Distributed Vector Approximate Message Passing

Distributed VAMP is an iterative algorithm for solving linear inverse problems. This algorithm is more robust towards heterogenity with a tradeoff with accuracy of convergence.

• In this repository, We implemented D-VAMP with different experimentations with increasing Noise Correlation, Heterogenous Agents, Attacking agents.
• Proved better robustness with increased number of heterogenous agents suggesting wide range of practical application.

How to Run

For NRMSE vs Number of Iterations

• Run the python files in the MONTE CARLO Folder, for gaussian and binary prior.

For Experimentation with Noise Correlation and Heterogenity

• Run the respective python files in Experiments Folder.