGraph()` generates a weighted Erdos-Renyi graph with link weights sampled from a distribution of choice.
+    * `WeightedRandomGraph()` generates a weighted random graph with link weights sampled from a distribution of choice.
+* New **example notebooks** added:
+    * [RichClub_Undirected.ipynb](Examples/RichClub_Undirected.ipynb) shows how to identify the presence of a rich-club in empirical (di)graphs.
+    * [WeightedGraphs_Intro.ipynb](Examples/WeightedGraphs_Intro.ipynb) illustrates how to use the new functions to work with weighted (di)graphs.
+* Minor changes and bug fixes:
+    * In *models.py* module, function renamed from `ModularHeterogeneousGraph()` to `ModularGraph()`.
+    * Bug fix in `ErdosRenyiGraph()` function that prevented addition of self-loops even with `selfloops = True` option set.
+    * `outdtype` option removed from graph generation functions in *models.py* module. For consistency, all graph generators return 2D arrays of `np.uint8` type, for the binary cases and `np.float64` for the weighted (di)graphs.
+
+
 ##### November 10, 2025 (Release of Version 2)
 
 Stable version 2.0 checked, validated and released.
@@ -233,42 +252,6 @@ Stable version 2.0 checked, validated and released.
 * Bug fixes to adapt to the various changes in Python and NumPy since last release.
 * Sample and validation scripts in the "*Examples/*" folder revised and adapted to recent changes in Python and NumPy. 
 
-##### March 14, 2024
-Small bugs fixed:
-
-- Normalization of `galib.metrics.Modularity()` function corrected.
-- Fixed the new  aliases for `int` and `float` in *Numpy*. All arrays are now declared as `np.int64` or `np.float64`, and individual numbers as standard Python `int` or `float`. 
-
-##### February 7, 2022
-Minor bug fixes. A remaining Python 2 to Python 3 conversion error was fixed, since standard library function `range()` no longer returns a list, but an iterator object.
-
-##### June 15, 2020
-Docstrings corrected. Function `k_DensityW()` was added in module *metrics.py* to calculate the k-density in networks with weighted links, which is needed to evaluate potential formation of rich-club structures in weigthed networks.
-
-##### July 12, 2019
-Version 1.1.0 released. Section for classic and deterministic graphs added to the *models.py* module. New generators `PathGraph()`, `StarGraph()` and `CompleteGraph()` included.
-
-##### July 6, 2019
-For clarity, function `RichClub()` has been splitted into two functions: `RichClub()`and `k_Density()`. The reason is that the output of old `RichClub()` was basically the k-density for all degrees, from 0 to kmax. Now this is done by `k_Density()` and the new `RichClub()` function identifies the set of nodes (hubs) for which k-density overcomes a given value.
-
-##### June 15, 2019
-GAlib has been registered in PyPI ([https://pypi.org/project/galib/](https://pypi.org/project/galib/)). Direct installation and version management using `pip` is thus available.
-
-##### January 29, 2019
-New in Version 1.0.1:
-
-- Minor corrections overall.
-- Function to generate Ravasz-Barabási hierarchical networks added to *models.py* module.
-
-##### December 3, 2018
-Release of Version 1.0.0 of pyGAlib. The library is now shaped as a proper Python package and is installable using standard tools. The structure of the package has been renewed and contains the following modules: 
-
-- *metrics.py*: Common graph metrics (degrees, clustering, graph distance, etc)
-- *models.py*: Generation of synthetic networks and randomization.
-- *tools.py*: Miscelaneous helper functions.
-- *metrics_numba.py*: Uses the Numba package to accelerate calculation of some metrics.
-- *models_numba.py*: Uses the Numba package to accelerate generation of some graph models.
-- *extra.py*: Additional measures and functionalities related to network analysis.
  
 See the file *CHANGELOG.md* for a complete history of changes.
 

{galib-2.0.dev0 → galib-2.1}/src/galib/__init__.py

@@ -219,7 +219,7 @@ from .models import*
 
 
 # Some metadata
-__version__ = "2.0.dev0"
+__version__ = "2.1"
 
 
 

{galib-2.0.dev0 → galib-2.1}/src/galib/extra.py

@@ -338,7 +338,7 @@ def FunctionalComplexity(corrmatrix, nbins=50, datarange=[0,1]):
 
     """
     # 0) Security checks
-    if len(np.shape(corrmatrix)) != 2:
+    if corrmatrix.ndim != 2:
         raise ValueError('Input data not a correlation matrix. Data not alligned.')
     if corrmatrix.min() < datarange[0]:
         raise ValueError('Input data not in range. Values smaller than range found.')

{galib-2.0.dev0 → galib-2.1}/src/galib/metrics.py

@@ -19,39 +19,44 @@ measures will be added to GAlib in future releases.
 
 BASIC CONNECTIVITY DESCRIPTORS
 ------------------------------
-Density
-    Returns the density of links in a network.
-Degree
-    Computes the number of neighbours of every node.
-Intensity
-    The total strength of a node in a weighted network.
-Reciprocity
-    Computes the fraction of reciprocal links to total number of links.
-ReciprocalDegree
-    Returns the reciprocal degree and excess degrees of every nodes.
+is_directed
+    Checks whether a (weighted) matrix represents a directed or undirected  graph.
+is_symmetric
+    Checks whether a (weighted) matrix is symmetric or not.
+
 AvNeighboursDegree
     Average neighbours' degree of nodes with given degree k, for all k.
 Clustering
     Returns the clustering coefficient and the local clustering of every node.
+Degree
+    Computes the number of neighbours of every node.
+Density
+    Returns the density of links in a network.
+Intensity
+    The total strength of a node in a weighted network.
 k_Density
     Computes the density of subnetworks made of nodes with degree >= k',
     for all k' = 0 to kmax.
-RichClub
-    Identifies the subset of hubs with dense interconnectivity.
 k_DensityW
     Computes the ratio of link weights among the nodes with strength > s',
     for s' = 0 to s' = smax.
 MatchingIndex
     Computes the number of common neighbours of every pair of nodes.
+Reciprocity
+    Computes the fraction of reciprocal links to total number of links.
+ReciprocalDegree
+    Returns the reciprocal degree and excess degrees of every nodes.
+RichClub
+    Identifies the subset of hubs with dense interconnectivity.
 
 PATHS AND GRAPH DISTANCE FUNCTIONS
 ----------------------------------
+AllShortestPaths
+    Finds all the shortest paths between two nodes.
 FloydWarshall
     Computes the pathlength between all pairs of nodes in a network.
 PathsAllinOne
     Returns pathlength and betweenness. Finds all shortest paths and cycles.
-AllShortestPaths
-    Finds all the shortest paths between two nodes.
 
 COMMUNITIES, COMPONENTS, K-CORES, ...
 -------------------------------------
@@ -59,12 +64,12 @@ AssortativityMatrix
     Returns the assortativity matrix of network given a partition of nodes.
 ConnectedComponents
     Finds all the connected components in a network out of a distance matrix.
-Modularity
-    Computes the Newman modularity given a partition of nodes.
 K_Core
     Finds the K-core of a network with degree k >= kmin.
 K_Shells
     Returns the K-shells of a network for all k from kmin to kmax.
+Modularity
+    Computes the Newman modularity given a partition of nodes.
 
 ROLES OF NODES IN NETWORKS WITH MODULAR ORGANIZATION
 ----------------------------------------------------
@@ -92,7 +97,6 @@ Hubness_GA
 
 """
 # Standard library imports
-import types
 # Third party imports
 import numpy as np
 # Local imports
@@ -101,6 +105,45 @@ from . import tools
 
 ############################################################################
 """CONNECTIVITY AND DEGREE STATISTICS"""
+def is_directed(adjmatrix):
+    """Checks whether a (weighted) matrix represents a directed or undirected graph.
+
+    Parameters
+    ----------
+    adjmatrix : ndarray of rank-2
+        A (weighted) adjacency matrix of a network. Weighted links are ignored.
+
+    Returns
+    -------
+    out : boolean
+        True if `adjmatrix` represents a directed graph, and False if `adjmatrix`
+        represents an undirected graph.
+    """
+    mask = adjmatrix.astype(bool)
+    out = (mask ^ mask.T).any()
+    return out.item()
+
+def is_symmetric(adjmatrix):
+    """Checks whether a (weighted) matrix is symmetric or not.
+
+    Parameters
+    ----------
+    adjmatrix : ndarray of rank-2
+        A (weighted) adjacency matrix of a network.
+
+    Returns
+    -------
+    out : boolean
+        True if `adjmatrix` represents a (weighted) undirected graph whose
+        link weights are all symmetric. False if any link weight `adjmatrix[i,j]`
+        differs from its reciprocal `adjmatrix[j,i]`. This can happen both if
+        the network is undirected but weights are not symmetrics, or if the
+        network is directed.
+    """
+    out = np.allclose(adjmatrix, adjmatrix.T)
+    return out
+
+
 def Density(adjmatrix):
     """Returns the density of links in a network.