5632 research outputs found
Sort by
Graph neural networks at the Large Hadron Collider
No full textFrom raw detector activations to reconstructed particles, data at the Large Hadron Collider (LHC) are sparse, irregular, heterogeneous and highly relational in nature. Graph neural networks (GNNs), a class of algorithms belonging to the rapidly growing field of geometric deep learning (GDL), are well suited to tackling such data because GNNs are equipped with relational inductive biases that explicitly make use of localized information encoded in graphs. Furthermore, graphs offer a flexible and efficient alternative to rectilinear structures when representing sparse or irregular data, and can naturally encode heterogeneous information. For these reasons, GNNs have been applied to a number of LHC physics tasks including reconstructing particles from detector readouts and discriminating physics signals against background processes. We introduce and categorize these applications in a manner accessible to both physicists and non-physicists. Our explicit goal is to bridge the gap between the particle physics and GDL communities. After an accessible description of LHC physics, including theory, measurement, simulation and analysis, we overview applications of GNNs at the LHC. We conclude by highlighting technical challenges and future directions that may inspire further collaboration between the physics and GDL communities
Dominant Two-Dimensional Electron-Phonon Interactions in the Bulk Dirac Semimetal Na₃Bi
No full textBulk Dirac semimetals (DSMs) exhibit unconventional transport properties and phase transitions due to their peculiar low-energy band structure, yet the electronic interactions governing nonequilibrium phenomena in DSMs are not fully understood. Here we show that electron–phonon (e–ph) interactions in a prototypical bulk DSM, Na3Bi, are predominantly two-dimensional (2D). Our first-principles calculations reveal a 2D optical phonon with strong e–ph interactions associated with in-plane vibrations of Na atoms. We show that this 2D mode governs e–ph scattering and charge transport in Na₃Bi and induces a dynamical phase transition to a Weyl semimetal. Our work advances the quantitative analysis of electron interactions in Na₃Bi and reveals a dominant low-dimensional interaction in a bulk Dirac semimetal
The Giant Accreting Protoplanet Survey (GAPlanetS)—Results from a 6 yr Campaign to Image Accreting Protoplanets
No full textAccreting protoplanets are windows into planet formation processes, and high-contrast differential imaging is an effective way to identify them. We report results from the Giant Accreting Protoplanet Survey (GAPlanetS), which collected Hα differential imagery of 14 transitional disk host stars with the Magellan Adaptive Optics System. To address the twin challenges of morphological complexity and point-spread function instability, GAPlanetS required novel approaches for frame selection and optimization of the Karhounen–Loéve Image Processing algorithm pyKLIP. We detect one new candidate, CS Cha "c," at a separation of 68 mas and a modest Δmag of 2.3. We recover the HD 142527 B and HD 100453 B accreting stellar companions in several epochs, and the protoplanet PDS 70 c in 2017 imagery, extending its astrometric record by nine months. Though we cannot rule out scattered light structure, we also recover LkCa 15 "b," at Hα; its presence inside the disk cavity, absence in Continuum imagery, and consistency with a forward-modeled point source suggest that it remains a viable protoplanet candidate. Through targeted optimization, we tentatively recover PDS 70 c at two additional epochs and PDS 70 b in one epoch. Despite numerous previously reported companion candidates around GAplanetS targets, we recover no additional point sources. Our moderate Hα contrasts do not preclude most protoplanets, and we report limiting Hα contrasts at unrecovered candidate locations. We find an overall detection rate of ∼36⁺²⁶₋₂₂%, considerably higher than most direct imaging surveys, speaking to both GAPlanetS's highly targeted nature and the promise of Hα differential imaging for protoplanet identification
Orbital evolution of satellite galaxies in self-interacting dark matter models
No full textDark matter self-interactions can leave distinctive signatures on the properties of satellite galaxies around Milky Way–like hosts through their impact on tidal stripping, ram pressure, and gravothermal collapse. We delineate the regions of self-interacting dark matter parameter space — specified by interaction cross section and a velocity scale — where each of these effects dominates and show how the relative mass loss depends on the satellite’s initial mass, density profile, and orbit. We obtain novel, conservative constraints in this parameter space using Milky Way satellite galaxies with notably high central densities and small pericenter distances. Our results for self-interacting dark matter models, in combination with constraints from clusters of galaxies, favor either velocity-dependent cross sections that lead to gravothermal core collapse in the densest satellites or small cross sections which more closely resemble cold and collisionless dark matter
Data-driven segmentation of cortical calcium dynamics
No full textDemixing signals in transcranial videos of neuronal calcium flux across the cerebral hemispheres is a key step before mapping features of cortical organization. Here we demonstrate that independent component analysis can optimally recover neural signal content in widefield recordings of neuronal cortical calcium dynamics captured at a minimum sampling rate of 1.5×10⁶ pixels per one-hundred millisecond frame for seventeen minutes with a magnification ratio of 1:1. We show that a set of spatial and temporal metrics obtained from the components can be used to build a random forest classifier, which separates neural activity and artifact components automatically at human performance. Using this data, we establish functional segmentation of the mouse cortex to provide a map of ~115 domains per hemisphere, in which extracted time courses maximally represent the underlying signal in each recording. Domain maps revealed substantial regional motifs, with higher order cortical regions presenting large, eccentric domains compared with smaller, more circular ones in primary sensory areas. This workflow of data-driven video decomposition and machine classification of signal sources can greatly enhance high quality mapping of complex cerebral dynamics
Isotropic plasma-thermal atomic layer etching of aluminum nitride using SF₆ plasma and Al(CH₃)₃
No full textWe report the isotropic plasma atomic layer etching (ALE) of aluminum nitride using sequential exposures of SF₆ plasma and trimethylaluminum [Al(CH₃)₃]. ALE was observed at temperatures greater than 200∘C, with a maximum etch rate of 1.9 Å/cycle observed at 300∘C as measured using ex situ ellipsometry. After ALE, the etched surface was found to contain a lower concentration of oxygen compared to the original surface and exhibited a ∼35% decrease in surface roughness. These findings have relevance for applications of AlN in nonlinear photonics and wide bandgap semiconductor devices
The Origin and Evolution of the Normal Type Ia SN 2018aoz with Infant-phase Reddening and Excess Emission
No full textSN 2018aoz is a Type Ia SN with a B-band plateau and excess emission in infant-phase light curves ≲1 day after the first light, evidencing an over-density of surface iron-peak elements as shown in our previous study. Here, we advance the constraints on the nature and origin of SN 2018aoz based on its evolution until the nebular phase. Near-peak spectroscopic features show that the SN is intermediate between two subtypes of normal Type Ia: core normal and broad line. The excess emission may be attributable to the radioactive decay of surface iron-peak elements as well as the interaction of ejecta with either the binary companion or a small torus of circumstellar material. Nebular-phase limits on Hα and He ɪ favor a white dwarf companion, consistent with the small companion size constrained by the low early SN luminosity, while the absence of [O ɪ] and He ɪ disfavors a violent merger of the progenitor. Of the two main explosion mechanisms proposed to explain the distribution of surface iron-peak elements in SN 2018aoz, the asymmetric Chandrasekhar-mass explosion is less consistent with the progenitor constraints and the observed blueshifts of nebular-phase [Fe ɪɪ] and [Ni ɪɪ]. The helium-shell double-detonation explosion is compatible with the observed lack of C spectral features, but current 1D models are incompatible with the infant-phase excess emission, Bₘₐₓ-Vₘₐₓ color, and weak strength of nebular-phase [Ca ɪɪ]. Although the explosion processes of SN 2018aoz still need to be more precisely understood, the same processes could produce a significant fraction of Type Ia SNe that appear to be normal after ∼1 day
Image Reconstruction without Explicit Priors
No full textWe consider solving ill-posed imaging inverse problems without access to an explicit image prior or ground-truth examples. An overarching challenge in inverse problems is that there are many undesired images that fit to the observed measurements, thus requiring image priors to constrain the space of possible solutions to more plausible reconstructions. However, in many applications it is difficult or potentially impossible to obtain ground-truth images to learn an image prior. Thus, inaccurate priors are often used, which inevitably result in biased solutions. Rather than solving an inverse problem using priors that encode the explicit structure of any one image, we propose to solve a set of inverse problems jointly by incorporating prior constraints on the collective structure of the underlying images. The key assumption of our work is that the ground-truth images we aim to reconstruct share common, low-dimensional structure. We show that such a set of inverse problems can be solved simultaneously by learning a shared image generator with a low-dimensional latent space. The parameters of the generator and latent embedding are learned by maximizing a proxy for the Evidence Lower Bound (ELBO). Once learned, the generator and latent embeddings can be combined to provide reconstructions for each inverse problem. The framework we propose can handle general forward model corruptions, and we show that measurements derived from only a few ground-truth images (O(10)) are sufficient for image reconstruction without explicit priors
The Southern Ocean mixed layer and its boundary fluxes: fine-scale observational progress and future research priorities
No full textInteractions between the upper ocean and air-ice-ocean fluxes in the Southern Ocean play a critical role in global climate by impacting the overturning circulation and oceanic heat and carbon uptake. Remote and challenging conditions have led to sparse observational coverage, while ongoing field programmes often fail to collect sufficient information in the right place or at the time-space scales required to constrain the variability occurring in the coupled ocean-atmosphere system. Only within the last 10 years have we been able to directly observe and assess the role of the fine-scale ocean and rapidly evolving atmospheric marine boundary layer on the upper limb of the Southern Ocean's overturning circulation. This review summarizes advances in mechanistic understanding, arising in part from observational programmes using autonomous platforms, of the fine-scale processes (1–100 km, hours-seasons) influencing the Southern Ocean mixed layer and its variability. We also review progress in observing the ocean interior connections and the coupled interactions between the ocean, atmosphere and cryosphere that moderate air-sea fluxes of heat and carbon. Most examples provided are for the ice-free Southern Ocean, while major challenges remain for observing the ice-covered ocean. We attempt to elucidate contemporary research gaps and ongoing/future efforts needed to address them.
This article is part of a discussion meeting issue 'Heat and carbon uptake in the Southern Ocean: the state of the art and future priorities'
Partial Tidal Disruptions of Main-sequence Stars by Intermediate-mass Black Holes
No full textWe study close encounters of a 1 M_⊙ middle-age main-sequence star (modeled using MESA) with massive black holes through hydrodynamic simulations, and explore in particular the dependence of the outcomes on the black hole mass. We consider here black holes in the intermediate-mass range, MBH = 100–10⁴ M_⊙. Possible outcomes vary from a small tidal perturbation for weak encounters all the way to partial or full disruption for stronger encounters. We find that stronger encounters lead to increased mass loss at the first pericenter passage, in many cases ejecting the partially disrupted star on an unbound orbit. For encounters that initially produce a bound system, with only partial stripping of the star, the fraction of mass stripped from the star increases with each subsequent pericenter passage and a stellar remnant of finite mass is ultimately ejected in all cases. The critical penetration depth that separates bound and unbound remnants has a dependence on the black hole mass when M_(BH) ≲ 10³ M_⊙. We also find that the number of successive close passages before ejection decreases as we go from the stellar-mass black hole to the intermediate-mass black hole regime. For instance, after an initial encounter right at the classical tidal disruption limit, a 1 M_⊙ star undergoes 16 (5) pericenter passages before ejection from a 10 M_⊙ (100 M_⊙) black hole. Observations of periodic flares from these repeated close passages could in principle indicate signatures of a partial tidal disruption event