cuTradeNet 0.1.2

GPU-Accelerated Kinetic Wealth Exchange Models on Complex Networks

cuTradeNet library provides classes to easily create & run kinetic wealth exchange models on complex networks.

Leads the user to set one (or ensemble) of complex networks as a contact structure agents use to trade about. The following wealth exchange models were implemented:

• Yard-sale model
• Merger-Spinoff model
• Dragulescu and Yakovenko
• Constant model
• Chatterjee, Chakrabarti and Manna
• "All in" model

It is written in Python and uses Cuda module from Numba package to accelerate the simulation runnin in GPU, paralelizing some transaccions in the same graph and paralelizing runs in multiple graphs, leading to easier & faster averaging of system properties. It's completely abstracted from the CUDA knowledge for the user, so you can use it as a regular Python library.

How to use

There is a Demo notebook in the repository that can be tryed in it's Google Colab version too (you can use the package there if you don't have a NVIDIA gpu).

There is also a General explanation of Kinetic Wealth Exchange Models used.

How to install

You can install it from PyPi with the following command:

pip install cuTradeNet

Repository&Questions

The repository is in GitHub, and you can ask questions or contact us in the Discussions section.

CUDA dependencies

In order to use this library in your personal computer you should have a CUDA capable gpu and download the CUDA Toolkit for your OS. If you don't fulfill this requirementes you can always use it in the cloud. Don't hesitate to contact us to get help!