This page will provide the reusable scientific outputs of the project. The work-package reports explain how the database and surrogate models are constructed; this page will provide the released artifacts, their documentation, version information, and citation instructions.
Simulation Database
The database will contain transport results generated with the T3ST test-particle code over selected equilibrium, particle, and turbulence parameter ranges.
- Variable definitions, units, and admissible ranges
- Database size and sampling strategy
- Training, validation, and test partitions
- Download link, version, checksum, and DOI
Files and metadata will be added after validation and release.
Neural-Network Ensemble
The surrogate will consist of an ensemble of trained feed-forward neural networks for fast prediction of the transport quantities represented in the database.
- Architecture and ensemble size
- Input and output normalization
- Training configuration and validation metrics
- Native model files and portable exports, where available
Model files will be published together with their applicability limits.
Software & Examples
Supporting material will demonstrate how to load the data, reproduce preprocessing, evaluate the ensemble, and interpret its predictions.
- Minimal inference example
- Input-validation and preprocessing routines
- Example dataset or notebook
- Source-code repository
Citation & Licensing
Each public release will state the authors, license, associated publications, release date, version identifier, and preferred citation.
Citation and licensing details will be finalized before publication.
Relationship to the Work Packages
WP3 documents the construction and validation of the simulation database. WP4 documents predictions and physical interpretation. This page is the distribution point for the final database, trained surrogate ensemble, and the materials required to reuse them.