Official Shardian Documentation
Welcome to the official technical documentation for Shardian, the high-performance physics-informed artificial intelligence suite for computational fluid dynamics (CFD) and mesoscale atmospheric modeling.
Our suite is designed for engineers, researchers, and climate data scientists who require maximum physical fidelity with minimal computational overhead.
Suite Overview
The Shardian suite is split into two specialized and independent simulation engines:
1. Shardian Aero (Computational Fluid Dynamics - CFD)
- Solver:
AdvancedZonalModelfor OpenFOAM. - Purpose: High-fidelity zonal turbulence modeling in internal and external flows.
- Technology: Symbolic expressions optimized and compiled natively in C++ (eliminating runtime dependencies on Python/JAX), optimized under the traditional RANS framework in OpenFOAM.
- Go to Section: Shardian Aero Documentation
2. Shardian Atmos (Meteorology & Mesoscale - WRF)
- Solver: Coupled
EML-SR(Planetary Boundary Layer & Surface Layer) physics schemes for WRF (Weather Research and Forecasting). - Purpose: Boundary layer modeling and precise thermodynamic surface coupling with adaptive canopy clamping.
- Technology: Solvers integrated into a portable Singularity/Apptainer container (
shardian_atmos.sif) to eliminate dependency hell and optimize execution in HPC clusters. - Go to Section: Shardian Atmos Documentation
Core Philosophy
- Zero CPU Overhead: The neural and symbolic solvers are compiled as closed algebraic equations in C++ and Fortran. This yields deep learning-level accuracy with the exact same CPU cost as traditional empirical models (such as \(k-\omega\) SST or MOST).
- Maximum Portability: Shardian Atmos is delivered as a single container, compatible with any workload manager (SLURM, PBS) and MPI environment.
- Developer Experience (DevX): A single automated command configures the environment and license keys in seconds.
Getting Started
To begin working with Shardian, proceed to the Installation & License Configuration page.