Description
Organisers: Sarah Johnston, Alastair Basden, Carlton Baugh, Sownak Bose
Modern computing capabilities are allowing us to delve further and learn more about our universe. Ongoing developments to simulations allow us to model larger problems and do so in greater detail than ever before, and our instrumentation advancements are leading to telescopes which produce massive amounts of data. To accommodate this, astronomy codes are relying more on High Throughput Computing or High Performance Computing systems to support their workload. With this move to highly parallel, data or memory intensive systems, the world of astronomy codes and tool-chains must adapt to follow the trend. With more and more codes looking to reach the petascale and exascale regime, utilising powerful compute systems is becoming more important.
We also want to ensure the long-term usability of our codes and this includes making sure it is forward compatible with future systems e.g. by fully utilising available hardware and leveraging the highest efficiency and performance. There are also increasing environmental impacts to consider when it comes to making sure the future of supercomputing is sustainable.
With many different approaches being taken across the community from the use of GPUs and Machine Learning, to novel infrastructure approaches, this session aims to bring code and software developers together from across astronomy. This includes updates and discussions on the future of computing within astronomy from a technical perspective. This could be current code development or porting initiatives, data management or processing pipelines, or sustainability efforts within the computational community. Submissions from active developers and ECRs are particularly encouraged.
Dyablo is a new high-performance hydrodynamical code designed for large-scale astrophysical simulations, from solar system dynamics to galaxy formation and cosmic reionization. Unlike many existing codes, Dyablo has been designed from scratch to run entirely on the available hardware, whether NVIDIA and AMD GPUs or CPUs accelerated with OpenMP. This architecture enables significant potential...
We present ExoSim 2, a next-generation, fully modular simulator for astronomical observations, initially developed to support the study of transiting exoplanets in the context of the Ariel space mission. Implemented in Python 3 and built on an object-oriented design, ExoSim 2 is publicly available and offers a flexible, open framework for simulating time-resolved spectro-photometric data from...
Radiative transfer is a cornerstone of astrophysics, providing a key tool to model and interpret observations of distant structures, for most of which any form of in situ measurement in impossible. Whilst simplifying approximations are often possible, there are many instances where the observed radiation forms in optically thick plasma outside of local thermodynamic equilibrium (LTE)...
As observational frontiers push into new territory, we must make apples-to-apples comparisons when comparing theory to observation to ensure robust conclusions are drawn. This means transforming models into the observer frame, where we can pinpoint observational biases, explore uncertain parameter spaces, and demystify what โmissing physicsโ really means (if anything). This process requires...
High-performance N-body simulations are crucial for astrophysics, cosmology, and computational physics. Odisseo (https://github.com/vepe99/Odisseo) is a differentiable N-body simulator designed for efficiency, scalability, and long-term usability. Implemented in JAX, it leverages just-in-time (JIT) compilation, automatic differentiation, and GPU/TPU acceleration
to provide fast, vectorized,...
We present jf1uids (https://github.com/leo1200/jf1uids), a differentiable magnetohydrodynamics (MHD) simulator written in JAX, scaling to multiple GPUs. Our open source code features a modern inherently divergence free approach to MHD, (near-)energy-conserving self-gravity stable at discontinuities, a conservative geometric formulation for radially symmetric simulations and multiple physics...
The New Robotic Telescope (NRT) represents a significant advancement in autonomous astronomical observation, with core science goals requiring response times under 30 seconds for transient events. Traditional scheduling approaches often rely on fixed priority rankings that inadequately address time-sensitive observations, especially as astronomical surveys like LSST generate unprecedented...
ExaGRyPE is a suite of solvers for numerical relativity based on ExaHyPE 2, our second-generation Exascale Hyperbolic PDE Engine. This solver tackles the Einstein equations/relativistic hydrodynamics equations under a 3+1 foliation, with a focus on compact object spacetimes. The implementation utilizes a block-structured Cartesian grid with higher-order Finite Difference schemes and adaptive...
Developing software for astronomical data analysis is fundamental for both observers and theorists. As big data and large collaborations become the norm, astronomers must prioritise statistically rigorous, reliable, and maintainable software. High-impact research increasingly relies on open, reusable code, with 36% of [recent Physics and Astronomy papers][1] referencing software. Journals are...
Non-equilibrium thermo-chemistry plays a crucial role in shaping the properties of the interstellar medium, from galactic to protoplanetary scales, particularly within molecular clouds. However, accurately modeling its effects in astrophysical simulations remains a significant challenge due to the complexity of the associated systems of ODEs, with chemistry often being by far the most...
As a result of being exposed to the night sky, Imaging Atmospheric Cherenkov Telescopes (IACTs), designed to observe astrophysical gamma-rays, are sensitive to background optical-wavelength illumination. This Night Sky Background (NSB) limits the operational time of most IACTs, introduces systematic uncertainty, and is a source of data/Monte-Carlo mismatch during event reconstruction. Building...
Over ten years after the survey was concluded, research into the WISE catalogue is still being undertaken. However, one area that has been lacking exploration is applying unsupervised machine learning to the catalogue, especially in the case of outlier detection. Using unsupervised machine learning for outlier detection can be a very powerful tool, allowing you to find underlying structure and...
Findable, accessible, reliable and repeatable. These words describe imperative attributes of software necessary for evaluating modern science questions. In this talk, I will present open-source code, SimSpin, written to generate mock observations of simulated galaxies in a format consistent with integral field spectroscopic (IFS) observations. This open-source software has been written using...
Identification of tidal streams in galaxy images is a difficult task as they are very faint compared to other stellar components. Additionally, the large influx of images expected from upcoming large-scale sky surveys will make it increasingly unrealistic to expect to be able to identify stellar streams by visual examination alone.
Here we will present a machine learning classifier trained...
Recent spectral observatories stand to revolutionise our ability to study exoplanets on a larger population scale than ever before. Analysing this data requires extracting information about the planetary atmosphere from the spectra. For anything outside a small number of targets this is very computationally resource intensive, which is a large barrier to entry as we move into larger scale...
