Convergent data‑driven workflows for open radiation calculations: an exportable methodology to any field

Abstract

The fast growth worldwide of linkable scientific datasets supposes significant challenges

in their management and reuse. Large experiments, such as the Latin American

Giant Observatory, generate volumes of data that can benefit other kinds of studies.

In this sense, there is a modular ecosystem of external radiation tools that should

harvest and supply datasets without being part of the main pipeline. Workflows for

personal dose estimation, muongraphy in volcanology or mining, or aircraft dose

calculations are built with different privacy policies and exploitation licenses. Every

numerical method has its own requirements and only parts could make use of the

Collaboration’s resources, which implies the convergence with other computing

infrastructures. Our work focuses on developing an agnostic methodology to address

these challenges while promoting open science. Leveraging the encapsulation of

software in nested containers, where the inner layers accomplish specific standardization

slices and calculations, the wrapper compiles metadata and data generated

and publishes them. All this allows researchers to build a data-driven computer

continuum that complies with the findable, accessible, interoperable, and reusable

principles. The approach has been successfully tested in the computer-demanding

field of radiation-matter interaction with humans, showing the orchestration with the

regular pipeline for diverse applications. Moreover, it has been integrated into public

or federated cloud environments as well as into local clusters and personal computers

to ensure the portability and scalability of the simulations. We postulate that this

successful use case can be customized to any other field.