Search for: All records

Integrated Sensing and Communication (ISAC) enables ubiquitous Human Activity Recognition (HAR) but raises severe resource and raw data-localization challenges by centralizing raw sensing data. To address this, we propose an Explainable Split Federated Learning (SplitFL) framework for effective ISAC beamforming. Our architecture partitions the neural network at the edge, compressing raw signals into low-dimensional, non-invertible ``smashed data'' before transmission. Experimental results on the 6G Internet of Things (IoT) Intelligent Management Dataset demonstrate a 5.0$$\times$$ reduction in uplink payload and 99.95\% computation offloading to the server, resulting in an ultra-lightweight client footprint (0.36\,KB). The model achieves 97.0\% accuracy in small-cluster settings, while our novel Split Grad-CAM mechanism validates that decisions are driven by physical channel attributes rather than artifacts. Our method, SplitFL, is an efficient and transparent paradigm for next-generation HAR systems empowered by ISAC and the state-of-the-art distributed learning technology. 

The Vera C. Rubin Observatory’s Legacy Survey of Space and Time (LSST) is expected to obtain observations of over ten million quasars. The survey’s exceptional cadence and sensitivity will enable a significant fraction of these objects to be monitored in theugrizybands, spanning observed wavelengths of approximately 0.3 − 1.0 μm. The unprecedented number of sources makes spectroscopic follow-up for the vast majority of them unfeasible in the near future, so most studies will have to rely on photometric redshift estimates which are traditionally much less reliable for Active Galactic Nuclei (AGNs) than for inactive galaxies. This work presents a novel methodology to constrain the photometric redshift of AGNs that leverages the effects of cosmological time dilation, and of the luminosity and wavelength dependence of AGN variability. Specifically, we assume that the variability can be modeled as a damped random walk (DRW) process, and we adopted a parametric model to characterize the DRW timescale (τ) and asymptotic amplitude of the variability (SF_∞) based on the redshift, the rest-frame wavelength, and the AGN luminosity. We constructed variability-based photometric redshift (photo-z) priors by modeling the observed variability using the expected DRW parameters at a given redshift. These variability-based photo-z(VAR-PZ) priors were then combined with traditional spectral energy distribution (SED) fitting to improve the redshift estimates from SED fitting. Validation was performed using observational data from the Sloan Digital Sky Survey (SDSS), demonstrating significant reduction in catastrophic outliers by more than 10% in comparison with SED fitting techniques and improvements in redshift precision. The simulated light curves with both SDSS and LSST-like cadences and baselines confirm thatVAR-PZwill be able to constrain the photometric redshifts of SDSS-like AGNs by bringing the outlier fractions down to below 15% from 32% (SED alone) at the end of the survey. 

Abstract In situ measurements of coronal mass ejections (CMEs) when they pass over an interplanetary probe are one of the main ways we directly measure their properties. However, such in situ profiles are subject to several observational constraints that are still poorly understood. This work aims at quantifying one of them, namely, the aging effect, using a CME simulated with a three-dimensional magnetohydrodynamical code. The synthetic in situ profile and the instantaneous profile of the magnetic field strength differ more from each other when taken close to the Sun than far from it. Moreover, out of three properties we compute in this study (i.e., size, distortion parameter, and expansion speed), only the expansion speed shows a dependence of the aging as a function of distance. It is also the property that is the most impacted by the aging effect as it can amount to more than 100 km s⁻¹for CMEs observed closer than 0.15 au. This work calls for caution when deducing the expansion speed from CME profiles when they still are that close to the Sun since the aging effect can significantly impact the derived properties. 

Abstract Simultaneous in situ measurements of coronal mass ejections (CMEs), including both plasma and magnetic field, by two spacecraft in radial alignment have been extremely rare. Here, we report on one such CME measured by Solar Orbiter (SolO) and Wind on 2021 November 3–5, while the spacecraft were radially separated by a heliocentric distance of 0.13 au and angularly by only 2.2°. We focus on the magnetic cloud (MC) part of the CME. We find notable changes in theRandNmagnetic field components and in the speed profiles inside the MC between SolO and Wind. We observe a greater speed at the spacecraft farther away from the Sun without any clear compression signatures. Since the spacecraft are close to each other and computing fast magnetosonic wave speed inside the MC, we rule out temporal evolution as the reason for the observed differences, suggesting that spatial variations over 2.2° of the MC structure are at the heart of the observed discrepancies. Moreover, using shock properties at SolO, we forecast an arrival time 2 hr 30 minutes too late for a shock that is just 5 hr 31 minutes away from Wind. Predicting the north–south component of the magnetic field at Wind from SolO measurements leads to a relative error of 55%. These results show that even angular separations as low as 2.2° (or 0.03 au in arc length) between spacecraft can have a large impact on the observed CME properties, which raises the issue of the resolutions of current CME models, potentially affecting our forecasting capabilities. 

Context.Coronal mass ejections (CMEs) are large-scale structures of magnetized plasma that erupt from the corona into interplanetary space. The launch of Solar Orbiter (SolO) in 2020 enables in situ measurements of CMEs in the innermost heliosphere, at such distances where CMEs can be observed remotely within the inner field of view of heliospheric imagers (HIs). It thus provides the opportunity for investigations into the correspondence of the CME substructures measured in situ and observed remotely. We studied a CME that started on 2022 March 10 and was measured in situ by SolO at ∼0.44 au. Aims.Combining remote observations of CMEs from wide-angle imagers and in situ measurements in the innermost heliosphere allows us to compare CME properties derived through both techniques, validate the estimates, and better understand CME evolution, specifically the size and radial expansion, within 0.5 au. Methods.We compared the evolution of different CME substructures observed in images from the HIs on board the Ahead Solar Terrestrial Relations Observatory (STEREO-A) and the CME signatures measured in situ by SolO. The CME is found to possess a density enhancement at its rear edge in both remote and in situ observations, which validates the use of the signature of density enhancement following the CMEs to accurately identify the CME rear edge. We also estimated and compared the radial size and radial expansion speed of different substructures in both observations. Results.The evolution of the CME front and rear edges in remote images is consistent with the in situ CME measurements. The radial expansion (i.e., radial size and radial expansion speed) of the whole CME structure consisting of the magnetic ejecta and the sheath is consistent with the in situ estimates obtained at the same time from SolO. However, we do not find such consistencies for the magnetic ejecta region inside the CME because it is difficult to identify the magnetic ejecta edges in the remote images. 

Abstract In situ measurements from spacecraft typically provide a time series at a single location through coronal mass ejections (CMEs), and they have been one of the main methods to investigate CMEs. The CME properties derived from these in situ measurements are affected by temporal changes that occur as the CME passes over the spacecraft, such as radial expansion and aging, as well as spatial variations within a CME. This study uses multispacecraft measurements of the same CME at close separations to investigate both the spatial variability (how different a CME profile is when probed by two spacecraft close to each other) and the so-called aging effect (the effect of the time evolution on in situ properties). We compile a database of 19 events from the past 4 decades measured by two spacecraft with a radial separation of <0.2 au and an angular separation of <10°. We find that the average magnetic field strength measured by the two spacecraft differs by 18% of the typical average value, which highlights nonnegligible spatial or temporal variations. For one particular event, measurements taken by the two spacecraft allow us to quantify and significantly reduce the aging effect to estimate the asymmetry of the magnetic field strength profile. This study reveals that single-spacecraft time series near 1 au can be strongly affected by aging and that correcting for self-similar expansion does not capture the whole aging effect. 

The BigCode community, an open-scientific collaboration working on the responsible development of Large Language Models for Code (Code LLMs), introduces StarCoder and StarCoderBase: 15.5B parameter models with 8K context length, infilling capabilities and fast large-batch inference enabled by multi-query attention. StarCoderBase is trained on 1 trillion tokens sourced from The Stack, a large collection of permissively licensed GitHub repositories with inspection tools and an opt-out process. We fine-tuned StarCoderBase on 35B Python tokens, resulting in the creation of StarCoder. We perform the most comprehensive evaluation of Code LLMs to date and show that StarCoderBase outperforms every open Code LLM that supports multiple programming languages and matches or outperforms the OpenAI code-cushman-001 model. Furthermore, StarCoder outperforms every model that is fine-tuned on Python, can be prompted to achieve 40% pass@1 on HumanEval, and still retains its performance on other programming languages. We take several important steps towards a safe open-access model release, including an improved PII redaction pipeline and a novel attribution tracing tool, and make the StarCoder models publicly available under a more commercially viable version of the Open Responsible AI Model license. 

Abstract The dispersive sweep of fast radio bursts (FRBs) has been used to probe the ionized baryon content of the intergalactic medium 1 , which is assumed to dominate the total extragalactic dispersion. Although the host-galaxy contributions to the dispersion measure appear to be small for most FRBs 2 , in at least one case there is evidence for an extreme magneto-ionic local environment 3,4 and a compact persistent radio source 5 . Here we report the detection and localization of the repeating FRB 20190520B, which is co-located with a compact, persistent radio source and associated with a dwarf host galaxy of high specific-star-formation rate at a redshift of 0.241 ± 0.001. The estimated host-galaxy dispersion measure of approximately $${903}_{-111}^{+72}$$ 903 − 111 + 72 parsecs per cubic centimetre, which is nearly an order of magnitude higher than the average of FRB host galaxies 2,6 , far exceeds the dispersion-measure contribution of the intergalactic medium. Caution is thus warranted in inferring redshifts for FRBs without accurate host-galaxy identifications.