Data model as agile basis for evolving calibration software

TL;DR

This paper presents a machine-readable data model for calibration software design, enabling automated generation of documentation, prototypes, and archives to support flexible development and hardware-software trade-offs in astronomical instrumentation.

Contribution

It introduces a detailed, machine-readable design approach that improves consistency and efficiency in calibration software development for the ELT MICADO instrument.

Findings

Automated generation of design documentation and prototypes.

Enhanced consistency between design and implementation.

Facilitated early-phase software development and hardware-software trade-offs.

Abstract

We design the imaging data calibration and reduction software for MICADO, the First Light near-IR instrument on the Extremely Large Telescope. In this process we have hit the limit of what can be achieved with a detailed software design that is primarily captured in pdf/word documents. Trade-offs between hardware and calibration software are required to meet stringent science requirements. To support such trade-offs, more software needs to be developed in the early phases of the project: simulators, archives, prototype recipes and pipelines. This requires continuous and efficient exchange of evolving designs between the software and hardware groups, which is hard to achieve with manually maintained documents. This, and maintaining the consistency between the design documents and various software components is possible with a machine readable version of the design. We construct a detailed design that is readable by both software and humans. From this the design documentation, prototype pipelines and data archives are generated automatically. We present the implementation of such an approach for the calibration software detailed design for the ELT MICADO imager which is based on expertise and lessons learned in earlier projects (e.g. OmegaCAM, MUSE, Euclid).