Introduction to IMAS

What is IMAS?

The Integrated Modelling & Analysis Suite is the collection of physics software that will be used to support ITER operations and research as defined in the ITER Integrated Modelling Programme...

... it comprises plasma modelling tools for systematic planning (evaluate candidate plasma operating scenarios and assist in the development of plasma control strategies) and analysis of each ITER discharge.

A brief history

• 2003: the Integrated Tokamak Modelling (ITM) starts discussing common data ontology in EU
• 2011: start of IMAS design and development at the ITER Organization (IO), building upon ITM concepts
• 2015: first IMAS prototype published ¹ (version controlled ontology: 3.2.0, 30 classes)
• 2017: WEST tokamak uses IMAS in production for its plasma reconstruction chain ²
• 2021: Heating and current drive workflow in IMAS ³
• 2023: IMAS workflow for energetic particle stability ⁴
• 2024: end of version 3 development (3.42.0, 81 classes), introducing version 4

IMAS software stack

IMAS software stack

Data model

• Covers simulation and experimental data
  • Made of Interface Data Structures (IDS)
  • Described in XML in the Data Dictionary
• Made machine agnostic by design
  • Used in experiments (e.g. WEST, ITER)
  • Mapping of existing experiments data
  
  • Support validation of ITER applications
• Can serve as a data standard for magnetic confinement fusion

Generic tools

data access

API

MATLAB® Java™

I/O
imas-core

ASCII

FlexBuffers

visualisation

workflows

databases

Applications

Energetic particles

Synthetic diagnostics

SOL and PWI

High-fidelity plasma simulator

Open sourcing IMAS

A long journey

https://github.com/iterorganization

IMAS-Data-Dictionary (DD)

For the definition and documentation of IMAS IDS

• IDS schemas definition
• Light Python code (build, CLI tool)
• For all questions and requests
• Documentation on readthedocs
• Package available on PyPI

IMAS-Python

For accessing and manipulating IDS in Python

• Source code under
• Documentation, package on PyPI
• DD selection & conversion at runtime
• Lazy loading, metadata, ducktyping
• I/O with imas-core not open yet
• Self-described I/O with netCDF format

		  
		    # create and explore IDS
		    import imas
		    from imas.util import inspect
		    ids = imas.IDSFactory("4.0.0")
		    equil = ids.equilibrium()
		    inspect(equil)
		    print(equil.time.metadata.units)
		    
		    # fill-in data in IDS
		    equil.ids_properties.homogeneous_time=1
		    equil["ids_properties/comment"]="simple test"
		    equil.time=[1.0]
		    
		    # save IDS into netCDF
		    fout=imas.DBEntry("myequil.nc","w")
		    fout.put(equil)
		    fout.close()
		    
		    # read IDS from netCDF
		    with imas.DBEntry("myequil.nc","r") as fin:
		       eq = fin.get("equilibrium",lazy=True)
		       print(eq.ids_properties.comment)
		  
		

Available soon on GitHub

IDS-Validator

• Python CLI tool for checking validation rulesets on IDS values
• Rules implemented as functions with decorators
• Generic or use-case specific rules (tokamak, code)

GGD-VTK plugin

• ParaView plugin to visualise meshes stored as GGD² in IDS
• More readers: 1D/2D profiles, beams, lines of sight...
• Access to data via imas-python

SimDB + Dashboard

• Tools to build, share and explore databases of IMAS simulations
• CLI client, remote server, REST API
• Web dashboard to query & compare simulations

Challenges and opportunities

Software with background intellectual property (BIP)

• IMAS relies on software developed within ITER Members (BIP)
• IO invests effort in using/combining/extending codes in IMAS (GIP)
• IO asked code owners about the possibility to release their BIP as open source → a general trend in fusion?

Contributing back

A key aspect for project sustainability, can take several forms

Adapting a code to IMAS

Why?

• Code shall run within ITER suite
• Benchmarking exercise with similar codes
• Standardize machine description

How?

• I/O data must be defined by following the IMAS DD
• Non-intrusive adaptation with a wrapper
• Depends on targetted use (standalone, coupling, ...)

License

• Always add a license for your source code
• Consider open licenses and open repositories

Mapping existing data in IMAS

Motivations

• Prepare ITER operation: validate tools, training
• Towards multi-machines databases
• Use available IMAS application to improve modelling capability in given areas

Caveats

• Involve a large effort from experts (sparse resources)
• Non-backward compatible changes in IMAS DD
• Lack of immediate pay off

Mitigations

• Targetted effort, identify success stories
• Share mappings and experience on open platforms^6,7

Migrating to a new developer platform

Limitations and perspectives

Towards a data standard for the fusion community