Abstract Suppose we are given a system of coupled oscillators on an unknown graph along with the trajectory of the system during some period. Can we predict whether the system will eventually synchronize? Even with a known underlying graph structure, this is an important yet analytically intractable question in general. In this work, we take an alternative approach to the synchronization prediction problem by viewing it as a classification problem based on the fact that any given system will eventually synchronize or converge to a non-synchronizing limit cycle. By only using some basic statistics of the underlying graphs such as edge density and diameter, our method can achieve perfect accuracy when there is a significant difference in the topology of the underlying graphs between the synchronizing and the non-synchronizing examples. However, in the problem setting where these graph statistics cannot distinguish the two classes very well (e.g., when the graphs are generated from the same random graph model), we find that pairing a few iterations of the initial dynamics along with the graph statistics as the input to our classification algorithms can lead to significant improvement in accuracy; far exceeding what is known by the classical oscillator theory. More surprisingly, we find that in almost all such settings, dropping out the basic graph statistics and training our algorithms with only initial dynamics achieves nearly the same accuracy. We demonstrate our method on three models of continuous and discrete coupled oscillators—the Kuramoto model, Firefly Cellular Automata, and Greenberg-Hastings model. Finally, we also propose an “ensemble prediction” algorithm that successfully scales our method to large graphs by training on dynamics observed from multiple random subgraphs.
more »
« less
This content will become publicly available on August 23, 2024
Firefly: The Case for a Holistic Understanding of the Global Structure and Dynamics of the Sun and the Heliosphere
This white paper is on the HMCS Firefly mission concept study. Firefly focuses on the global structure and dynamics of the Sun's interior, the generation of solar magnetic fields, the deciphering of the solar cycle, the conditions leading to the explosive activity, and the structure and dynamics of the corona as it drives the heliosphere.
more »
« less
- Award ID(s):
- 2229336
- NSF-PAR ID:
- 10495007
- Author(s) / Creator(s):
- ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; more »
- Publisher / Repository:
- Bulletin of the American Astronomical Society
- Date Published:
- Journal Name:
- Bulletin of the AAS
- ISSN:
- 00027537
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
We propose firefly neural architecture descent, a general framework for progressively and dynamically growing neural networks to jointly optimize the networks' parameters and architectures. Our method works in a steepest descent fashion, which iteratively finds the best network within a functional neighborhood of the original network that includes a diverse set of candidate network structures. By using Taylor approximation, the optimal network structure in the neighborhood can be found with a greedy selection procedure. We show that firefly descent can flexibly grow networks both wider and deeper, and can be applied to learn accurate but resource-efficient neural architectures that avoid catastrophic forgetting in continual learning. Empirically, firefly descent achieves promising results on both neural architecture search and continual learning. In particular, on a challenging continual image classification task, it learns networks that are smaller in size but have higher average accuracy than those learned by the state-of-the-art methods.more » « less
-
null (Ed.)NanoLuc is a bioluminescent protein recently engineered for applications in molecular imaging and cellular reporter assays. Compared to other bioluminescent proteins used for these applications, like Firefly Luciferase and Renilla Luciferase, it is ~150 times brighter, more thermally stable, and smaller. Yet, no information is known with regards to its mechanical properties, which could introduce a new set of applications for this unique protein, such as a novel biomaterial or as a substrate for protein activity/refolding assays. Here, we generated a synthetic NanoLuc derivative protein that consists of three connected NanoLuc proteins flanked by two human titin I91 domains on each side and present our mechanical studies at the single molecule level by performing Single Molecule Force Spectroscopy (SMFS) measurements. Our results show each NanoLuc repeat in the derivative behaves as a single domain protein, with a single unfolding event occurring on average when approximately 72 pN is applied to the protein. Additionally, we performed cyclic measurements, where the forces applied to a single protein were cyclically raised then lowered to allow the protein the opportunity to refold: we observed the protein was able to refold to its correct structure after mechanical denaturation only 16.9% of the time, while another 26.9% of the time there was evidence of protein misfolding to a potentially non-functional conformation. These results show that NanoLuc is a mechanically moderately weak protein that is unable to robustly refold itself correctly when stretch-denatured, which makes it an attractive model for future protein folding and misfolding studies.more » « less
-
more » « less
Abstract We present Firefly, a new browser-based interactive tool for visualizing 3D particle data sets. On a typical personal computer, Firefly can simultaneously render and enable real-time interactions with ≳10 million particles, and can interactively explore data sets with billions of particles using the included custom-built octree render engine. Once created, viewing a Firefly visualization requires no installation and is immediately usable in most modern internet browsers simply by visiting a URL. As a result, a Firefly visualization works out-of-the-box on most devices including smartphones and tablets. Firefly is primarily developed for researchers to explore their own data, but can also be useful to communicate results to researchers and/or collaborators and as an effective public outreach tool. Every element of the user interface can be customized and disabled, enabling easy adaptation of the same visualization for different audiences with little additional effort. Creating a new Firefly visualization is simple with the provided Python data preprocessor that translates input data to a Firefly-compatible format and provides helpful methods for hosting instances of Firefly both locally and on the internet. In addition to visualizing the positions of particles, users can visualize vector fields (e.g., velocities) and also filter and color points by scalar fields. We share three examples of Firefly applied to astronomical data sets: (1) the FIRE cosmological zoom-in simulations, (2) the SDSS galaxy catalog, and (3) Gaia Data Release 3. A gallery of additional interactive demos is available at
alexbgurvi.ch/Firefly . -
Correlated Spatio-temporal Evolution of Extreme-Ultraviolet Ribbons and Hard X-Rays in a Solar Flaremore » « less
Abstract We analyze the structure and evolution of ribbons from the M7.3 SOL2014-04-18T13 flare using ultraviolet images from the Interface Region Imaging Spectrograph and the Solar Dynamics Observatory (SDO)/Atmospheric Imaging Assembly (AIA), magnetic data from the SDO/Helioseismic and Magnetic Imager, hard X-ray (HXR) images from the Reuven Ramaty High Energy Solar Spectroscopic Imager, and light curves from the Fermi/Gamma-ray Burst Monitor, in order to infer properties of coronal magnetic reconnection. As the event progresses, two flare ribbons spread away from the magnetic polarity inversion line. The width of the newly brightened front along the extension of the ribbon is highly intermittent in both space and time, presumably reflecting nonuniformities in the structure and/or dynamics of the flare current sheet. Furthermore, the ribbon width grows most rapidly in regions exhibiting concentrated nonthermal HXR emission, with sharp increases slightly preceding the HXR bursts. The light curve of the ultraviolet emission matches the HXR light curve at photon energies above 25 keV. In other regions the ribbon-width evolution and light curves do not temporally correlate with the HXR emission. This indicates that the production of nonthermal electrons is highly nonuniform within the flare current sheet. Our results suggest a strong connection between the production of nonthermal electrons and the locally enhanced perpendicular extent of flare ribbon fronts, which in turn reflects the inhomogeneous structure and/or reconnection dynamics of the current sheet. Despite this variability, the ribbon fronts remain nearly continuous, quasi-one-dimensional features. Thus, although the reconnecting coronal current sheets are highly structured, they remain quasi-two-dimensional and the magnetic energy release occurs systematically, rather than stochastically, through the volume of the reconnecting magnetic flux.
