Search

We describe how the Fink broker early supernova Ia classifier optimizes its ML classifications by employing an active learning (AL) strategy. We demonstrate the feasibility of implementation of such strategies in the current Zwicky Transient Facility (ZTF) public alert data stream. We compare the performance of two AL strategies: uncertainty sampling and random sampling. Our pipeline consists of 3 stages: feature extraction, classification and learning strategy. Starting from an initial sample of 10 alerts (5 SN Ia and 5 non-Ia), we let the algorithm identify which alert should be added to the training sample. The system is allowed to evolve through 300 iterations. Our data set consists of 23 840 alerts from the ZTF with confirmed classification via cross-match with SIMBAD database and the Transient name server (TNS), 1 600 of which were SNe Ia (1 021 unique objects). The data configuration, after the learning cycle was completed, consists of 310 alerts for training and 23 530 for testing. Averaging over 100 realizations, the classifier achieved 89% purity and 54% efficiency. From 01/November/2020 to 31/October/2021 Fink has applied its early supernova Ia module to the ZTF stream and communicated promising SN Ia candidates to the TNS. From the 535 spectroscopically classified Fink candidates, 459 (86%) were proven to be SNe Ia. Our results confirm the effectiveness of active learning strategies for guiding the construction of optimal training samples for astronomical classifiers. It demonstrates in real data that the performance of learning algorithms can be highly improved without the need of extra computational resources or overwhelmingly large training samples. This is, to our knowledge, the first application of AL to real alerts data., Comment: accepted for publication in Astronomy and Astrophysics

Fink is a broker designed to enable science with large time-domain alert streams such as the one from the upcoming Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST). It exhibits traditional astronomy broker features such as automatised ingestion, annotation, selection and redistribution of promising alerts for transient science. It is also designed to go beyond traditional broker features by providing real-time transient classification which is continuously improved by using state-of-the-art Deep Learning and Adaptive Learning techniques. These evolving added values will enable more accurate scientific output from LSST photometric data for diverse science cases while also leading to a higher incidence of new discoveries which shall accompany the evolution of the survey. In this paper we introduce Fink, its science motivation, architecture and current status including first science verification cases using the Zwicky Transient Facility alert stream., Comment: accepted in MNRAS

We study two types of active (self-propelled) macroscopic particles under confinement: camphor surfers and hexbug crawlers, using a combined experimental, theoretical, and numerical approach. Unlike widely studied microscopic active particles and swimmers, where thermal forces are often important and inertia is negligible, our macroscopic particles exhibit complex dynamics due expressly to active non-thermal noise combined with inertial effects. Strong confinement induces accumulation at a finite distance within the boundary and gives rise to three distinguishable dynamical states; both depending on activity and inertia. These surprisingly complex dynamics arise already at the single particle level -- highlighting the importance of inertia in macroscopic active matter., Comment: 8 pages, 6 figures

We examine the interplay of motility and information exchange in a model of run-and-tumble active particles where the particle's motility is encoded as a bit of information that can be exchanged upon contact according to the rules of AND and OR logic gates in a circuit. Motile AND particles become non-motile upon contact with a non-motile particle. Conversely, motile OR particles remain motile upon collision with their non-motile counterparts. AND particles that have become non-motile additionally "reawaken", i.e., recover their motility, at a fixed rate $\mu$, as in the SIS (Susceptible, Infected, Susceptible) model of epidemic spreading, where an infected agent can become healthy again, but keeps no memory of the recent infection, hence it is susceptible to a renewed infection. For $\mu=0$, both AND and OR particles relax irreversibly to absorbing states of all non-motile or all motile particles, respectively. The relaxation kinetics is, however, faster for OR particles that remain active throughout the process. At finite $\mu$, the AND dynamics is controlled by the interplay between reawakening and collision rates. The system evolves to a state of all motile particles (an absorbing state in the language of absorbing phase transitions) for $\mu>\mu_c$ and to a mixed state with coexisting motile and non-motile particles (an active state in the language of absorbing phase transitions) for $\mu<\mu_c$. The final state exhibits a rich structure controlled by motility-induced aggregation. Our work can be relevant to biochemical signaling in motile bacteria, the spreading of epidemics and of social consensus, as well as light-controlled organization of active colloids.

We study active run-and-tumble particles with an additional two-state internal variable characterizing their motile or non-motile state. Motile particles change irreversibly into non-motile ones upon collision with a non-motile particle. The system evolves towards an absorbing state where all particles are non-motile. We initialize the system with one non-motile particles in a bath of motile ones and study numerically the kinetics of relaxation to absorbing state and its structure as function of the density of the initial bath of motile particles and of their tumbling rate. We find a crossover from fractal aggregates at low density to homogeneous ones at high density. The persistence of single-particle dynamics as quantified by the tumbling rate pushes this crossover to higher density and can be used to tune the porosity of the aggregate. At the lowest density the fractal dimension of the aggregate approaches that obtained in single-particle diffusion limited aggregation. Our results could be exploited for the design of structures of desired porosity. The model is a first step towards the study of the collective dynamics of active particles that can exchange biological information., Comment: 6 pages, 4 figures

Building on the striking similarity between the structure of the spindle during mitosis in living cells and nematic textures in confined liquid crystals, we use a continuum model of two-dimensional nematic liquid crystal droplets, to examine the physical aspects of cell division. The model investigates the interplay between bulk elasticity of the microtubule assembly, described as a nematic liquid crystal, and surface elasticity of the cell cortex, modelled as a bounding flexible membrane, in controlling cell shape and division. The centrosomes at the spindle poles correspond to the cores of the topological defects required to accommodate nematic order in a closed geometry. We map out the progression of both healthy bipolar and faulty multi-polar division as a function of an effective parameter that incorporates active processes and controls centrosome separation. A robust prediction, independent of energetic considerations, is that the transition from a single cell to daughters cells occurs at critical value of this parameter. Our model additionally suggests that microtubule anchoring at the cell cortex may play an important role for successful bipolar division. This can be tested experimentally by regulating microtubule anchoring., Comment: 10 pages, 7 figures

Locomotion and generation of flow at low Reynolds number are subject to severe limitations due to the irrelevance of inertia: the "scallop theorem" requires that the system have at least two degrees of freedom, which move in non-reciprocal fashion, i.e. breaking time-reversal symmetry. We show here that a minimal model consisting of just two spheres driven by harmonic potentials is capable of generating flow. In this pump system the two degrees of freedom are the mean and relative positions of the two spheres. We have performed and compared analytical predictions, numerical simulation and experiments, showing that a time-reversible drive is sufficient to induce flow., Comment: 5 pages, 3 figures, revised version, corrected typos

We construct a family of 1-convex threefolds, with exceptional curve C of type (0,-2), which are not embeddable in C^m \times CP_n. In order to show that they are not Kaehler we exhibit a real 3-dimensional chain A whose boundary is the complex curve C., Comment: 7 pages

We extend a result of to Esnault-Levine-Viehweg concerning the Chow groups of hypersurfaces in projective space to those in weighted projective spaces.