Explore projects

ZipGAN is a deep learning framework for compressing and decompressing 3D Direct Numerical Simulation (DNS) datasets of turbulent flows. It uses a Super-Resolution GAN to reconstruct full-resolution DNS fields from compressed (filtered + downsampled) representations, achieving compression ratios of up to 512x while preserving fine-scale turbulence features.

Package for analysis and manipulation of xyz coordinates files.

Finite Difference based Porous Media Anisotropic Permeability Solver for Stokes flow