vibeflow-nektar

name	vibeflow-nektar
description	Use when preparing, running, debugging, post-processing, or extending VibeFlow/VibeCAE Nektar++ spectral/hp element workflows from 0-caseDict/caseDict, including XML session files, high-order meshes, incompressible/compressible solvers, FieldConvert/NekMesh, boundary conditions, polynomial order, filters, and reusable high-order CFD/PDE templates.

VibeFlow Nektar++

This skill treats Nektar++ as the high-order spectral/hp element CFD and PDE backend for VibeCAE. It is suitable for high-accuracy flow, wave, stability, and controlled-scale high-order discretization studies.

• Usage mode: generate Nektar++ mesh/session XML from caseDict, run the solver, and post-process with FieldConvert.
• Development mode: develop session templates, high-order mesh workflows, parameter sweeps, FieldConvert pipelines, or Nektar++ extensions.

When to Use

• A high-order spectral/hp method is needed instead of a conventional low-order FVM/FEM workflow.
• The task involves incompressible or compressible flow, vortex dynamics, stability, waves, or verifiable high-accuracy PDE work.
• Tools such as NekMesh, FieldConvert, and IncNavierStokesSolver are required.

First Decide: Usage Mode or Development Mode

• Current high-order case setup, execution, and post-processing: use usage mode.
• XML templates, mesh converters, parameter sweeps, plugins, or solver extensions: use development mode.
• Nektar++ commands or library paths are unavailable: hand off to vibeflow-toolchain first.
• Geometry or high-order meshes are not prepared: hand off to geometry or mesh first.

What to Read First

• PDE, boundaries, initial conditions, Re/Ma, time integration, and output objectives in 0-caseDict/caseDict.
• references/nektar-playbook.md
• references/nektar-practical-tips.md: software knowledge base, key concepts, failure interpretation, and adapter design notes.
• High-order mesh or convertible mesh under B-mesh/.
• Existing C-run/nektar/ and Ref/ material.

Input

• 0-caseDict/caseDict
• Nektar++ XML mesh/session files, Gmsh meshes, parameter tables, initial fields, or restart files

Output

• C-run/nektar/case-*/mesh.xml
• C-run/nektar/case-*/session.xml
• C-run/nektar/case-*/Allrun / Allclean
• .fld, .chk, .vtu, and logs
• D-post/nektar_postprocess.py

Recommended Structure

C-run/nektar/
`-- case-1/
    |-- mesh.xml
    |-- session.xml
    |-- logs/
    |-- results/
    |-- Allrun
    `-- Allclean

Usage-Mode Workflow

1. Map caseDict to the Nektar++ session: solver, parameters, variables, boundary regions, initial conditions, time integration, filters, and I/O.
2. Prepare or convert the mesh, and check that composites and boundary regions match the caseDict boundaries.
3. Choose polynomial order, quadrature/integration rules, time step, and stabilization settings.
4. Run a short-time or short-iteration smoke test.
5. Use FieldConvert to convert .fld/.chk to .vtu or to extract profiles.
6. Post-process residuals, energy, force coefficients, field views, and error metrics.
7. Write the run status and results back to caseDict.

Development-Mode Workflow

1. Define the development target: session template, high-order mesh pipeline, solver parameter sweep, FieldConvert post-processing, or Nektar++ module extension.
2. Write the development note first, recording the Nektar++ version, commands, XML interface, boundary-region mapping, and validation example.
3. For high-order mesh workflows, record curved-boundary order and geometry error.
4. For solver extensions, start from official examples and do not directly modify the system installation.
5. Validate at least solver startup and FieldConvert on a small mesh.

Core Rules

• High-order methods are sensitive to boundary regions, curved geometry, time step, and aliasing.
• The XML session is an execution artifact; caseDict is the source of truth for physical intent.
• Do not claim high-order geometry accuracy from only a low-order linear mesh; curved-boundary requirements must be stated.
• FieldConvert outputs are the post-processing entry point, but they do not replace preserving the original .fld/.chk files.

Validation

• Nektar++ solver, NekMesh, and FieldConvert are available.
• XML files parse successfully.
• The solver can start and generate .fld or .chk files.
• FieldConvert can generate the target format.
• Post-processing can read results and check key quantities.

Fallback Rules

• Boundary regions are unclear: return to the mesh stage.
• PDEs or parameters are unclear: return to vibeflow-case-dict.
• The environment is unavailable: return to vibeflow-toolchain.