We investigate the impact of tokeniser pretraining on the accuracy and efficiency of physics emulation. Modern high-resolution simulations produce vast volumes of data spanning diverse physical regimes and scales. Training foundation models to learn the dynamics underlying such data enables the modelling of complex multiphysics phenomena, especially in data-limited settings. The emerging class of physics foundation models typically aims to learn two tasks jointly: (i) extracting compact representations of high-resolution spatiotemporal data, and (ii) capturing governing physical dynamics. However, learning both tasks from scratch simultaneously can impede the effectiveness of either process. We demonstrate that pretraining the tokeniser with an autoencoding objective prior to training the dynamics model enhances computational efficiency for downstream tasks. Notably, the magnitude of this benefit depends on domain alignment: pretraining on the same physical system as the downstream task yields the largest improvements, while pretraining on other systems provides moderate gains. In-domain pretraining reduces VRMSE by 64% after 10,500 training steps compared to training from scratch. To our knowledge, this is the first systematic investigation of tokeniser pretraining for physics foundation models. We further introduce flexible spatiotemporal compression operations that extend causal convolutions to support runtime-adjustable compression ratios, enabling efficient adaptation to diverse downstream tasks. Our findings provide practical guidance for training efficient physics emulators and highlight the importance of strategic pretraining data selection.
We present a comprehensive analysis of four near-Eddington black hole accretion models computed by solving the GRMHD equations with full radiation transport. This study focuses on the dynamical effects of magnetic field topology and black hole spin. Two stable near-Eddington solutions emerge in these models: a thin thermal disk embedded within a magnetic envelope when sufficient net vertical magnetic flux is present (e.g., vertical field $\gtrsim 5\times10^5$ G at $20r_g$), and a magnetically elevated disk when the net vertical flux is weak or absent. One model initialized without net vertical flux is found to evolve into the thin disk solution, as strong, anisotropic radiation feedback at high accretion rates promotes the accumulation of vertical magnetic flux in the inner disk. In the thin thermal disk, accretion is driven primarily by mean-field Maxwell stress and proceeds largely within the magnetic envelope, while heat dissipation is spatially decoupled and concentrated near the midplane. However, in the magnetically elevated disk, accretion occurs throughout the disk body and is comparably driven by mean-field and turbulent stresses; heat dissipation therefore occurs locally through turbulence. Radiation transport is diffusion-dominated, enabling efficient radiative cooling ($\sim$4-10%). An optically thin wind is launched from the disk surface by combined radiative and magnetic forces, with its strength increasing with black hole spin and vertical magnetic flux. Both strong and weak jets are produced in these models: strong jets are persistent, highly relativistic, and magnetically driven, while weak jets are intermittent, mildly relativistic, and powered by a combination of magnetic and radiative forces.
Nearly all previous binary black hole searches in LIGO--Virgo--KAGRA (LVK) gravitational wave data have assumed that the component spins are aligned with the orbital angular momentum, thereby neglecting spin-precession effects in the waveform, which can lead to potentially missing interesting signals. Precessing searches are challenging, because the extra degrees of freedom due to misaligned spins lead to: $(i)$ a much larger number of templates compared to the aligned-spin configurations, $(ii)$ an increased rate of background triggers. To address this, we develop novel precessing signal template banks using mode-by-mode filtering and marginalization methods. We use the precession harmonic decomposition from Fairhurst et al. (2019) and filter each precessing harmonic separately with the data. We then marginalize over the SNRs from different harmonics in our detection statistic. We also use machine learning methods to improve our search efficiency: $(i)$ we use singular value decomposition together with random forest regressor to reduce redundancy in the dominant precessing-harmonic templates; $(ii)$ we use normalizing flows to generate optimal prior samples for harmonic SNRs for the marginalized statistic. We show that marginalizing (instead of maximizing) over the harmonic mode SNRs increases the search sensitive volume by $\sim 10\%$. Results from searching in LVK data using this framework will be reported in a companion paper.
Symbiotic Binaries contain a white dwarf accreting material from a red giant star through a wind. We present the results of a search for outbursts from Symbiotic Binaries using photometric data obtained using the GOTO all-sky survey taken from 2023 onwards. After identifying ten candidate outbursting systems, we used ATLAS photometry to characterise their photometric behaviour before 2023, leaving five systems which showed photometric behaviour consistent with an outburst. The ATLAS data showed how important the photometric history of an object is in determining whether a photometric feature is a likely outburst event. The outburst from LMC N67 is the first reported Z And-type outburst from a Symbiotic binary in the LMC. OGLE SMC-LPV-4044 and HK Sco show previously unreported outbursts. QW Sge and V4141 Sgr show outbursts starting in 2024, which have already been reported and are ongoing. By better identifying and characterising Z And-type outbursts from many systems, it will be possible to better understand the physics of these events, which are still not fully understood.
Substantial amounts of air-shower simulations are needed to derive the instrument response for analyzing Imaging Air Cherenkov Telescope (IACT) data. This process is both computationally intensive and requires repetition under varying observation conditions, due to detector aging, changes in the atmosphere, or the instrument hardware. Generative models offer an efficient alternative, significantly accelerating simulations while compactly storing extensive simulation libraries, and providing a differentiable surrogate model of the instrument. However, their applicability has so far been limited in gamma-ray astronomy, particularly for modeling hadronic showers that dominate the background and exhibit significant intrinsic fluctuations that are challenging to model. In this study, we present the first application of score-based diffusion models to generate monoscopic $\gamma$-ray and proton shower images with nearly 2,000 pixels and benchmark the performance against Wasserstein GANs using H.E.S.S. simulations. We examine quality using both low-level parameters and well-established shower-shape observables, and assess analysis readiness via state-of-the-art $\gamma$-hadron separation. While GAN-based approaches can reproduce $\gamma$-ray showers with high fidelity, they fail to generate proton events of comparable quality, leading to a measurable degradation in analysis performance. In contrast, score-based diffusion modles achieve significantly superior quality for $\gamma$-ray and proton showers, accurately reproducing high-level correlations and generating events that are statistically indistinguishable from simulations at the analysis level. These results establish diffusion-based models as the first analysis-ready surrogate model of a single IACT, opening new prospects for fast instrument response generation, detector optimization, and connected downstream tasks.
this https URL . Accepted for publication in The Astrophysical Journal, volume 1000, issue 1, article 38. DOI: https://doi.org/10.3847/1538-4357/ae44f3
arXiv:2408.05210 , 5 pages, 1 figure, uses elsarticle