Gartner, Thomas III; Zhang, Linfeng; Piaggi, Pablo; Car, Roberto; Panagiotopoulos, Athanassios; Debenedetti, Pablo
Abstract:
This dataset contains all data related to the publication "Signatures of a liquid-liquid transition in an ab initio deep neural network model for water", by Gartner et al., 2020. In this work, we used neural networks to generate a computational model for water using high-accuracy quantum chemistry calculations. Then, we used advanced molecular simulations to demonstrate evidence that suggests this model exhibits a liquid-liquid transition, a phenomenon that can explain many of water's anomalous properties. This dataset contains links to all software used, all data generated as part of this work, as well as scripts to generate and analyze all data and generate the plots reported in the publication.
The multi-scale, mutli-physics nature of fusion plasmas makes predicting plasma events challenging. Recent advances in deep convolutional neural network architectures (CNN) utilizing dilated convolutions enable accurate predictions on sequences which have long-range, multi-scale characteristics, such as the time-series generated by diagnostic instruments observing fusion plasmas. Here we apply this neural network architecture to the popular problem of disruption prediction in fusion tokamaks, utilizing raw data from a single diagnostic, the Electron Cyclotron Emission imaging (ECEi) diagnostic from the DIII-D tokamak. ECEi measures a fundamental plasma quantity (electron temperature) with high temporal resolution over the entire plasma discharge, making it sensitive to a number of potential pre-disruptions markers with different temporal and spatial scales. Promising, initial disruption prediction results are obtained training a deep CNN with large receptive field ({$\sim$}30k), achieving an $F_1$-score of {$\sim$}91\% on individual time-slices using only the ECEi data.
The data provided in this DataSpace consists of sample training data to be used for Fluorescence Reconstruction Microscopy (FRM) testing. We provide a subset of the keratinocyte (10x magnification) dataset used in our paper, in which interested parties may find more complete information about our data collection methods. Matched pairs of phase contrast and fluorescent images are given. The nuclei were stained using Hoechst 33342 and imaged using a standard DAPI filter set.
The data provided in this DataSpace consists of sample training data to be used for Fluorescence Reconstruction Microscopy (FRM) testing. We provide a subset of the MDCK (20x magnification) dataset used in our paper, in which interested parties may find more complete information about our data collection methods. Matched pairs of DIC and fluorescent images are given. The cells stably expressed E-cadherin:RFP which enabled imaging of junctional fluorescence, while the nuclei were stained using Hoechst 33342 and imaged using a standard DAPI filter set.
We provide all the test data and corresponding predictions for our paper, “Practical Fluorescence Reconstruction Microscopy for High-Content Imaging”. Please refer to the Methods section in this paper for experimental details. For each experimental condition, we provide the input transmitted-light images (either phase contrast or DIC), the ground truth fluorescence images, and the output predicted fluorescence images which should reconstruct the ground truth fluorescence images.
Woods, B. J. Q.; Duarte, V. N.; Fredrickson, E. D.; Gorelenkov, N. N.; Podestà, M.; Vann, R. G. L.
Abstract:
Abrupt large events in the Alfvenic and sub-Alfvenic frequency bands in tokamaks are typically correlated with increased fast-ion loss. Here, machine learning is used to speed up the laborious process of characterizing the behavior of magnetic perturbations from corresponding frequency spectrograms that are typically identified by humans. The analysis allows for comparison between different mode character (such as quiescent, fixed frequency, and chirping, avalanching) and plasma parameters obtained from the TRANSP code, such as the ratio of the neutral beam injection (NBI) velocity and the Alfven velocity (v_inj./v_A), the q-profile, and the ratio of the neutral beam beta and the total plasma beta (beta_beam,i / beta). In agreement with the previous work by Fredrickson et al., we find a correlation between beta_beam,i and mode character. In addition, previously unknown correlations are found between moments of the spectrograms and mode character. Character transition from quiescent to nonquiescent behavior for magnetic fluctuations in the 50200-kHz frequency band is observed along the boundary v_phi ~ (1/4)(v_inj. - 3v_A), where v_phi is the rotation velocity.
One of the most promising devices for realizing power production through nuclear fusion is the tokamak. To maximize performance, it is preferable that tokamak reactors achieve advanced operating scenarios characterized by good plasma confinement, improved magnetohydrodynamic (MHD) stability, and a largely non-inductively driven plasma current. Such scenarios could enable steady-state reactor operation with high \emph{fusion gain} --- the ratio of produced fusion power to the external power provided through the plasma boundary. Precise and robust control of the evolution of the plasma boundary shape as well as the spatial distribution of the plasma current, density, temperature, and rotation will be essential to achieving and maintaining such scenarios. The complexity of the evolution of tokamak plasmas, arising due to nonlinearities and coupling between various parameters, motivates the use of model-based control algorithms that can account for the system dynamics. In this work, a learning-based accelerated model trained on data from the National Spherical Torus Experiment Upgrade (NSTX-U) is employed to develop planning and control strategies for regulating the density and temperature profile evolution around desired trajectories. The proposed model combines empirical scaling laws developed across multiple devices with neural networks trained on empirical data from NSTX-U and a database of first-principles-based computationally intensive simulations. The reduced execution time of the accelerated model will enable practical application of optimization algorithms and reinforcement learning approaches for scenario planning and control development. An initial demonstration of applying optimization approaches to the learning-based model is presented, including a strategy for mitigating the effect of leaving the finite validity range of the accelerated model. The approach shows promise for actuator planning between experiments and in real-time.
This dataset contains input and output files to reproduce the results of the manuscript "Homogeneous ice nucleation in an ab initio machine learning model" by Pablo M. Piaggi, Jack Weis, Athanassios Z. Panagiotopoulos, Pablo G. Debenedetti, and Roberto Car (arXiv preprint https://arxiv.org/abs/2203.01376). In this work, we studied the homogeneous nucleation of ice from supercooled liquid water using a machine learning model trained on ab initio energies and forces. Since nucleation takes place over times much longer than the simulation times that can be afforded using molecular dynamics simulations, we make use of the seeding technique that is based on simulating an ice cluster embedded in liquid water. The key quantity provided by the seeding technique is the size of the critical cluster (i.e., a size such that the cluster has equal probabilities of growing or shrinking at the given supersaturation). Using data from the seeding simulations and the equations of classical nucleation theory we compute nucleation rates that can be compared with experiments.