IMASHelper — IMAS Access & MHD Mapping Layer

Overview

IMASH is a lightweight C++ library that extracts plasma and machine data from IMAS (Integrated Modelling & Analysis Suite) and maps it into solver-ready data structures. The actual C++ wrapper tends to increase the compile time when used in conjunction with accelerator compilers (e.g. nvcc). This library completely isolates the IMAS structure from the accelerator compilers.

It is designed as a bridge layer between IMAS and numerical solvers (e.g. AMReX-based FDTD codes), while preserving a Python-verifiable workflow via file outputs.

---

Running the IMASH test executable

The executable requires the IMAS path and a time (in seconds):

./imash <imas_path> <time> [output_dir]

Example:

./imash ~/iter/134174/117/ 60.0 ./

This will read IMAS data at t = 60.0 s and write output files to the current directory.

What IMASH Does

IMASH performs three core operations:

1. Data Extraction from IMAS

Reads data from standard IMAS IDS:

• core_profiles
• equilibrium
• edge_profiles (GGD)
• wall

Example:

IdsNs::IDS ids;
ids.open("imas:hdf5?path=...", OPEN_PULSE);

---

2. Mapping to Physical Space

IMAS data is not directly solver-ready. IMASH reconstructs physical fields:

Core plasma (structured grid)

Maps 1D profiles → 2D equilibrium grid:

$$ \rho(R,Z) = \sqrt{\frac{\psi(R,Z)-\psi_{\text{axis}}}{\psi_{\text{bnd}}-\psi_{\text{axis}}}} $$

• Produces:
  • $n_e(R,Z)$
  • $T_e(R,Z)$
• Uses interpolation in $\rho$

---

Edge plasma (unstructured mesh)

• Extracts cell-centered data from GGD mesh
• Reconstructs:
  • $(R,Z)$ from node geometry
  • $n_e, T_e$ from field values

Optional:

• Projection to equilibrium grid via $(\rho,\theta)$

---

Magnetic field

From equilibrium:

• $B_R(R,Z)$
• $B_Z(R,Z)$
• $B_\phi = B_T \frac{R_0}{R}$

---

Wall geometry

• Extracts limiter and vessel contours
• Produces polylines in $(R,Z)$

---

3. Packaging into Solver-Ready Structures

IMASH builds clean C++ objects:

• PlasmaData
• WallData
• EdgeCellPoint
• EqGrid2D

These are independent of AMReX or any solver.

---

Design Philosophy

Separation of concerns

Layer	Responsibility
IMASH	Read & reconstruct physics
Solver	Discretization & time evolution

IMASH:

• does not know about MultiFab, MPI, or GPUs
• produces pure C++ data structures

---

Dual workflow

IMASH supports:

✅ Debug / validation mode

Write outputs to file:

WriteCore2D(ids, time, "core.dat");
WriteEdge2D(ids, time, "edge.dat");

Used with Python:

plt.pcolormesh(...)
plt.tricontourf(...)

✅ Production mode

Return in-memory structures:

PlasmaData plasma = ExtractPlasmaData(ids, time);

---

Data Conventions

Structured grid (equilibrium)

Flattening convention:

$$ k = j \cdot n_R + i $$

• $i$: radial index
• $j$: vertical index

---

Edge data

• Stored as unstructured points
• No flattening
• Derived from mesh connectivity

---

Units

• Length: meters
• Density: $\text{m}^{-3}$
• Temperature: eV
• Magnetic field: Tesla

---

Typical Usage

IdsNs::IDS ids;
ids.open("imas:hdf5?path=...", OPEN_PULSE);

double time = 60.0;

// Full plasma extraction
PlasmaData plasma = ExtractPlasmaData(ids, time,
                                      true,  // core
                                      true,  // B field
                                      true,  // edge projection
                                      true,  // merge density
                                      true); // merge temperature

// Debug output
WriteCore2D(plasma, "core.dat");
WriteBfield2D(plasma, "bfield.dat");

---

Integration with Solvers (e.g. AMReX)

IMASH is designed to feed directly into simulation codes:

PlasmaData plasma = ExtractPlasmaData(ids, time);
FillMultiFabFromPlasma(plasma, ...);

• IMASH handles physics reconstruction
• Solver handles:
  • domain decomposition
  • GPU transfer
  • time stepping

The key step takes place in the reader functions which will connect the data used by the solver and the data owned by IMAS. The reader functions must be designed in a way where IMASH.H is present only in the private part of the code, i.e. the cpp file. The public part where the function headers reside, the H file, should only have TokamakData.H. The H and cpp file should have no headers requiring accelerator compile or the problem will still exists. The isolation must be complete, and only pure C++ functions must be used in the reader functions.

---

Supported Features

• ✅ Core profile mapping
• ✅ Edge GGD reconstruction
• ✅ Edge → grid interpolation ($\rho,\theta$)
• ✅ Magnetic field extraction
• ✅ Wall geometry extraction
• ✅ Python-compatible outputs

---

Future Extensions

• STL wall import (CAD geometry)
• Reflectometer / antenna geometry
• 3D toroidal reconstruction
• Direct AMReX ingestion utilities
• Time-dependent IMAS sequences

---

Dependencies

• IMAS C++ Access Layer (ALClasses)
• C++17
• Standard library only (no solver dependency)

---

Why IMASH Exists

IMAS provides rich physics data, but:

• data is heterogeneous
• grids are not solver-ready
• edge uses unstructured meshes

IMASH solves this by providing:

A consistent, physics-aware mapping from IMAS → simulation-ready fields

---

Summary

IMASH is a thin but critical layer that:

• converts IMAS data into physical fields
• unifies core + edge descriptions
• remains solver-agnostic
• preserves validation through file output