Kevin D. Smith - Software

Implicit Flow Networks

Implicit Flow Networks (IFN) are an implicit neural network model for the problem of network flow estimation, proposed in my NeurIPS 2022 paper.

Source	https://github.com/KevinDalySmith/implicit-flow-networks
First Release	October 2022
Relevant Papers	“Physics-Informed Implicit Representations of Equilibrium Network Flows”

High-Order Tomography

High-Order Tomography (HOT) is a Python prototype for the topology inference method proposed in my IEEE Transactions on Networking paper.

Project Page	https://high-order-tomography.readthedocs.io/en/latest/
Source	https://github.com/KevinDalySmith/high-order-tomography
First Release	April 2022
Relevant Papers	“Topology Inference with Multivariate Cumulants: The Möbius Inference Algorithm”

PyMoments

PyMoments is a Python library for computing multivariate k-statistics (i.e., minimum-variance unbiased estimators of multivariate cumulants). This project started as a utility function for High-Order Tomography, but the applications are general enough that I released it as a standalone package on PyPI. For an overview of the underlying mathematics, see my tutorial paper, “A Tutorial on Multivariate k-Statistics and their Computation”.

Project Page	https://pypi.org/project/PyMoments/
Source	https://github.com/KevinDalySmith/PyMoments
First Release	May 2020
Relevant Papers	“A Tutorial on Multivariate k-Statistics and their Computation”