📢 TipTop v1.5.1, P3 v1.6.2 and MASTSEL v1.5.2 released!
TipTop v1.5.1, P3 v1.6.2 and MASTSEL v1.5.2 are now available! 🎉
This release focuses on improving the software architecture and interoperability of the TipTop ecosystem while reducing latency for service-oriented deployments. These developments provide the software foundations required to support the growing range of scientific and operational applications built around TipTop, including exposure-time calculation, observation preparation, and future PSF reconstruction services.
✨ What's new
🏗️ Modular software architecture
The main evolution of this release is the introduction of an abstract simulation layer through the new AbstractSimulation base class.
The simulation workflow—configuration handling, orchestration, FITS generation, PSF profiles, and performance metrics—is now implemented once within the common infrastructure, while individual simulation backends only need to implement the physical computations specific to their analytical model.
This modular design considerably simplifies maintenance, reduces code duplication, and makes it straightforward to integrate alternative PSF engines while preserving the same user interface.
The first applications of this architecture are already underway with:
- TipTorch, a differentiable analytical PSF model
- AOPERA (coming soon), providing advanced descriptions of Fourier-filtering wavefront sensors and optical-gain effects.
Developers creating their own backend only need to implement a small number of abstract methods while reusing the complete TipTop infrastructure.
🔄 Improved interoperability
TipTop, P3 and MASTSEL now exchange their configuration directly through a shared config_dict object instead of relying on configuration files and path discovery.
This simplification:
- reduces unnecessary I/O,
- removes the need for automatic path detection,
- improves robustness,
- simplifies embedding TipTop inside external applications and services.
This mechanism is now used throughout the TipTop ecosystem.
⚡ Faster service deployments
P3 now supports optional GPU warm-up through two environment variables:
export P3_GPU_WARMUP=TRUE
When enabled, CuPy is initialised during package import by running a lightweight dummy simulation. Subsequent simulations therefore start immediately without paying the usual GPU initialisation overhead.
This feature is particularly useful for low-latency services such as exposure-time calculators or observation-preparation tools.
🧮 Numerical improvements
Several numerical improvements have also been introduced in MASTSEL, including:
- improved handling of multi-wavelength PSF generation,
- more robust support for odd and even sampling grids,
- various performance optimisations.
These changes further improve the stability and reliability of the analytical PSF generation pipeline.
🚀 Looking ahead
This release represents an important step in the evolution of TipTop from a standalone AO PSF simulator into a modular analytical framework supporting a broad ecosystem of scientific and operational applications.
Future releases will continue extending both the analytical capabilities and the operational deployment of the framework.
🚀 Update now
pip install --upgrade astro-tiptop
Check it out on PyPI 👉