heterogeneous spatial networks

A repository of research codes that computationally synthesize and analyze heterogeneous spatial networks. The networks of focus include (in alphabetical order): artificial bimodal end-linked polymer networks, artificial polydisperse end-linked polymer networks, artificial uniform end-linked polymer networks, Delaunay-triangulated networks, spider web-inspired Delaunay-triangulated networks, and Voronoi-tessellated networks.

Setup

Once the contents of the repository have been cloned or downloaded, the Python virtual environment associated with the project needs to be installed. The installation of this Python virtual environment and some essential packages is handled by the virtual-environment-install-master.sh Bash script. Before running the script, make sure to change the VENV_PATH parameter to comply with your personal filetree structure. Alternatively, a Conda environment can be installed with the required packages. All required packages are listed in the requirements.txt file.

Structure

The core functions in this repository are modularly distributed in Python files that reside in the following directories:

• file_io
• helpers
• networks
• topological_descriptors

The core functions can then be called upon in Python files or Jupyter notebooks for various tasks. The following directories contain various Python files that synthesize and analyze the aforementioned types of heterogeneous spatial networks:

• abelp
• apelp
• auelp
• delaunay
• swidt
• voronoi

Each of these directories contain a YAML file defining a wide variety of parameter configuration settings. Moreover, each of these directories contain two sub-directories, topology and descriptors. Within these sub-directories are YAML files that define parameter configuration settings specifically related to network topology and descriptor calculations. The Hydra package is employed to load in the settings from the YAML files.

The following Python files synthesize various heterogeneous spatial networks and calculate topological descriptors when run in the order provided:

• In the abelp directory: abelp_network_topology_synthesis.py -> abelp_network_topology_descriptors.py
• In the apelp directory: apelp_network_topology_synthesis.py -> apelp_network_topology_descriptors.py
• In the auelp directory: auelp_network_topology_synthesis.py -> auelp_network_topology_descriptors.py
• In the delaunay directory: delaunay_network_topology_synthesis.py -> delaunay_network_topology_descriptors.py
• In the swidt directory: swidt_network_topology_synthesis.py -> swidt_network_topology_descriptors.py
• In the voronoi directory: voronoi_network_topology_synthesis.py -> voronoi_network_topology_descriptors.py

In addition, several Python files are supplied in the abelp, apelp, and auelp directories that plot the spatially-embedded structure of a given network and analyze the statistics of various network topological features. Time and bandwidth permitting, such functionality will also be extended to the delaunay, swidt, and voronoi networks.

Usage

Before running any of the code, it is required that the user verify the baseline filepath in the ``filepath_str()`` function of the ``file_io.py`` Python file in the ``file_io`` directory. Note that filepath string conventions are operating system-sensitive.

Example timing benchmarks

The following contains some timing benchmarks for the network synthesis and topological descriptor Python files on my Dell Inspiron computer with cpu_num = 8 for the topology\20250102A and descriptors\20250102A parameter configuration settings:

• abelp_network_topology_synthesis.py -> abelp_network_topology_descriptors.py: ~ 60 seconds (for abelp_network_topology_synthesis.py) + ~ 30 minutes (for abelp_network_topology_descriptors.py)
• apelp_network_topology_synthesis.py -> apelp_network_topology_descriptors.py: ~ 105 minutes (for apelp_network_topology_synthesis.py) + ~ 25 minutes (for apelp_network_topology_descriptors.py)
• auelp_network_topology_synthesis.py -> auelp_network_topology_descriptors.py: ~ 30 seconds (for auelp_network_topology_synthesis.py) + ~ 30 minutes (for auelp_network_topology_descriptors.py)
• delaunay_network_topology_synthesis.py -> delaunay_network_topology_descriptors.py: ~ 10 seconds (for delaunay_network_topology_synthesis.py) + ~ 80 minutes (for delaunay_network_topology_descriptors.py)
• swidt_network_topology_synthesis.py -> swidt_network_topology_descriptors.py: ~ 20 seconds (for swidt_network_topology_synthesis.py) + ~ 25 minutes (for swidt_network_topology_descriptors.py)
• voronoi_network_topology_synthesis.py -> voronoi_network_topology_descriptors.py: ~ 30 seconds (for voronoi_network_topology_synthesis.py) + ~ 35 minutes (for voronoi_network_topology_descriptors.py)

Note that the Voronoi-tessellated network timing benchmarks are measured with respect to running the topology\20250102A, topology\20250102B, descriptors\20250102A, and descriptors\20250102B parameter configuration settings altogether. Also note that essentially every type of topological descriptor is calculated for each network in the topological descriptor Python files (for the sake of benchmarking).